Annals of Biomedical Engineering, Vol. 23, Suppl. l , pp. S-1-S-134, 1995

0 0 9 0 - 6 9 6 4 / 9 5 $10.50 + .00 Copyright 9 1995 Biomedical Engineering Society

Printed in the U S A . All rights reserved.

Abstracts of the Biomedical Engineering Society 1995 Annual Fall Meeting, October 6-8, 1995 Industry/Society

1

2

CLINICAL GAIT ANALYSIS: A TOOL IN ORTHOPAEDIC SURGICAL PLANNING

TISSUE COMBUSTION POLLUTANTS tN THE OPERATION ROOM Ott DE School of Engineering, Mercer University,Macon, Georgia

R.B. Davis. Ph.D.. and P.A. DeLuca. M.D. Gait Analysis Laberato~. Departme.nl of O~hopaedics. Newington Children's Hospital. Newington, Connecticut

Operating room air quality is affected by production of creation of toxic and hazardous materials dudag surgery by use of laser, e|ectrosurgical and mech~eal devices. Toxic chemicals aad potentially infectious material are produced by combustion, mechanics] or vibratory mechanisms interacting with tissue. The aet'~osolscreated are potentially hazardous earcinogonic, mutegeaie and infectines mteriai in various conceatratinus and levels of danger. Formaldehyde, hydrogea cyanide, beazeae, aerolein and poly-aromatie hydrocarbons ere produced from tissue combustion. The greatest impact on reducing dangerous conceoteatinan of noxinus materialscurrently is proper use of smoke evacuation unite. Surgical ~ct-m at best protect against particulates to O. 1 microns but do not control volatile chemicals. Exposure reduction by masks to threshold limit value levels require full gas numk apparatus. Hazard reduction user compliance and comfort are aecer,ssry to create the best solution for a more perfect breathing system in coDtnminnted r

Over the past fifteen years, technoIogy and protocols have been developed to allow for the mutlne assessment of pathological gait in the clinical selling. Data can be readily gathered that are associated with the instantaneous spatial position of markers placed on the patient's body segments. Computation strategies, such as Euler angles, can then provide objective documentation of the three-dimensional angular orientation of the segments or the relative position of the lower exWemityjoints. With simultaneously collected data &~ociatsd with ground reaction Ioedlog, Newtonian mechanics can estimate net joint moments and Inrees. This information can then be used in conjunction with electmmyographic data to facilitate a better understanding of the underlying mechanisms employed in walking. It is important to apprecinte that gait analysis results ale subject to ~ e n t error associated with the equipment limitations or improper pretoonis, as well as random errors associated with excessive soft tissue movement or insufficient anatomical modeling. In addition, the compotational models include simplifying assumptions and anthropomctric nstimat~ that also reduce the relative pt'ecision of the Infocmation. Consequently. the fundamental cbailengns in the inteqnetation of clinical gait data for surgical decision making are 1) the integration of the infomlation contained in all of these data types, identifying consistent relationships while resolving apparent contradictions, and 2) the avoidance of the over interpretation of the data through the imowledge of the limitations of the mensomraont technology, the models employed, and the particular patient cbaractmistins. It is important to recognize that ctusent clinical gait analysis approaches provide information for inteq3eetefionthat ~ t s the best available ~ o f the subject's performance. The presentation willInclede a brief de..q~pfion of the proPseols used at the Newington Children's Hospital Gait Analysis Laboratory with pertinent clinical illusWa~ons.

S-1

Education/History

S-2 3

4 BIOMEDICAL ENGINEERING: A BIBLIOGRAPHIC MATRIX OF ACTIVITY IN THE 1960S AND 1970S Paul H. Fagette, Jr., Ph,D. Department of History, Arkansas State University

Abstract Not Available

A matrix of publishing activity by authors in the first live years of Ar, nals of Biomedical En_o'lr,e~riqg has been constructed. Approximately 4,.500 abstracts from Medline 09641977) and 220 articles (volumes 1-5 of Annals) were surveyed. This investigation has sought to create a direction of historic development by using what was defined as biomedical engineering by practitioners of the time. First level general area categories were defined physiologically or by application, e.g~ neurophys/ologlcal or rehabilitation. The next level of order was within a bioengineering context ~ t up by Annals. e.g~ mechanics, mass I~ansport, or cellular. The last category was if a model was proposed. If so what were its parameters of measurement and the level at which it worked, e.g., multicel|? Additional information on biographical background was collected as available. Results show a decided emphasis on circulatory and electrical studies throughout the 1960s and most of the 1970s. Annals studies tended to focus on modeling/simulation versus experimenL In fi-,elater part of the 70s a definite expansion of rite notion of biomedical engineering emerged as new areas such as cellular and prosthetic efforts were added. This augmentation was also reflected in a staged evolution in the first five years of Annals. Years three and five reveal the broadening of the definition and inclusion r a wider array of science and engineering fields. Analysis reveals a significantly higher proportion of models in Annals as compared to accompanying journals.

H e m o g l o b i n - b a s e d R e d Cell Substitutes

S-3

5

8

P O L Y E T H Y L E N E G L Y C O L CONJUGATED BOVINE H E M O G L O B I N (PEG-HB) AS A OXYGEN CARRIER. C.D. Conover, L. Sedlatschek, R. Linberg, K. Shum and R.G.L. Shorr Enzon, Inc., 20 Kingsbridge Road, Piscataway, NJ 08854

TRANSPLANTING HUMAN ~IEMOGLOBIN" INTO HEMOGLOBIN FRONTICELLL C. U. Maryland Med. School, Baltimore, MD 21201

The advent of viral contamination of the donor blood supply has highlighted the need for a safe and effective red blood cell substitute. Enzon, Inc. has focussed its attention to the research and development o f polyethylene glycol (PEG) conjugated bovine hemoglobin. Attachment of PEG to proteins has been demonstrated to prolong circulating half-life, alter biodistribution by fa~,oring vascular retention, and blunt any immune or allergic response. PEG-Hemoglobin has a circulating half-life o f 18 hours in rats and 50 hours in dogs. PEG-Hb has been shown to cause no alterations in cardiovascular function, hemoglobinuria, gross morphological renal damage, abnormal blood chemistry or allergic reactions in rats, rabbits, clogs and pigs. In addition to evaluation of the safety of PEG-Hb formulations and their ability to deliver oxygen, efficacy models have been examined. These include the use of PEGHb as a sensitizer to radiation in the treatment of hypoxic tumors.

Bovine hemoglobin [HbBv] has a lower oxygen affinity than human hemoglobin [HbAI regulated by physiological concentrations of CI'. This mechanism of O 2 affinity regulation represents an alternative to the 2,3-DPG regulation exhibited by HbA. Fronticelli (Biophys. Chem. 37:141,1990) has proposed that the CI" regulation in HbBv is introduced by particular amino acid residues located in the A and E helices of the [?,-subunits. In accordance with this proposal we have constructed two mutant human hemoglobins 13(VIM+H2deI+T41+P5A), [PB4], and J~(VIM+I-I2deI+T4I+PSA+A76K), IPB5]. In the absence of CI', PB4, PB5 and HbBv have the same 02 affinity, three fold lower than HbA. In the presence of 200mM CI', the 02 fft'mity of PB5 and HbBv are three fold lower than in H b A and PB4. This indicates that the substitutions in PB4 are sufficient for introducing in HbA the intrinsic low O2 affinity o f IToBv, however, these substitutions decrease the sensitivity to CI'. In turn, CI" sensitivity is re-introduced in the molecule by the substitution I3(A76K). Crystallographic analyses suggest that in the deoxy conformation, l~K76 bridges with [3K8 through CI ions. This

BOVINE

interaction is not present in oxy conformation.

6

9

HEMOGLOBIN-BASED RED CELL SUBSTITUTES (HBRCS): AN OVERVIEW H.W. Kim Brown University/ The Miriam Hospital, Providence, RI.

PREoCLINICAL SAFETY AND EFFICACY STUDIES OF THE EXPERIMENTAL RED CELL SUBSTITUTE, LIPOSOME ENCAPSULATED HEMOGLOBIN Alan S. Rudolph

Red cells from donor blood is most effective and blood transfusion is a currently accepted therapy for severe anemic conditions such as hemorrhagic shock. Donor blood is, however, limited, requires typing and crossmatchingbefore use, and carries a risk of disease transmission. Therefore, a red cell substitute that is readily available, effectively dellvers oxygen, and yet pose no risk of disease transmission would be highly desirable. Currently, hemoglobin based oxygen carrying solutions ere being evaluated as potential red cell substitutes. In this overview, experimental and theoretical basis of HBRCS, current methods of hemoglobin modification (Intra- and intermolecular crosslinking, polymerization/conjugation, and encapsulation}, oxygen binding and delivery characteristics, physiological effectiveness and problems, and issues regarding clinical efficacy and safety w i l l ~ briefly described.

7 EFFICACY AND SAFETY EVALUATIONS OF HEMOGLOBINBASED OXYGEN CARRIERS C. Robert Valeri Naval Blood Research Lab., Boston Univ. Sch. of Med. Over the past ~ years significant progress has been made to demonstrate the safety of hemoglobin-based oxygen carriers in animals, normal volunteers and patients. Previous studies have reported that hemoglobin-based oxygen carriers produced renal insufficiency, coagulation disorders, hypertension and anaphylactic type reactions. Several companies have developed methods to purify and modify tetrameric hemoglobin and to prepare polymerized hemoglobin from human hemoglobin, bovine hemoglobin and recombinant stroma-free hemoglobin produced in E. coli and in yeast. Polymerization of cross-linked tetrameric hemoglobin can prevent renal excretion and tissue distribution into the extravascular space. The efficacy of the hemoglobin-based oxygen carrier is being compared to red blood cells, colloid solutions, and crystalloid solutions in animals and in patients. Several indications for the use of the hemoglobin-based oxygen carriers in clinical medicine are being investigated.

Canter for Bio/Moleculer Science and Engineering, Code 6900, Naval Rcseerch Laboratory Washington, DC 20375-5348 One atrctegy to deliver hemoglobin in an oxygen carrying fluid is based on the sequestration of bemoglobin in blocompatible carriers. Much of the current work in this field is focused on the use ofliposomes. Lipoanmes are biodegradable capsules which ponnit the diffusion of oxygen while encapsulating hemoglobin within an aqueous environment. The

encapsulation of hemoglobin in liposomes extends the circulation persistence and alters the blodlstributinn of free or chemically modified hemoglobin. Numerous animal studies have shown effective oxygen deliveW by liposomes with hemoglobin in models of hemorrhagic shock. The focus of much work to define consequences of encapsulated hemoglobin administration has been on the effects of large dose lipoanme application. This work has been directed at discerning the effects of lipoanme encapsulated hemoglobin on organ.~of the retlculoandothelial system, particularly the liver and apleen as these are the principal sites of lipesomc accumulation. A number of mmsiant effects ere observed following the administration of clinically relevant doses of liposome encapsulated hemoglobin such as a rise in liver enzymes, thrombocytopenia, lenkocytusis, and complement activation. More recent studies have examined differences in the vasooctivity of encapanlated vs. free hemoglobin. These studies indicate that encepanlation markedly attenuates the vasoactivity of hemoglobins in an isolated aortic ring model. Large sale manufacturing methods m produce sterile filtered preparations have been developed which increase the opportunities for GMP production"and commercial development. Recent progess in the pre-clinical utfaty and efficacy studies of Iiposomc encapsulated hemoglobin will be presented.

S-4

Targeted Drug Delivery

9a

10a

COMPUTA~ONAL APPROACHES TO SEARCHING SEQUENCE SPACE Alan Lapedea Theoretical Division, LANL and Santa Fe Institute

Purified Hematopoietic Stem Cells: Targets for Gene Delivery?

Methods of combinatorial chemistry have the capability to produce a tremendous number of sequences that have been selected for a given function. However, practical limitations make it unfeasible to experimentally search all of sequenoe spaoo. In this talk we discuss methods that use a finite amount of experimental data produced in combinatorial chemistry experiments to "loam' the map relating sequence to function. This approximate map or 'landscape" can then be searched by computational means to suggest new sequences with enhanced function that did not exist in the original experimentally produced data set.

9b IX](~INNANDPKI~IDE LIC~U~}S B r i a n K. Kay, Andrew B. S p a r k s , Noah O. Koffman, James E. R i d e r University of North Carolina at Chapel Kill

Margaret A. Goodell and Richard C. M~Uigan Whitehead Institute for Biomedical Research Massachusetts Institute of Technology Hematopoietic stem cells (HSC) are the ideal target for gene therapy of blood disorders due to their persistent role in replenishIng all the differentiated cell types of the adult blood. We have developed a novel strategy for purificationof these cells from murine bone marrow based soley On an unusual dual wavelength analysis of the fluorescence of the vital dye Hoechst 33342. In one step, we obtain a population which is virtually homogeneous for cell surface markers shown previously to be present on routine HSC. The purificationstragey has also led to the identification of a small subset of HSC which are replicating (1-3%), and which m a y be superior targetsfor retroviral-mediatedgene transfer. W e are developing methods for gene transferInto these cells.W e are also investigatingthe applicationof th3spurificationstrategyto stem cells from other species such as humans.

Many proteins involved in signal transdoction and the cytoskeleton contain a structural motif known as the SH3 domain. This 60-70 amino acid long domain resides in such proteins as Sre, Abl, Nck, Grb, cortaotln# and spootrin and most likely plays a role in proteln-proteln Interactions inside the eukaryotie cell. To define the llgand specificity of SN3 domains, we have utillzed M13 phage-dlsplayed random peptide libraries. Individual phage san be isolated from libraries composed of billlons of different recombinents, based on their ability to blnd to target molecules of interest, and the primary structure of the displayed peptide can be deduced by sequenoing the appropriate region of the viral genome. By this method, we have Identified peptide llgands for over II different SK3 domains. AS the group, the peptide Iigands are short and prolIne-rioh, but differ In primary structure for each SB3 domain.

10

II

THE BIOPRECURSOR APPROACH TO TARGET DRUGS TO THE BRAIN M.G. Paifreyman, LA. McDonald and P. Bey SCRIPTGEN Pharmaceuticals,Inc., 200 Boston Ave. Medford, MA 02155, SIBIA 505 Coast Blvd. S., Ida Jolla, CA 9203"/and Marion Merron Dew, 2110 E. Galbraith Road. Cincinnati, OH 45215

TARGETED DELIVERY OF POLYNUCLEOTIDES TO H E P A T O C Y T E S G Y W u . Y. Z h a n g , a n d C . H . W u D e p a r t m e n t o f M e d i c i n e , D i v i s i o n o f G a s t r o e n t e r o l o g y - H e p a t o l o g y , U n i v e r s i t y of C o n n e c t i c u t S c h o o l o f M e d i c i n e , F a r m i n g t o n , CT, U S A .

The blood brain barrier (BBB) is an effective mechanism to protect the brain against noxious substances, It is also a potent impediment for penetration of potentially valuable therapeutic agents, Multiple approaches have been devised to target drugs to the brain and these include altering lipopbilicity,opening the BBB with chemicals or using specific transport mechanisms. This latter approach has been funber refined to target specific subpopulations of neurons in the brain by application of the bioprccursor approach to site-selective drug targeting, Amicoacids are readily transported into the brain with an active uptake system; their metabolic products such as amine and acid neurotransmlitars are not. We have developed a series of enzyme inhibitors and receptor antagonists that exploit the aminoecid uptake system and are then converted into the active moieties by enzymes that are located specifically in the neuronal systems of the brain. (E)-B-Fluoromethylene-m-tyrosine is a taken up by the brain aminoacid transport system and then decarbexylated to (E)-B-fluoromethylene-m.tyramine by aromatic L amino acid decar'ooxylnse in catecholamine neurons. The product is a potent mechanism based in'everaibieinhibitor of monoamine oxidase A and has potential as an antidepressant. A related approach to target serotoninergic neurons uses the bloprecursor (E)-B-fluoromethylcne-5hydroxytryptophan to generate the corresponding serotonin analog. 7, and 5,7-Chlorokynurenic acids are potent antagonists of the glycine site on the N-methyl-D.nspartate subtype of glutamate receptors which are implicated in acute and chronic neurodegeneratlon, cognition and epilepsy. They do n6t penetrate the brain; their bioprecursors, L-4-ssd 4,6-chlorokynureninns do. The bloprecursprs are then transaminated by conical kynutenine aminotransfarese to the active antagonists. Thnse examples illustrate the power of targeting bioprecursors to the brain.

We have shown previously that foreign DNA can be targeted to hepatocytes resulting in new gene expression. The delivery process Is efficient, but so Is subsequent degradation. To determine whether the expresslon of receptormediated gene delivery could be enhanced by incorporaUon of a simple bacterial protein into the DNA carrier. 1) A liver-targetable DNA delivery system is based on the presence of receptors on hepatocytes t h a t bind a n d internalize galacteae-termlnal (astalo-)glyeopmteins (AsG) 2) Listerlofysin O (LLO}, a bacterial protein, can lyse endosomes in a pH-dependent process. A DNA-carrier, aslaloorosomucold (AsOR) coupled to polylysine (PL), was subsequently linked to LLO. The LLO-conJugate w a s complexed in pCMV-luc containing the luciferase gene and expression was determined by lucfferase assay. After 48 hrs exposure to Huh7, AsG receptor (+) cells. DNA complexes containing LLO showed targeted gene expression greater t h a n 100-fold more than for DNA complexes alone. SKHepl. AsG receptor (-) cells had no significant expression under any condition. The enhancement was completely blocked by an excess of an asialoglycoprotein. Mixing of LLO with complex increased expression only 4-fold. There was no evidence of toxicity. Conehmlons: Incorporation of a pH-dependent endosomolyUc bacterial peptld. into an asialoglycoprotein-based DNA carrier system results in d ~ a f l c a ~ v enhanced foreign gene expression that r e t a i n s its hepatocyte-directed cell specificity, (Supported in part by grants: NIH DK 42182. March of Dimes 901236, a n d TargeTech, Inc. GYW and CHW hold equity in the Immune Response Corp.)

Biological

Applications

of Mass

Spectrometry

12

15

RECENT DEVELOPME.NTS IN THE STRUCTURE DETERMINATION OF PEPTIDES, PROTEINS AND SULFATED OLIGOSACCHARIDES Klaas Blemann Department of Chemistry, Mass~hasetts iostitate of Technology, Cambridge, MA

TANDEM HIGH-RESOLUTION MASS SPECI'ROMETRY OF LARGE BIOMOLECULES. Fred W. McLafferty Baker Chemistry Laboratory, Comell University, Ithaca, NY 14853

Ever since the early" 1980s. new ionizationmethods have revolutionizedmass spectrometry making it one of the important methods in biochemistry and some aspects of biology. For the detailed and reliable sequencingof peptides, collision-induceddissociation nf fast-atom bombardment generated [M + Hi§ ions in a four-scorns tandem mass spectrometer allowed the determination of the primary smgture of proteins entizelyby mass spocuomeuy. Since the late 1980s, matriz-assistod laser desoq)tion ionization(MALDI) made it possible to determine the molecularweights of proteins and other "biopolymen"in the l0~ 106 Da range with pioomolesensitivity. This method is also very useful at lower masses (kDa range), wha~ the sensitivitymay even be censidambly higher. We are also utilizing MALDI in a novel way to determine the moleculm"weights of compound classesthat are too polar to be ionized by conventionalmeans,such as molecules containing a number of --C-SO3E or -O-SO3H and -NHSO3H 8rottps. The latter two ale present as substituentson the polysaccharidechains(giycosamioogiycans,GAG) of hepedn. beparan, and related biologically and clinically impotlant molecale~of as yet ill-defined smlctsn~.

The combination of el~ctrospray ionization (ESI) with FourieroUansform mass spectrometry (FTMS) has unasaal capabilities, recording many (e.g., 5000) ions simultaneously (< I s) over a broad mass range (>105 De) at >105 resolving power using <10-I 5 tool of introduced sample. ESI/FI'MS can now provide accurate (+l Da) MW data and extensive sequence information on low-purity large proteins (-70 kDa) and nucleotides (-30 kDa). The three active components of rabbit creatino kinas~ (43 kDa) have the same MW values within +1 Da (deamidatinn?), with no amino acid exchange isomers detected by MS n. Reaction with C6H5COCHO adds 116 Dam each Component, indicating all are enzymatically active, with MSn restricting this site to <25% of the molecule. "Pure" thiaminas~ (42 kDa), instead, shows three major MS components, MSn demonstrates that these differ by an extra A]a and AlaGly on the N-ten-ainus, and corrects the DNA
13

16

DEPENDENCE OF PEPTIDE FRAGMENTATIONON SIZE AND SEQUENCE V.H. Wysocki, A. Somogyi, A.R. Oongr6, J,L Jones, H. Nair Dopartment of Chemistry, Virginia Commonwealth University, Richmond, VA 23284.2006

ELECTROSPRAY IONIZATION AND TENDEM MASS SPECTROMETRY OF ~ T E S Bruce Reinhold, Song Ya and Vernon Reinhold Mass Spectrometry Facility, Boston University School of Medicine

Collisions of iow-energy (eV) ions with a surface lead to structurally cheractedstic fragmentation by a process called surface.induced dissociation, SID. In the wod( to be desedhed, surface-induced dissociation Is applied to determine mechanisms and anergetics of peptide fragmentation. The surfaces fnveetigatedconsist ct monolayer films of alkanathiols chemisorbed on gold (JACS, 1991,113, 8967; 8969). The extent of fragmentation varies significantly with tha nature of the serf-assembled monoisyar film, with a fluodnated film providing much more extensive fragmentation than an alkenathiol film. For the pepfidss, fragmentation has been explored as a function of pepfide size and sequence, collision energy, and type of surface. Electroapray ionization Is combined with surface-induced dissociation to determine fragmentation efficiency cuwss is fragment ion intensity/a tOtal ion Intansity vs. collisioo energy] (JACS, 1994, 116, 8368). The~e curves depend strongly on the umioo acid ssquenca and size and provide insight Into other atructural features that may make dissociation of pattioular sequences more difficult than others, The data show an excellent linear correlation between the onset energy for Iragmentation and the gas-phase baslctty of the most basic residue in the pepude. All of the data are in agreement with a mode{ In which the peptide is odginelly protonated at the most basic site in the molecule, with proton transfom to ~ basic sites and aubssquent fragmantation initiated by coifision I with the surface. In two related projects, largo multiply-charged peptides are fragmented by collision with the surfacn and thermal activation of prctonated dtmars is' compared with surface-induced dissociation to obtain binding energies.

The combination ot electrospray ionization and tandem mass spectrometry Is e now and powedul tool for 1he analysis of biological glycoconjugates. Electrospray (ES) generates molecular ions with low internal energies, making it possible to ionize by forming weak noncovalent complexes between the molecule and solvatod ions. This allows considerable flexibility in experimental protocols. Blocking polar groups by alkylaUon and ionizing by adducting metal cations provides an especially effective method for oligossccharide analysis and some motives for this will be presented. Ion fragmentation analysis allows a detailed structural characterization of many ollgosaccharidas and glycoconjugatee. A discussion of the structural problem confronting the carbohydrate biochemist will omphasiza the topological aspect of the analysis because this information Is uniquely accessible by mass epectromelry and reference to topological features underlies the interpretation of fragmentation spectra. The relation between methylatlon, topology and glycosidic cleavage will be discussed as will linkage analysis through fragmentation pathways involving opening of the glycosyl ring.

14

17

MASSSPECrROMETRICINVESTIGATIONSOF NONCOVALENTINTBRACTIONSIN PROTEINS Paul9Lei, CynthiaPitseaberger, lviiahsolEater ring,Jonathan Ametef Department of Chemistry,Universityof Georgia,Athens,GA 30602-2556

PROTEIN STRUCTUREAND INTERACTION

Mass apeGtrometryis now known to providepowerfuleapabilltisefor ehtr~tefi~g the prlma~ tmxtctoreof protein. The mada~ innizatioomethodsthat are maDdfor protein mtm ~ , r niectrespray ionizationand mstnx-mmistndlater demrption ionizationneutlly dimtpt nogtmvalant beadingin biomolectdes. Under"typicalopefttlng nanditioas,a proteinthat is composedof more than one subunitwillyield 9 mass ~pnetmmonlyof the individualeubuai~, mui not of the mdive molneule~ In the same fashina, metal-~entera that are held by nonoovalnatintecsotienaia the intm'iorof protaine are usually~ duringma~ ~-'treme~y analysis. U~ng nonde~turiag solutions,one one obtain elsotrnepr9 man spectra of the nttive protalne. Thi==apabilitywill be ilium9149for the protein ~ which is bningiavastigatadia our laboratory. Thinia an octamario, iron.r proteinthat is nempoNd of two typosof Babtmlta. El~troR~rsy iani~ttlonof this protein usingmmderdce~litlc~asyinldea mt~ gpec~un with paska ceaespomiingthe two monomericsubuaitswithouttheir iron atoms. V/hemexaminedusing naodmatudng coaditious,silpais oonmpandi~ to the oetamarioprctaln m~e in the m m spectrum. Tbe netamernunbe "hneted"in the mmmspnetrometerto breaksomeof the wnekeatof the bondsthat holdthe complextogether,to yieldmon~el~ spo~es that now &-efomulto retain theirmetal atom& The capabilitiesand limitationsof qnatem~ ~ analynitby mmm q)sotrm~etrywillbe dismum~

NEW MASS SPECI'ROMETRIC APPROACHES TO THE ELUCIDATIONOF B.T. Chalt The Rockefeller Univursity,1230York Ave, NY, NY 10021 Recentadvancesin mass spectrometryare providing powerful new tools for studying proteins. Two techniquesin particular - matrix-assistedlaserdesot'pfiotVionization mass spectrometryand electraspray ionization massspectrometry-- are beginning to make a significant inpect on oar ability to solve challenging biological proberne. Problems that can be addiessed by these new tools include the rapid identificationof proteins, the elucidation of protein primary structures with an emphasis on the elucidation of post-translational modifications, and the determination of sites of protoolyticprocessing. We willdiscuss a selectionof such applications. In addition. we will discuss some relatively new eses of mass spnctromet~, including rapid protein ladder sequencing (from the amino- and car~xyl-tetmini) and the identificationof bindingepitopas and regionsof specificprotein-protein and proteiooligoancleotideinteraction.

S-5

S-6

Molecular Design and Structure/Function of Ion Channels

18

20

S E L F - A S S E M B L I N G P E P T I D E N A N O T U B E S AS ARTIFICIAL TRANSMEMBRANE ION CHANNELS AND PORES M. Rcza Ghadiri, Departments of Chemistry and Molecular Biology, The Scripps Research Institute, La Jolla, California 92037. A new strategy for the design of functional transmcmbranc ion channels and pore structures is described. The approach is based on the de novo design of fiat ring-shaped cyclic pcptide subunits having the propensity to undergo spontaneous self-assembly, upon incorporation into lipid bilaycrs, to form water-filled transmembranc [5-sheet like cylindrical ensembles with predetermined internal diameters. Such transmembraue structures display remarkably efficient ion and molecular transport efficiencies. The underlying design principles, characterization schemes, and functional properties will be presented.

19

Abstract

Not Available

"Palpating K+ ch~mncls with Peptide Toxins, Sulfur, and Silver" Christopher Miller, Howard Hughes Medical Institute Graduate Dept. of Biochemistry, Brandeis University The gross molecular architecture of voltage-gated K+ channels is now pretty well known. The channel forms as a tewamcr of identical (or very similar) subunits. The external vestibule and pore-lining sequences are found between the 5th and 6th transmembrane helical segments. Omrybdotoxin (CTX), a 37-residue peptide, binds specifically in a receptor site located at the outer opening of the pore. By using complementary mutagenesis of both CTX and Shaker K" channels, residues in the chan~l's vestibule that contact known toxin residues have been identified. This locates these channel residues in three dimensional space. In addition, the sidechaln projections of the "deep-pore" residues have been identified by cysteine-scnnning mutagenesis of the channels, which are then challenged with a solfhydryl-active reagent that is also a K" analogue.

Poster Presentations

S-7

21

23

DOSIMETRIC ANALYSIS OF RADIOIMMUNODETECTION AND THERAPy USING A PHYSIOLOGICALLY BASED PHARMACOKINETIC MODEL Hui Zhu, Laurence T. BaxW.r,and Rakesh K. Jsin. Department of Radiation Ontology, Massachusetts General Hospital and Harvard Medical School, Boston, MA 02114

DISTRIBUTION OF BCNU AND TRACERS IN THE RABBIT BRAIN FOLLOWING INTERS'ITrlAL DELIVERY BY BIODEGRADABLE POLYMER IMPLANTS J.F. Stsasser, L. Fung, S. Eller, S.A. Grossman and W.M. Saltzman Johns Hopkins University, Baltimore, MD 21218

The difficulty of making accurate dosimetsy measurements in humans has led to a scarcity of date necessary in planning schemes for cancer detection and treatment. The difficulty of estimating dose rate or total absorbed dose in radioimmunotberapy is compounded by uncertainty in the location of the radioisotope after injection. We have developed 9 physiologically based pharmacokinetic model capable of predicting monoclonal antibody biodistribetion in humans by scaling up from mice. In this study, the model was used to explore the optimal diagnostic and therapeutic potentials of radioimmunodetection and therapy. Using the antibody pharmacoklnetics generated by the model, the absorbed doses were calculated in tumor and normal tissues for radiolabeled (67Cu, 903(, 1311, and 188Re) antibody (enti-CEA ZCE025) and its fragments using the Medical Internal Radiation Dosimetry scheme. Organs considered inchfde the tumor, heart, lung, liver, kidney, spleen, bone, muscle, skin, G.I. tract, plasma, and urine. The model calculations showed that 131I combined with ZCE025 F(ab') 2 fragments provided the highest tumor:bone marrow absorbed dose ratio at the optimized antibody injection doses, with 90y+F(ab')2 another good choice. In addition, model sensitivity analysis demonstrated that antigen expression in normal tissues, such es bone marrow, had little contribution to the absorbed doses under the baseline physiological conditions. The analysis also identified that tumor antigen expression, tumor antigen-antibody binding, and transvescniar antibody exchange were the major factors limiting antibody accretion in tumor. Thus, the model suggested that radioimmunodetecritn and therapy could be improved by upregulating tumor antigen expression and increasing vascular permeability.

Intracranini tumors, such as giioblastuma mnitiforme and astrocytomas, are among the most aggressive and difficult to cure. In the present study, we evaluated the intracranial distribution of released agents during the first three days following implantation. Polymer implants containing [3H]BCNU (l,3-bis(2-chloroethyl)-l*nitrosourea), [3H]dextran (M w 70,(}00), or [14C]IAP were implanted into the brains of rabbits; autoradiogeaphy was used to measure the distribution of radiolahels within the brain at 6, 24, and 72 hr after implantation. For all of the agents studied, the majority of the radioactivity was found within the region 1-2 mm from the surface of the polymer. Dextran, however, penetrated farther into the brain than either IAP or BCNU. The distribution of radiolabel on an anteroposterinr axis was determined by examining serini coronal images: after 72 In-. significant radioactivity (<2 standard deviation above background) extended >17 mm in animals with [3H]dcxtran implants and --6 mm in animals receiving [3H]BCNU or [14C]IAP. Concentration profdes were also measured on coronal images obtained at the implant site: radioactivity dropped to a 10% maximum value 1.7 mm from the surface of the pellet in [3HI dexlran-trented animals and <1.2 mm in [3H]BCNU or [14C]lAP-treated animals. Measured concentration profiles near the polymer were compared to mathematical models of drag diffusion and elimination. These results demonstrate that agents delivered into the brain by intracrnninily implanted polymers penetrate into the tissue within 1-2 mm of the implant and that the size of the treated region depends o~ physicocbemical properties of the agents.

22

24 C O M P U T I ~ A ] ~ ] ~ M O D I ~ I N G OF HUMAN AND BOV]Ng G~OWTH H O R M O h ~

SOLID FREEFORM FABRICATION OF DRUG DELIVERY DEVICES

B,M. Wu I.A. Hirahara i,L.G. Cima 2,M.J. Cima i Uaiverslty, Aaron, ~

4.q701

Ahalnd The Bovine growth bermone(bGH) and the Human growth bermone(hGH) have beea modeled using 9 3-dimensional protein modeling software package developed in-bet.lse. The graphical package was developed using the engineering modeling system tad a C language interface on the CAD/CAM system and coupled with the Charmm21 molecular modeling pmlFam to perform energy minimization and molecular dynamics. Twelve different mutetlom in the third alpha helix oftha bGH and hGH have been studied using computer modeling and the results have been compared to the experimeatel results obtained by the Edison biotzehnology institute. The 3-D structure of the mutated zequcoco where predicted using gtafistical and neural network algorithms developed in-house. Thin research wuperformed iathel~omolecniar modeling facilityat Ohio Unive~ity, which utilizes an Intorgeaph CAD w/stem connected to the Ohio Supereomputor. The resee~h is ixut oftbe on-going developmmts in the field ofbiomolocular cogincwaing and drug design at Ohio University in collaboration with the chemical eagineering and the biotechnololD" ~ t . The.interdisciplinary research efforts are focused on the design and production of recombinant molefules through 9 gsucture/fimction approach. The initial efforts have fo~zed oa the design, production and biological evaluation of bGH and hGH amdogs.

i Department of Materini Science and Engineering, and 2 Department of Chemicni Engineering, Massachusetts Institute of Tecimology, Cambridge, Massachusetts Three dimensional printing (3DP) is a solid freeform fabrication method where beth the macro and micrnstructure of the constructed device can be controlled since objects are built in a layer by layer fashion by joining very fine panicles together with liquid binder. The cun'ent project focuses on the application of this novel technology for fabrication of polymeric drug delivery systems. The structure dimension of the device is fwst specified in a computer model and then built using the 3DP process. Potentially complex drug delivery regimes can be created in this way, such as prescriptive multi-drug release, or multiphasic relense of a single drug. For prnof of concept purpuses, dyes are used as model drugs in a matrix of biocompatible polymers. The dye release profile can be controlled by 3DP by specifying the spatial dislribmion of the dye, the local composition of device structure, and the local microstructure within the devices.

S-8

Respiratory Airway Mechanics I

25

28

NON-UNIFORM DISTENTION CAN GENERATE CLEARANCE FLOW IN THE LUNG F. F. Espinosa and R. D. Kamm Massachusetts Institute of Technology, Cambridge. MA The airway liquid lining exhibits a surface tension iu the trachea of about 30 dyo/cm, whereas the alveoli experience a surface tension between 0 and 25 dyn/cm. Thus a surface tension difference exists which may assist the mucc-ciliary escalator in clearing liquid and particulate debris from the lung. Several factors might ceutribute to this difference, one of which is nonuniform surface distention in combination with surface tension hysteresis during breathing. Where strain is small (e.g., central airways), surface tensions remain close to equilibrium values. Where strain is larger (e.g., periphery) cyclic distention produces surface tensions ranging from slightly above equilibrium to near zero. To investigate this effect, we numerically solve the oue-dimeesional conservation equations for continuity, momentum and sarface concentration for viscous flow over a distensible membrane, the strain of which is a linear function of distance.A linear equation-of-state relates surface concentration to surface tension. Srsaption Of suffactant to the liquid-air interface is determined by a sosption model based on the behavior of normal lung surfaetant. The governing parameters are the average membrane strain, c , a ratio of cycle time to viscous time, %/r,,, a ratio of desoq)tion time to adsorp2tion rime, "t~/~ and a ratio of viscous time to desorption time, ~j~.v __ where x,,=-~tL/boAt', it is viscosity, L is membrane length, ho is liquid layer thickness and ff is surface tension. Flows toward the non-distensible end of the membrane were simulated over one cycle. A family of eorves for different values of xn/'d ' were generated, showing net volume transported as a function of %/*.,, for values of ~ ffi0.1 and %/~ ffi0.024. Maximum net flow was generated when % / ~ - O(1), which increased with increasing values of ~/'d'. To mimic physiological conditions, we set Iio= I ~tm, It = 0.01 dyn s cm "2. % ffi4. s/breath, f. ffi0. I, aud L = I cm, resalting in near maximal surface velocities of 0.1 mm/s. "I~ese are comparable to the velocities associated with transport due to cilia of about 0.2 mm/s. (Support from the NHLBI, HL33009, and the Freeman Foundation is gratefully acknowledged.)

NON-LINEAR VISCOELASTIC MODEL OF TRACHEAL WALL DYNAMICS. N Aljuri, L Freilag, JG Veuegas. Massachusetts General Hospital, MIT, Harvard Medical School,Boston MA, and Ruhrlandklinik, EasenGermany. The relationship between transmuml pressure (Ptm) and tracheal ornss-sectioanl area (Air) (the tube law, TL) measured under static conditions or using small amplitude osclllatiues can nol describe the non-liuear dynamic behavior of the tracheal walls during large amplitude perturbations such as in forced expiration (FE) or high frequency oscillatory ventilation (HFV). To describe such behavior we developed a non-linear viscoelastic model of dynamic airway collapse consisting of a non-linear compliance CI and a non-linear resistance R iu series and connected in parallel with a second non-linear compliance C2. The model assumed that the characteristic time constant RCI was independent of Ptm and Air. The non-linear characteristics of C1 and C2 could be obtained from the TL measured at two limits: immediately after the step change in Pan corresponds to the dynamic TL (DTL); and when time approached infinity corresponding to the static TL (STL). The model was implemented iu S1MULINK (The MathWorks Inc, Natick Mass.) and applied to optical measurements of TL's in excised human tracheas during b-E and during HFV. We found that the STL was significantly more compliant than the DTL. We also found thai the model described well the stiffer behavior of the tracheal wall dining FE and HFV at 5 Hz. However the model failed to describe the behavior at 15 Hz where the trachea collapsed even beyond the areas predicted by the morn compliant STL. We speculate that tracheal collapse during HFV is exaggerated when the period of osclllatioj1 is too short to allow for a full recovery of the distended viscoelastic tissues. (Supported i t part by grant HL35267)

26

29

A MODEL FOR EPITHELIAL FOLDING IN NORMAL AND ASTHMATIC AIRWAYS Conslarlth~ A. Himlsist , Barry R. Wiggs2. Jeffrey M. Drazea3, Roger D. gamin I tFldd Mechanic* Labotmory, ~ n s a ~ Institute Of Techuology, Cambridge, MA, USA 2phlmonary Research Labeslory, University Of British Columbia, Vanccovor, BC, Canada 3Harvasd SdmoI of Public Health, Boston, MA, USA

SERIAL DISTRIBUTION OF AIRWAY DIAME'IERS FROM INPUT IMPEDANCE AND COM~ TOMOGRAPHY. S.A. Wood, R.H. Habib, W. Mitznsx, and A.C. Jackson. Respiratory ~ Laboratory, Biomedical Engineering Department, BostOn University, Boston, MA and Division Of Physiology, The Johns Hopkins University, Baltimore, MD.

It has been ofn~rved that as the amonth muscle of the airway wall contracts, the epithelial cell layer folds into many lobes such that the internal pesimnies of the airway remaing CORStuOL This suggests the presonce of a relatively ttiff layer ~ the epithelium. In normal hmnnn~ this inyes, lhe lamian prepda, is about 4 microns thick and is ~ of dease cnilagen. In asthmatics the entire alnvay wall Is thickened, but this layer in particular is about twlen as thick_ II has also been observed that asthmatic airways exhibit far fewer epithelial folds than nmmal airways which extend more deeply into the lumen, eausin8 greater airflow obsmgtion. We have developed a sim#e mathematical model for simniation of this heckling phanomenesl using finite element methods. The model airway is composed of a thick and compliant outer layer sotmonding a thin but stiff inner layer. Both layers are assumed to be incompressible linear elastic solids. The action of the lanonth muscle is simulated by applying the appropriate boundary conditions external to the onter layer. The model predicts that ianreastng the thinimess of the Inner layer ~ y decrease* the expected number of folds while cinmgos hi lise thickness of the onter layer have Uttie effert. Cbunging tbe stiffness (Yoang'$ mndnius) of the inner layex relative to the antet layer has an latatmediare effect. For example, doubling the thickuees of Ihe thin stiff layer redaces Ihe expected anmbor of folds by about 50% while doubling Its modulus only reduces It by ~Doot 15%. This model provides insight into the physical mechanisms behind the epithelial folding and describes quantitatively how reanodniing Of tha airway wall in asthma effects the shape of airway collapse. (Supported by grants firm the NHLBI (HL33009) and fae Freeman Fonodation.)

27 THE INFLUENCE OF AIRWAY WALL MECHANICAL PROPERTIES ON AIRWAY REOPENING PRESSURES AND WALL STRESSES D P Gaver IlL M.L. Perun and D. Halpern t. Department of Biomedical Engineering, Tulane University, New Orleans LA, tDepsrtment of Mathematics, University of Alabama, Tuscaloosa AL, We model airway reopening as a bubble that advances steadily with velocity U into a liquid-filled, collapsible channel. The lining fluid has Newtonian viscosity V, surface tension y, and depth H in the collapsed region. The airway wall is an inextensible membrane under tension T, supported by a foundation of elasticity K to represent parenchyma. The following dimensionless parameters govern the behavior of the system: Ca=pU/y, the ratio of viscous to surface tension forces; rl=T/r, the ratio of wall longitudinal tension to surface tension, and I"=-KH2Iu the ratio of elastic to surface tension forces. We analyze the system with physical and computational models. Physical models indicate that a critical transpolmonary pressure, PnrJ, must be exceeded before reopening can occur. This pressure depends upon the external traction provided by pareuchymal tethering, the airway radius, R, and the surface tension, u A dimensionless parameter, ,t=P,~m/(y,qr determines this relationship. Analyses based upon this result indicate that Pe,m--depandant surface tension may be necessary to stabilize the lung. Computational models predict airway wall stresses as a function of reopening velocity and mechanical properties. Increasing Ca results in increasing stresses. Decreasing /" reduces reopening pressures, but significantly increasing the wall shear stress. In contrast, increasing causes the reopening pressures to inctra~, but decreases shear stresses. We will present predicdoes of ranpening stresses for"RDS", "Normal" and "Cystic Fibrosis" lungs.

Supported by NIH grant HL51334and NSFgrants BCS-9209558and BCS.9358207

Several non-invasive, indirect meamiremeots such as plethysmographic airway resistance and resph-atmy system input ~ (Z/O) at low frequenciea (f < 32 Hz) have been used to infer changes in ait~eay diameten. However, none of these tedmiqans are capable of differeatiating between central and poxiphetal airways. We recently developed a method that provides an estimate of the scrinl distribution of airway diameters, i.e, as a function of Horsfield airway order, from measurements of Zin at high frequencins (100 to 2,000 Hz) (Habib et ni. J. Appl. Physiol. 77:554-566, 1994). The cmrant study was conducted to detmniv~ the reliability Of this technique by comparing airway diameters estimated from Zin data to measurements Of airway diameters measured from high resolation omapatod tumogrephy (HRCT) ~ made oll an excised dog lang. Since Horsl'teld airway ~ could not be determined frmn the HRC'r images, diamete~ at farmtimal residual capacity (FRC) were detetmiued as a function of their diameter at total lung capacity OI.C). Tairty-fanr cerrespo,~ing airway segtaants were klaatified in me T I C and FRC, CT images. The FRC di~nete~ which rengod botwera 1.5 8rid 16 ram, wece plaead into 6 gtq~ups a~ording to thek TLC di~tneltcn. TI~ F~C l~Nvay ~ totemalt'M to lheif TLC disnletet~ w ~ qmmtitatively coualstent with those obtained from anpasate Zin measmemems in 5 dogs, and 5 Of the 6 diameter gronp~ were ant significamly differeat from the dlamatess estimated from 71n measmemonts tn oae dog (Maon-Whimey Rank Stan Test. P < 0.02). We conclude from these results that eppropdato analysis of Zin data provides reliable eralm~tw~, Of the seslal disuibetion of airway'diameters for airways having diameters greater than 1.5 wan..

30 C A P I L L A R Y / E L A S T I C I N S T A B I L T I E S IN L I Q U I D L I N E D FLEXIBLE TUBES WITH AN EXTERNAL

FORCING FUNCTION. J.A. Moriarty, D. Halpern* and J.B. Grotberg. Departments of Biomedical Engineering and Anesthesia, Northwestern University, Evanston, IL, 60202, and *Department of Mathematics University of Alabama, Tuscaloosa, AL, 35487. In the present study we examine the effect that forced oscillations can have on capillary/elastic instabilities which lead to airway closure. These instabilities are a combination of capillary forces at the air-liquid interface of the liquid lining of the lung, as well as the elastic forces of the flexible airway wall. A linear stability analysis of the effect of an oscillatory airflow in the core of the airway, as well as a nonlinear stability analysis to include the effects of an oscillatory airway, will be discussed within this context. Results indicate that these forced oscillations can, in many cases, expedite airway closure. Supported by NASA and NSF grant CTS 901383.

Respiratory

Gas Transport

S-9

31

34

USE OF COMPUTATIONAL FLUID DYNAMICS FOR INTERSPECIES COMPARISON OF NASAL AIR POLLUTANT DOSE

CHAOTIC MIXING OF ACINARFLOW IN THE PULMONARYACINUS A. Ts'ud~. J.P. Butler. F.S. Hem'v. S. Habet. I. Godleski. =rid K. Ok9 I H9149 Univ., RiomechanicsInst., Boston, MA; City Univ., London, U.K.; Technion, Hail9 Israel

JS KImbell, GM Kepler, and KT Morgan Chemical Industry Institute of Toxicology, Reseamh Triangle Park, NC Laboratory rodents and nonhuman pdmates have been used to determine upper respiratory tract toxicity of inhaled materials. In some cases, such as the reactive gas fom'mldehyde (HCHO), site-specific nasal lesions occur in the nasal passages of beth rats and rhesua monkeys. Thres-dimenslenal, anatomically accurate, computational fluid dynamics (CFD) models were constructed to test the hypothesis that differences in the distribution of HCHO-Induced nasal lesions between rats and rhesus monkeys is a consequence of Interspacles differences in regional nasal airflow patterns. Airflow and uptake of HCHO in the nasal passages of the F344 rat (anterior portion only) and rhesus monkey were simulated using the finite element software package FIDAP (Fluid Dynamics Intemational, Evanston, IL). Model geometry was based on digitized outlines of the nasal passages from sequential cross sections through fixed tissue specimens. Simulated inspiratory airflow was compared with dye-streakline measurements in hollow acrylic molds of rat and rhesus monkey nasal airways. Simulations of HCHO uptake incorporated Insplratory airflow vaicolties, air-phase diffusion, and airway wall permeability (modeled as HCHO concentration set equal to zero at airway walls). Regions in which high levels of dose (wall mass flux) were predicted generally corresponded to locations of HCHO-Induced nasal tissue damage in both species. Further resaamh is needed to investigate the roles of various volumetric flow rates end airway epithelial type on regional dosimatry predictions. The construction of a CFD model of the human nasal passages is currently In progress. These models provide a means to extrapolate regional airway dose from laboratory animals directly to humans, while accounting for the significant anatomical differences among these species, to reduce uncertainty in extrapolating exposure-response data from animals to people.

We developed 9 computationalanalysis to examine the hypothesis that clutotie mixing oc.r in the rhythmicellyexpanding and contracting aiveolatedduct structure tnd tested that hypothesisexperimentally. Computationalanalysis: Time-dapendeat low Reynolds number flow was solvednumerically in a rhythmicallyexpanding geometric model of an aiveolatedduet. We found that for a given geometry the ratio of the alveolar flow (Q,) to the ductal flow (Qd) played a major role in determining the flow pattum. For larger Q,/Qd, the flow in the alveolus was largely radial. For small Q,/Q~, the flow in the 9lveoleswas slowly rotatingand the velocityfieldnear the alveolaropaaingwas complex with 9 stagnationsaddle point, typical of chantie flow stmctores. PerfonmingLagrengian fluid particle tracking, we demonstrated that in such a flow stnteture the motionof fluid could be highly complex, irreversible,and intrinsicallyunpredictable even thoughit was governed by simple deterministic equations. Animal Experiments: Ultraqow viscosity polymerizable silicone fluids of two colors were instilled sequenti9 into excised rat lungsand allowed to convent deep in the lung during mechanical ~eilhtian. The mixing patterns of the two colors were examined on lung sections 9 curing, and showed the stretehed-&-folded patterns typical of chaos. To quantify the extent of mixing, we computed the entropy in an analyticalapproximationof chaotic flowwith diffusinn.Them was a critical number of folds at which the extent of mixing 9 changed and 9 which mixingwas essentiallycomplete. In conclusion, chaotic mixing can occur during normal breathlng.in the pulmonary acinns and is fundamentallydifferent from classical diffusivemixing. Supported by NIH HLA7428and HL33009.

32

35

MODELING THE BRONCHIAL CIRCULATION: APPLICATION TO EXHALED ETHANOL PROFILES. Steveo C. George, Albert L. Babb, and Michael P. Hlastala Departments of Chemical Engineering, Medicine, and Physiology and Biophysics, University of Washington, Seattle, WA

DETERMINATION OF CARDIACOUTPUTFROM SINGLEBREATH CO2 WASHOUT Peler W. Sc'herer 1, j. D. Neff t, and G. N. Neufeld 1,2

The steady state exchange of inert gases across an in situ canine trachea has recently been shown to be limited equally by diffusion and peffusion over a wide range (.01-350) of blood sohbilities ([3b, ml.ml-t-atm't). We hypothesize: 1) the exchange of ethanol (13b=1756) in the airways depends on the blood flow rate from the bronchial (Q~) and p.ulmonary circulations (Qp), and 2) the surface area for exchange from the capillary bed through the tissue is less than the surface area of the lumen. We incorporated the structure o f a tracheal circulation model into a larger model that describes the simultaneous exchange of heat and gas in the entire airway tree, then used the model to simulate experimental exhaled ethanol profiles from human subjects. There a ~ three free parameters which describe the bronchial circulation model: 1) the total flow rate, 0ex (l's't), 2) the fraction of luminal surface area available for diffusion from the capillary bed, 7, and 3) the mean residence time in the capillary, '~ (s). The exhaled profile is insensitive to '~. We fit the model to the exhaled profile using ~ea as the free parameter for 7=1.0 and at T--0.1. The mean Q=x increases from 0.35 to 0.78, and R2 improves from 0.93 to 0.98 when y is decreased from 1.0 to 0.1.. We conclude that airway ethanol, exchange depends on the blood flow rate, Qex is a combination o f Q ~ and (Qp), and the presence of discrete capillaries reduces the surface area for diffusion from the capillary bed through the tissue. (suppoged by NIH grants HI.,24163 and HL09117).

1. Depaament of Bioengineering,Universityof Pennsylvania,PhiladelphiaPA 19104 2. Department of Anesthesia, VAMC, PhiladelphiaPA 19104 The single path model(SPM) of airway gas exchange has been very successful in simulatingsingle breath CO2 (SBCO2) airway washout under a variety of conditiens. Once airway structure and breathingpattern are fixed, the equation governingthis model contains only two other paramemm, the cardiac output QB and the mixed venous pCO2, which determine the shape of the washout curve. We fix lung structure by matchingthe normalized phase Ill CO2 washout slope vs. exhaled tidalvolume between a given subject and the standard Weihel lung using a le,~t squares algorithm. Using the SPM for this scaled lung, we then compute the volumeof CO2 evolvedper breath and the normalized slope of phase II1 of the SBCO2 curve over a range of cardiac outputs and mixed venous pCO2 valu~ at the same tidal volumeand frequency breathed by the subject. Using the simplex method, we then minimizethe absolutevalue of the dilferenze h e t w ~ the exparh'nantally measured valuas for these parameters and those oomputed by the SPM over the 2-D QB, pCO2 space. Preliminarycomparison of our noninvasiveresults with there: found from catheterized subjectsshow good agreement.

35a ~'qIRA-AIRWAYGASTRANSPORTDURINGAL'I'HT,NATIVEMODESOF Vt~/I'K.A'I'ION N. Gavndy, D.M.~ ; O.P. Gav~II1,and J.B.~ TerhnianH,=faIsrad; ~ & Bicmed.Eng. NonhwestenaU. ChicagoIL; Bicmed.Eng.,TulaneU. New Ode9 LA33 O Z O N E TRANSPORT IN H U M A N M A L E A N D FEMALE LUNGS Bush, M., Ben-Jcbria,A, and Ultra9 J.g. Pennsylvania StateUniversity,UniversityPark, PA 16802 National Ambient Air Quality Standards have been based primarily on data obtained from healthy male s'ubjcots. However,recent studies eslng female sobjeets have indicated that women may exhibit larger responses to ozone than males at equal total dosage levels. We hypothesize that observedgender differences are a consequence of lung anatomy differences between men and women. Since the cotuinettag airway region of the lung "seruha" the majority of inhaled ozone, the volume and surface of this region greatly affnet ozone tramport. To sludythe ialhenco of lung aantomy on ozone dnsimeW/we have measured the longitudinaldistribulionof ozone in the lung using the Ixfl~-response me0md. Additionally, the velumc of the conducting alnvay region was estimated using the standard N2 wuho~ method of Fowler. Data ebtained from a populationof ten male and ten female mbjnets indicate a strong correlation between ozone transport in the lungs and the conducting airway volume.

t n e a - ~ w gas a s m ~ (Q) ~ a g r~gh r - ~ q m ~ Valance 6WV), r~gh r-~qumcy Vi'mntionVmtiladon(HFVV),andCon,c~t Row Vanalatiou Q = A'D,# ~/~V (CFV)may ha moddledby the modred l~ckcquatien: Where A isthelocal~a. De.I isthelocalfffectivediffmivity,and o33/0VisthelocalvoheneUic gmcom:annfiongradientWe mcastaedlocalinlra-ai~vayAZDegbytheholmd i ~ u n method sod by die SingieBreathInertGasWashout(SBWO)mtshod,and r by the SBWOmethod md by mmmnngtheintra-mwaygmconcmu~iengradimtnsingspadally~ om~nuin dm~e lumm ca~e~. tO0-- WithIfVV

g~

s

5o ..

0

5

10

T~me[s~)

15

0

0,I

,

,

0.2

0.3

F..x~ed volumepit=]

Bolm Dispe~en Method SingleBreathWasbcot ~ The dam fr~a these~imimmts show central-to-pmpharaldism'b~ienofeffeaiven~s~(~'V, ~ ' V md I-nWV,m~e,~vely.

S-IO

Respiratory Tissue Mechanics I

36

39

T H E MAGNETIC TWISTING CYTOMETER: A NOVEL T O O L T O PROBE RECEPTOR-SPECIFIC TRANSMEMBRANE MECHANICAL COUPLING AND CYTOMECHANICS Ning Wang, Jeffrey J. Fredberg~ Donald E. lugber Harvard School of Public Health, Children's Hospital, Boston, MA 02115

EFFECTS OF BRADYKININ ON CYTOSKELETAL MECHANICS OF ADHERENT HUMAN AIRWAY SMOOTH MUSCLE CELLS R.Hubmayr, S.Shore, J.Fredbefg, E.Planus, R.Panettieri & N.Wang Mayo Clinic. Rochester, MN; U. of Pennsylvania~ Philadelphia, PAl & Harvard School of Public Health. Boston, MA

We have recently developed a magnetic twisting cytometry technique which allows us to apply controlled mechanical stresses to specific cell surface receptors using llgnudcoated ferromagnetic microbeads and to simultaneously measure the mechanical response in cultured living cells. Using this technique, we have shown that: (1) ,81 integrin receptors mediate mechanical force transfer across the cell surface to the cytoskeleton whereas other receptors (acetylated low density llpoprotein receptors) do not; (2) the measured stiffness increases with the applied stress to integrin receptors; and (3) different cell shapes result in different mechanical responses. We have also shown that different integrin subsets differ in their ability to mediate stress transfer across the cell surface. Surprisingly, We found that the urokinase receptor, a nontransmembraue cell surface receptor, also mediates stress transfer across the cell surface. In all the cell types that we have examined, we found that stiffness increases as the level of applied stress to integrin receptors is raised. Thus the stiffening response appears to be a general feature of many cell types. Finally, a dynamic change in cell shape during cell spreading was accompanied by coordinated increases in stiffness. These data suggest that the magnetic twisting q.~ometer may be a nsoful tool to study molecular basis of transmembrane mechanical coupling to the cytoskele~on end cytomechands,

Since the nOn-contractile cytoskeleton (CSK) of smooth muscle cells is thought to be connected to sliding filaments, its stiffness ought to increase during muscle activation. This hypothesis was tested in relaxed and bradykinin-activated adherent human airway smooth muscle (HASM) ceils. The CSK of adherent HASM cells was deformed using magnetic twisting cytometry in which ferromagnetic beads hound to apical in~egrin receptors are rotated with magnetic fields. The ratio of applied stress to angular strain defined CSK stiffness (S). Cell shape was manipulated by varying the concentration of collagen 1 matrix from low CL, 10ng/well) through intermediate (1, 100ng/well) to high (H. 500ng/well). Cells grown on L were round and clustered; those grown on H appemed fully spread. S of unstimulated HASM cells was sigeificandy greater at H (126=1:16dyne/era2) than at either I (954-7 dyne/~a~2) or L (434.3 dyne/era2). Bradykinin 10~M incw.ased S under all expcrimcotai conditions, However, the htadyHnin-indaced stiffemng responses were most pronounced in cells grown on ff matrices in which S increased to 2194.18 dyn/cm= (an 854-16 % incrco~), less pmnourgaxl in I coils in which S increased to 1514-17 dyn/cm 2 (a 59d:14% increo.~), and least pronounced in L cells in which S increased to only 57:f.2 dyn/cm2 (a 33+11% increose). We conclude that bradyidnln increases the stiffness of the HASM CSK in a shape.decedent manner. Vlewed from a purely geometric perspective, this observation may bc attributed to shape-d~ndeat differences in aerie/myosin overlap. However, it remains m be de.temfined whether round cells express an adequate number of bradykinin receptors, whether they have functioning chemomech~nicai signal wansduction pathways, and whether actinand myosin are assembled as sliding filaments in them. (Supported by HL-49788, HL-02647, HL-33009)

37

40

MECHANICAL FRICTION AND CROSS BRIDGE CYCLING RATE IN CONTRACTILE TISSUES J Fr~serg, KA Jones, S Raboodi. YS PrakaslL SA Shore, G Sieck Harvard School of Public Health; Mayo Clinic Fo~ndatinu

THE EFFECTS OF STRETCH ON THE LUNG CELL LINE A549 Sarah BaptiSbNguyca ~, Lan'y V. MclntL~ t , Burton Dickey ~, m:d Joseph R. Rodarte ~ ~Cox ~ for Biomedical Engineering, Rice Unive~ity. H~ston, TX ~Baylor College of Medicine, Houston, TX

ARbeugh it is a bypredua of tone maintenan~, frictional stress modulates imlmt~nt facets of contractile tissue function, Considerations of cress bridge kinetics suggest that, following the onset of satstained stimulation, recn~trcent of numbers of cross bridges attached ao:ounts for the development of tissue clastatme. E, while conversion of rapidly cycling cross bridges to slowly cycling latch bridges accounts for the time course of tissue hystereaivity, ~. Frictional stress can be given as the product of t/ssneelastic guess and tlf~e hysteresivity. Changes in frictional stress uttfihotable to activation of contractile tissues would be inteq~rotable, thcrofore, as the product of changes of cross bridge numbers and their apparent cycling rates. Temporal changes in unloaded velneity of shut.tung and netin-netivated myosin ATP utilization have been ~ and are shown to be closely related in time to changes of~ and the product ~E respectively, These data iedicste that meteholic and mechanJc~ energy digs/patina are l/eked f~adamentally at the lovel of the contmC~le protein ~cling and, as such, support the idea that rig macrascele hystexr rt tracks underlying changes of cross bridge cycling rate continuously in time. In doing so, the~ data represent an indapendont line of evidence that tends to support the existence of the latch state and reveals one of its more relevant manifestations. Taken together, these results indicate that the rate of actorayesia detachment (in contrast to the classical viscous stress) seems to be of n ~ o r imporlance in the frictional stress developed within contractile tissues, (Supported by NIH HL33009.)

As oae ~the~ the lung changes volume, saetching and stimulating the ~cretloa of su~ac~st, a mixtorr ~ I~osphollpids and suffectsnt-specificpmtclne, This sur[actant is vitalfcr maintenance of lung complianee and for ~ gas exchange. O~ gnal isto anders~md how strotchaffects lung cell se~ction aod metal~ism. For our initial studica, we chose to study the effects ~ stretch on p ho~hollpid ~ t i c a a~d ~ r e gone of the ltmg cancer cell Line A549, which m~vhologicatty resemble alveolar type II cells b e c x ~ they coatsin rs'gauelles tcsemhting l,m~llarbedi~. Ourexpets appasates exposes ceils te uninxinl stretch. A549 cells are seeded at a density of I mllIoo cells per 18 cma 0a fihronectin-r polydimcthylslloxa~ membranes. Once Ihey w.ech cooflnenee, the cells mc subjectod te 24 h0tws of stretch, fluid motion, ~" no motion. Cell damage Ca"~ v a l is mO0itow.d by esnay of ~ dehyclmgenase (LDH) netivi~y of the media exposed to the celh. TI~ Idx~pholipids ~a~ extrac~ fi~m cells ~ d m,-~i~. Scpa~ted by ~ - I ~ y e s ~a~:aaatogra~y gnd ~ e.ifa~ by de~t~eak-t:y. LDH lcve~s in coe,tmLs lind stre~bed samples wege the ~en~ne,ladicsIng that cells are aor removed ~ ~ mcmlran~ by elongation (Lc. for 10% stretch and 0.18 Hz, sl~stch = 15.6:t~.9; mofio~ couwol 14.8:/:1.2; atahomwy control= 15.&kl.0; media control. 15.6:1:0.9;mean .-~EM for u = 6 e ~ ) At 10% stretch and 0.18 I-Ix,cell n~thr and cell phosphoIpid comp~ifion me net affneted(nffi5). The cells also do not dlign in tmpoeseto~fislevelofsh'etch. Media phosphoIpids, howev~, do change I~ c ~ i t t o n at this level of s~ctch. At 10% stretch and 0.18 Hz, p h O ~ d y l e ~ = m t ~ keels ate elevagod in sti'el~ed and motion control media when comp~aed with sta~iom~y co~n/cell media ( u . 5). Tbe effnets of o~e~ umplRudes 8nd fmquenc/~ o~ sUetch on Ihese ~lls will also be d i s c ~

38 RESISTANCE OF CELLS TO SHAPE DISTORTION: A MICROMECHANICAL APPROACH. Dimltr~je Stamenovi~, Jeffrey J. Fredberg, Ning W u g , and Doneld E. lugber. Ek~ton Uaiverdty, Be*ton MA 02216, H a r o l d School of P u b i c Health, Child~r Booplt~ ud R~d Medlr School, B~ton, MA 0211& Mecba~c~d force~ acting on, or generated by cells, influence organ functions. For example, temted extraeelluinz matrix of lung pnsenchym& is the ~Jdfoldln S upon which pulmanary cells of all typr sdhetr spread, contrast, locomote, proliferate mad remodel. On the other hand, smooth muscle coutracsion alters pul~onaxT .mscuinr ~aiJtance, airway r.~-txnee, lea S tuque te~tgaee, and lung einatidty. The shtatht:d approach in cell mechgnlea is ha~ed upon the continuum hypoth -~-. l a I ~ study, however, we comfider the impIeatio~ of am ~lteruative approach that empluutises the discrete nature of etrr bearing elements ia the cell u d is lanted on imown pgopertlcs of the eytmkeleton (CSK). We have noted prevlomdy that tea~gfity aschitect~ze Jcemi to gaptnse essenti~ qnldlt6five fcat e x u of CSK sha@r distortion in adhegeat r Hege we extend throe qulitativr fr into 9 formal mictmtructursl a m ~ . On the baals of tlmt x~slysis we •ttempt to identify unifying prlaelpleJ that might ~ m d ~ e shape 8t~hl]ity of the CSK. Per dmplldty we f e c u on 9 tem~egrlty straetm~ conta~nln B ~ c Slgld s t r a k interconnected by twenty fo~x linearly ehutle ~ b l ~ . C~Mm use mm~med to egny iaithd turn,on ( ' p r e l t r e u ' ) , coeatefl~daneed by e o m p ~ n of struts. The system w ~ then 8mhjected to nn;=~d atsete.klng. Udag the prindple of virtan] work, stretehlng force vezsus e x t e ~ o u and stiffnm versus force sehttloaJhitx were eeleu~ted for different p ~ d ~ g e s . The s t i ~ e u was found to ineses~ with i n e ~ e 2 ~ p m t r e ~ and, st 9 Kive9 p m t ~ , to ~ n ~ sppro~m&tely hae~]y with i n ~ s t r u t t i n g for~. Tkis behavior k c o . t e n t with olmrv~tions ia llHag smooth muscle cells and andothelin] cells e ~ to shear s t r e ~ s . A ~smlt of peat~uls~ i m p o ~ a a ~ ~ that shape stsl~ity of the cell depand~ more upon the p ~ , t r e ~ ia the ezl[ lind upol~ CSK sgekltect 9 ~ it doe* upon the stiffae~ of the individmd CSK fdsmcnts. (Supported by: NIH HL 33000 and CA4554g, ~ud NASA NAG-9-4g0.)

Respiratory

Tissue Mechanics

II

S-11

41

44

PULMONARY MECHANICS IN DOGS WITH CHRONIC UNILATERAL PULMONARY ARTERY LIGATION. Kelly, SM, Bates, JHT and Michel, RP. Mealdns.Christin Labs and Dept of Pathology, MeGill University, Royal Victoria Hospital, Montreal, Quebec, Canada.

NONUNIFORM EXPANSION OF CONSTRICTED DOG LUNGS R.D. Hubmayr, M. Hill, and T.A. Wilson Mayo Clinic & Foundation, Rochester, MN; University of Minnesota, Minneapolis, MN; USA

After chronic unilateral llgatioa of one pulmonary artery, the affected lung is perfased with systemic arterial blood and inspired gas is diverted to the contralatasal, non-llgated lung to optimize gas exchange. This study was designed to determine the mecbankm responsible for this shift of ventilation. We measured pulmonary resistance (RL) and elastance (EL), and minute ventilation (~/E) in the right and left lungs of s ~ dogs before, and 6 months after, left pulmonary artery llgation. We also examined the effect of CO2, atropine and isoprotereanl on RL and EL. We found, in the lung with the tigated pulmonary artery, that ~E was slgnificandy reduced. RL and E L were incrcasod but unresponsive to CO2, atropine, and isoprotereanl suggesting that bronchoconstrictlan was not responsible for the redirection of ventilation. To determine if the mechanical properties of the lung pasenchyma were rh~nZed, we measured EL, RL and airway (Raw) and tissue (Rti) resistances in five of the ligated dogs and in five control dogs. Separate measurements of left and right lung airway flow, tracheal pressure and alveolar pressure were made during mechanical ventilation at fi'equendes (f) between 5 and 40 brcoths/min. At all f, llgatod dogs had higher left (ligated) lung E L and Rti and lower right (normal) lung Rti but similar E L compared with the respective lungs from control animals. Raw was the same in both lungs. We conclude that, with chronic pulmonary artery figation, 1) the llgated lung has increased E L and RL, the latter due to increased Rti 2) there is a reduction of Rti in the contralateral lung, and 3) the shift of ventilation is a consequence of the changes in mechanical properties of both the llgated and contralatoral lungs.

The pamnchymal marker technique was used to measure regional tidal volumes of samples of lung parehchyma in four open-cbested supine dogs. Radio-opaque markers that had been implanted in the lower lobe were tracked by biplane videofluoroseopy during sinusoidal mechanical oscillation at tidal volumes of 20% of total lung capacity and frequencies of 1 to 40 breaths/minute both before and after methacholine was administered by aerosol. The volumes of tetrahedra with apices at four markers were computed, and sine waves were fit to the data for volume vs time for each tetrahedron. The ratio of mean lung volume to mean airway pressure decreased by 10% to 45% after exposure to methaeholine. Dynamic lung elastance and resistance of constricted lungs were larger than control and both were frequency dependent. Regional elastance and resistance varied considerably among tetrabedra, and these were also frequency-dependent. The data were fit by a model in which tissue elastance was uniform and nearly equal to elastance in the control state but small airway resistance was high and variable. We conclude that the lung contracts under bronchoconstriction but that the increased dynamic elastance and resistance of constricted lung are primarily the result of non-uniform increased airway resistance at the level of the terminal bronchioles. (Supported by HL-49788)

42

45

A N O N L I N E A R DISTRIBUTED M O D E L O F L U N G TISSUE ELASTICITY LH.T. Bates and G.N. Maksym. M~kins-Chrlstio Laboratoriesand Delmrtnmat of Biomedical Enginee~qg, McGillUniversity,Montreal, Quebec, Canada.

LOW-FREQUENCY R E S P I R A T O R Y I M P E D A N C E IN H E A L T H Y I N F A N T S Z. Hantes, M. Hayden*, F. Pettlk, P. D. Sly* Dept. of Medical Informatics and Engineering, A. Szent-Gy6rgyi Medical University, Szeged, Hungary, and *Institute for Child Health Research and Princess Margaret Hospital, Perth, Australia

elastic ixope~e6 of kingtissueare reflected in the relationshipbetween lung recoilpressure (P) and inflationvolume 00. This felatiowhipis otiea expressed in terma of the empirical reistinmhlpV ~ A- Bexp(-KP), which~ovatthatthethtsuebeconmsver/~Jffat isrgeV. Iris widelyaccepted that this elastic nonlinearityis produced in large pert by the fact that lung tissue is composedof a networkof oollagettand elastinfibe~. At low V most of the collagentRJenlare flaccid,leavingthe relativelyexte~ioloelastinfibers to generate stress. As V increases, the much stiffer collagen fiben become strltighteaed~ so take over the load bearing role. We have developed an analyticalmodelof lung tlmmmelasticitycompogedof 9 diatn'butinnof collageneisstlafiberImlnammgedla i~riea. Ati the elastinfiben itre of ideatlcalelasticltyand tmstressed length. ~ collagenfibree, however,are ioextensioleand assume a distn'butinnof lengths. The bulkelasticityoftha tissueat my volumeis thas due to the fraction of the colingeafibers that have reached their "stoplengths'. The shapeoftha stop lengthdistributionfunctionthus determines the overall poV behavior. The stop length diatri~on function corresponding to the above empirical relationshipbetween V and P is 9 decreasingexponential. Thus, according to this model, lung tissueelasticityis determinedby the elastance (E) of the unit elastic fiber and the amptimde (D) and tare-constant (u) of the exponentialstop length distribution. DecreasingE (correep~dingto ~ of ehuainI inereuea K but dosi not affect total lung capacity O'LC), wheretodecre~lagu ( ~ to lactating collagen)increasesK and decreau~ TLC. We slmcuistzthatdifferentialeffects on colingenand elastinsecount for the characteristic difference~ io A, B aod K seea in emlNayseumand fibrosis. (Suppertedby MRC Canada, FRS Quebec, RHNCE and J.T. CostetioMere. Res. Fund.)

43 EVIDENCE OF AN ARTIFACTUAL INCREASE IN TISSUE RESISTANCE DETECTED BY COMPARING IMPEDANCE MEASUREMENTS WITH I~WO DIFFERENT RESIDENT GASES

K.R. Lutchen. B. Suld, F. Petak, A. Adamicza, End Z. Hantca Dept. of Biomedical Engineering, Boston Univ., Boston, MA, USA. and Inst. Exp. Surgery, Szent-Gy~rgyi Medical Univ., Szeged Hungary The large increases In tissue dEmplng previously reported during bronchoconstrlctian may be artifactual, end 9 result of airway Inhomnganeitles. To help resolve this iSSUE, in 5 rats we measured lung Impedance ZL from 0.23 - 12 ~ with the lungs equilibrated on room air end then on 9 mixture of 80% Neon-20% Oxygen (NeOx). The ZL was acquired before End after stcady-atate bronchoconstrlction caused by methacholine (MCH). The rats were opon-chestnd within 9 body box. A broadband flow signal was delivered via a piston pump. The signal simultaneously ventilated the animal while compensating for the box gas compression. Pa~tionJng of airway End tissue properties WESachieved by fitting ZL with a model including an airways resistance, Raw, airway InErtsnca (law), tissue dEmping ((3), end tissue elestanca (H). The latter two parameters govern the frequency dependence of resistance (R) End elEstanca (E). The NeOx is a more viscous gas. HENCE,if InhomogeneRieswere not significant, the tissue propartleE would be Independent of the resident gas while the Raw would scale as the ratio o f viscosities. After MCH, the Raw end G increased markedly (60-100%) while the increase In H was smell (<20%). Dudng control the H(~Ox-to-~r Raw ratio (Raw./Raw,) was 1.22 + .06 while the (G./G*) End (I-t./1-1.)were both 1.04 • .04. This Indicated little or no Inhornngenetiy. During MCH, the (Raw/Raw*) stayed 1.23 • .065, the (H/H*) stayed 1.04 ~-.04, but the (GN/G*) Increased to 1.18 • .1. This is evidence that during MCH Induced constriction, 9 po~on of the Increase in tissue dEmping ((3) end lung E is artifactual end 9 ~onsequenca of the Increased frequency dependence associated with airway InhomogeneRles. (NIH HL50515 and NSF BCS-9309426).

The need for apneic conditions limits the availabilityof respiratory impedance (Zrs) data in humans at low frequencies including the spontaneous breathing rats. W e exploited the apneic phase elicitedby the Hering-Breuer reflex at the end-inspiratery airway occlusion in 5 sedated infants aged 9-12 months. Through use of a computercontrolled p u m p and solenoid valves, the supine infants were inflated through a face mask three times to a transrespiratory pressure of 20 cmH20, and the mask was then connected to a loudspeaker-in-box system. Small-amplitude pseudorandom oscillations (<1 cmI-120 poak-te-peak) containing noninteger-multiple components between 0.5 and 20 Hz were applied for 6 s. Four consecutive measurements were made in each infant, and the Zrs spectra were averaged. ARer omission of the frequency points corrupted by the heart frequency and its harmonics, the Zrs data were evaluated on the basis of a model containing frequency-indepondent airway resistance (Raw) and inertance (law), and the viscous damping (G) and elastance (H) parameters of the constantphase compartment of chest wall and parenchymal tissues.The measured Zrs values were consistent with the model up to 15 Ha, and the average fitting error was 0.905_-0.14(SD)cmH~O. The following parameter values were obtained: Raw=9.84~.4 cmIlaO.sil,law=0.064L'O.014 cmHsO.s2/l, G---30.2.'k4.9crnH20/l,H = 138.+.59crnl~O/l. The tissue hyateresivity(G/H) values were 0.249-L-0.103.Our results indicate that in short apneic periods evoked by the Hering-Breuer reflex reliable data can be collected to partition the tissuE and airway impedances in sedated infants. - Supported by grants NH&MRC 941252 and O T K A 2675.

S-12

Issues in Lung Macrovascular Blood Flow and Tissue Exchange

46

49

ABSENCE OF A TRANSIT TIME LIMITATION TO OXYGENATION AT HIGH BLOOD FLOWS IN ISOLATED PERFUSED RABBIT LUNGS. I. Ayappa, L.V. Brown, P. M. Wang, P. Houtz, E. N. Bruce and S. J. LaI-Fook Center for Bbmndioal Engineering, University of Kentucky, Lexington, KY 40506.

EFFECTS OF PERFUSION ON UREA EXTRACTION IN ZONE 3 SHEEP LUNGS R.J. Roselli, T.R. Harris and N.A. Pou. Oepartment of Biomedical Engineering and Pulmonary Circulation Center, Vanderbl!t University, Nashville, TN 37235.

Capillary transit time measured directly from tluoreecent dye transit through the subpleurul mlcroclrcutatlon of Isolated periused rabbit lungs using video microscopy averaged 0.6 s at basal blood flows of 80 ml. rain"1. kg-1 and Increased with a decrease In blood flow (J. Appl. Physiol. 72: 2420, 1992). Similar values for capillary transit time were obtained by measuring capillary blood volume by the CO dtltusing capacity method and dividing it by blOOdflow. Further studies showed that capillary transit time decreased to 0.14 + 0.01 (SD, n = 5) s as blood flow increased to 480 ml. min"1. kg"1, approximating the cardiac output observed in humans dudng heavy exemise. To study the effect of these short transit times on 0 2 saturation, we connected two isolated pedused lungs in sedes. Blood from one lung was 02 dasaturated by ventilating with 7.5% CO2 * 92.5% N2, and then saturated by the test lung ventilated with air. Ventilation was increased with blood 8ow to maintain a normal I~lmonary arlariat-to-venous PCO2 difference (6-t0 mmHg). 02 saturation was 100% lot blood flows of 80, 240 and 480 ml. min-1, kg-1 with no significant alvsolar gas-to-end capillary blood 02 gradients (A-aDO2). Mild edema of 30% increase in lung water in these lungs had no deleterious effect on oxygenation. Thus there was no evidence of a transit time or diffusion limitation at blood flows near values observed dudng heavy exemlse in humans. Only severe edema and hemcrrttaging at flows near 640 rnl. mln"1. kg"1 (n - 4) produced slgnitioant A-aO2 gradients (40-55 mmHg) and 02 desaturation (80-90%), consistent with observations repealed in some thoroughbred horses after a race. Supported by HL 36597 and HL 40362 from NHLBL

We examined the effects of pen'usion rate on urea extraction in/n .situ sheep lungs perfused with s mixture Of autolngous bloOd and Dextren 40. Lungs were peri'used under zone 3 conditions by keeping venous pressure positive relative to alneay pressure. Flow was vadnd in 3-6 steps between 1-5 I./min and multiple indicator dilution (MID) data (~4C-urea, 3H-water, S~Cr-erythrocytas and nSl-albumin) ware collected at each step. The Effective Diffustvity (~/DS) and Crone models (PS) ware applied to the MID data. Blood flow correlated ,needy with the difference between upstream and downstream pressure, suggesting full recruitment at each flow. MID results were grouped into low flow (< 2 Umin) and high flow (> 2 IJmin) categories. PS increased significantly with flow (5.14 vs. 8.66 mUs), while ~/DS did not change (8.91 vs. 8.69 ml/~/s). Urea integral extraction (E$) generally decreased with increasing flow, but in a few studies E, appeared to be independent of flow. However, in those studies the extrectiowlime relation for urea (E(t)) changed significantly as flow changed. E(t) for high flows was initially higber than E(t) for low flows, but this trend was reversed at later times. Since E, only accounts for extraction up to the peak of the reference curve, it may underestimate total urea extraction at low flows. Therefore E, may not change with flow, even though total extraction increases with decreasing flow. Compsdson of the shapes of extraction pettems for urea (coqcave up) and water (concave down) indicated that urea did not behave as a flow-limited trecar. We conclude that simple application of the Crone model using E, is unreliable II flow vades, and a better model such as the Effective Diffusivity model st~ould be used to estimate permeability. Suppoded by HL 51234.

47

5O

DISTRIBUTED iN TIME AND SPACE RECIRCULATiON MODEL OF THE PULMONARY ENDOTHELIAL UPTAKE OF 18F.FLUOROCAPTOPRI L Audi, S.H., J,H. Linehan, G.S. Krenz, D,P, Schuster, J, Markham and C.A, Dawson Marquette University; Medical College of Wisconsin and VAMC, Milwaukee, WI 53295 and Washington University, St. Louis. MO 63110

VISUAL REGRESSION OF MULTi-PAKAMETER LUNG VASCULAR-TISSUE EXCHANGE MODELS Sorel Bosan,ThomasR. Harris Department of BiomedienJ Englncet'tog, Vandethili Unive~ity, Nashville TN,37235

Dynamic imaging of the lung with Positron Emission Tomograpby (PET) after the administration of 18F-fluorocaptopdl has been used to quantify the kinetics of pulmonary ondothelial Angiotensin Convertin 8 Enzyme (ACE) using a compartmental receptor model (Schuster et al. J. Anol. PhvsioL 78:1169-1178, 1995). In the present study, we developed a distribu ted-in-timc-and-spac e model for evaluating the pET-generated 18F lung residue curves following 18F-lianrocaptopril infusion. The model accommodates the recirculation of 18Ffluorocaptopril through the systemic circulation and back to the lungs over the time course of a typical PET data collection period (20-25 mie). Although captopril exists in cis and rrans conformational forms, the model assumes that only the rrans form is the ligand for ACE. Within the systemic circulation portion of the model, there is cis to trans conversion and the tran~ form is extracted. As a partial validation of the model concepts, we earned out single pass bolus injection multiple indicator dilution experiments in isolatnd dog lungs and analyzed the data using only the lung portion of the model. We then used the ligand-onzyme kinetic parameters estimated from the isolated lungs in a simulation using the recirculation PET protocol. The resulting simulated data compared well to those measured using PET in intact dogs suggesting that the model is consistent with data from both the residue detection method (PET) and from the single pass iulet-ootlet method, and that such a model might be an alternative to the mare commonly used compartmental models.

Our goal was to apply computer visualization to the process of identifying the parameters of palumnmy vascular-tissue exchange models. Several regression routines exist for carrying out this analysis automatically. However, these methods are difliunlt too u.~ or yield unsatisfactory results in situations where a) the model does not describe the data vecy well, b) the data is very noisy, or c) there are underlying difficultins to Sanxitlvity and identifiabiBty. To apply visuafization, n-dimensional images of the sum of squared errors, as well as tho cnofl]cimt of variation as a function of the model parametam need to be generated. To illustrate the methodology, two c a p l l l m y ~ models which have been used for lung iodicator-dilution studies (Sangren-Sheppaed and PermeabiityDiffusivity) were analyzed. We applied this method to ~4C-urca ~ r t data from bointed peffused dog lungs. Our resells indicate thut tn l ~ troublesome r162 visust regreasinn can be nsed ns a means of obtatoing aneerate initial pmameter e~in~ates for automated procedar~. In more problematic situations, where convergence of the parameters is impossible to obtain, visual regression can be used as an altemetive to the amomated rontiue~ and still enable the exUact~n of some information regarding the paranmtam. Finally, otw re~lts also indicate that the images geoumtcd for visunl regref~on can also be used in aiding the global ldontifiabillty studies of modeh.

Supported by the Amcdenn Heart Association of Wiscoosin, Department of Veterans Affairs and NHLBI grants HL24349 and HL48550.

Acknowledgmem: This work was supported by NIH Grant RR06558-03

48 EFFECT OF HYPOXiA AND FERRICYANIDE ON THIAZiNE DYE REDUCTION AND UPTAKE IN A CELL-COLUMN MODEL OF THE PULMONARY MICROVASC ULATURE. L.E. Olson, M.P. Merker, R.L. Bongard, LH.Liuehan, C.A. Dawson Marquette Uulversity, Medical College of Wisconsin, and Zablocki V.A. Medical Center, Milwat~kue, 97I 53233. Methelyue Blue (MB) and Toluldlue Blue O (TBO) are two thiazine dyes which ate reduced and sequestered by lungs and cultured pulmonmy enduthelium (J. Appi. Physiol. 77:1480,1994, PASEB J. 8:A9i7,1995). Previously, we used a flow-through celi-coinmn system to characterize the f6netica of this redaction and uptake process in endothelium (FASEB L 9:A719,1994). These studies suggest that MB orTBO is first reduced by an endothelial surface rednctsse and the u:daced dye either auto-oxidizus in the perf,c~te or is taken up by the endothelium. In this study, we tested whether hypoxia would affect blue dye reduction or uptake by deceasing dye auto-oxidation. We also studied a potential competitive inhibitor of the endothelial surface rndactase, Potassium Fet'ricyanidr (FCN). Bovine pulmonary arterial endothelial nulls grown to confluoncr on gel-coated Biusilon beads (200/an (fla.) were placed in a chromatoBraphy column (3 c m x 0.7 cm din.). Multiple indicator dilution studies using FITC.dexmm as a referenee indicator and either MB or TBO were performed at basoltoe conditions, d u r i ~ hypoxto or to the presence of 200 ~ FCN, and after return to baseline. Effluent dme-cow.enuadon profiles were obtained by measuring optical absorbanee using a flow-through optical deusitometer. Perfusion of the cell-column with a hypoxic medium (PO2=6 mmHg) and return to normoxic medium did not change reductionand uptake of MB or TBO. FCN diminished the amountof MB or TBO removed from the medium, howeverdata that it was not a simple competitive inhibition. We concinde that auto-oxidsdon b not a nile-limiting step in the removal of MB or TBO by the cell.coinmn. Also, FCN can inhibit ME and TBO rednetion and permanent uptake, but the me*-hani~n is not dmple competitive inhibition. Supported by HI..24349 and Dept. of Vet. Ai~tirs

N o n l i n e a r D y n a m i c s a n d R e s p i r a t o r y Control 51

54

PRENATAL AND POSTNATAL BREATHING DYNAMICS IN SHEEP, BABOONS AND HUMANS

UPPER AIRWAYINFLUENCES ON RESPIRATORY PATTERN VARIABILITY E. N. Bruce Center for Biomedical Engineering, Unlverslty of Kentucky, Lexington, KY

Hazel H. Szeto, Department of Pharmacology, Cornell University Medical College, 1300 York Avenue, N.Y., N.Y. 10021 Spontaneous fetal brenthfng movements have been documented in both humans and animals. These movements are highly irregular and often interrupted by long apneic intervals. Apnelc pauses are also seen postpatally, especially in premature infants. The control mechanisms underlying developmental changes in breathing pattern are incompletely understood and may differ among species. Despite their random appearance, we have found evidence o f fractal scaring properties in ovine fetal breathing, suggesting a basis for nonlinear control. In this paper, evidence will be presented showing that fractal scaling is ubiquitous in respiratory cycling o f fetal and neonataI animals and humans,

and that similar maturational changes are observed in the ovine and primate species. Maturation is associated with a more correlated breathing signal, and may indicate a less flexible, or more rigid, control mechanism. Dynamical analyses using delay plots o f consecutive breath intervals from fetal lambs

revealed continuous motion among a large number of unstable periodicities. Large deviations from this motion were generally corrected immediately. Perturbation of the system with morphine resulted in a highly stable periodic behavior. Together, these findings suggest that respiratory cycling in early development may have some of the characteristics of a chaotic system.

S-13

The respiratory pattern of the anesthetized, tracheotomlzed m t is hlghly variable from breath to breath when end-explratory volmne (BEY) is reduced and very regular when EEV in elevated. We hypothesized that tracheotomy bypasses upper airway (UAW) mechardsms used to regulate alrtlow and regularize the pattern. Consequently we measured respiratory pattern variab~lty in t l anesthetized rats ( I I 0 0 m g / k g urethane} with intact UAW using a head-out plethysmograph, at various levels of continuous positive (CPPB} and negative (CNPB} mean plethyemograph ("box"} pressure. With CPPB of 3 or 6 cm I~O (presumed to reduce EEV] respiratory pattern variability increased as assessed from phase plots and coefficients of variation (e.g., of tidal volume). In contrast to tracheotomlzed ardmais, with elevated EEV (CNPB = -2 cm H=OI resptrato~ pattern was still variable, but less so than with CPPB of 6 c m H=O. Thus breathing through the UAW Increased respiratory pattern variability rather than reduclng It. We attempted to determine whether the ettucture of the valdabtlJty was slmilar with CNPB and CPPB. Estimates of correlation dimension were Inconclusive. Recurrence plots often seemed vlsua~y different but quantitative measures (% recurrence, % determin~ro, entropy) were not consistently different between CNPB and CPPB. Additional analyses {e.g.. spectra] degrees of freedom, appro~irnats entropy) may be helpful for addressing thls issue. (Supported by HL50374 and HL40369.)

52

55

NONLINEAR DYNAMICS IN RESPIRATORY PATI'ERN GENERATOR DURING DEVELOPMENT: IMPLICATIONS FOR SUDDEN INFANT DEATH SYNDROME. Chi-Sang Pcon. Harvard-MIT D/vision of Health Sciences and Technology, MJ.T.. Cambridge. MA

POSSIBLEFRACTALAND/ORCHAOTICBREATHINGPATTERNSIN RESTINGHUMANS. ILLHughson*~tdY.Yen'Bmoto~. 'FacultyofApplledHealthScionces,UniversityofWatedon,and 2Facultyof Education,Universityof Tokyo.

Respiratory rhythm in mammals is produced by a cemral panern generator (CPG) composed of an inte~Jmnncfingneural network in the brain stem. A basic mechanism of rhythmogenesis in the adult CPG is reciprocai inhibition between inspiretory and expiratory related neuroos, with general excitation from chemoreceptor afforents and an adaptationeffect in some neurom. These neural processes undergo progressive matomficu during iof=ncy, where respiratory rhythm may also be driven partly by endogr=mus pacemaker neurons. During this developmental period, respiratory rhythm may be influencad by the nonlinear interaction between the network CPG and the pacemaker(s) if they are strongly coupled together and are of comparable potmcy. To examine the effects of such nonlinear interacdua, we employed a compound oscillator model of the neonatal respiratory CPG in the form of a neural network oscillator coupled with a pacemaker. With strceg (supracritieal)pacemaker input sad/or weak sponisneous network oscillation, the network CPG is entrained to the pacemaker. However, for suberitieal pacemaker input the"resulting respiretoW rhythm exhibits various forms of disto~ua including quasiperindicity and chaos. Moreover, the ampli~de of the respiratory rhythm decress~ with tacmasteg deviation of the pacemaker frequeucy from the spontaneous frequency, resulting in hypoventflsdon (and possibly, spnea). Thus, infants may be at risk of sudden respiratory failure if the temporal maturations of the network C I ~ and the chemorecepWr system are not properly synchronized with the decay of respiratory pacemakers dining postnatal development. Supported by NIH grants ItL50641 and HL45261; ONR contract N00014-95-0414; and NSF grant BES-9216419.

Breath-by-btenth variationin breathingpanem ofhummtsat rest is weUdocumented. Whetherthis variationis simplywhitenoise,withrandomvmation about a desiredset point,or whetherthe~ is a colored poisecharactezisticof fngtel or pomiblychaoticsyszenmis lessclear. We haveexploredthe pesm~oilitythattheremighthe shea- and long4ermcon~afionswithinthe breathingfrequency (re) and tidal volume (Vv),and theirproduct, minuteventilation(Ve) by analysisof the fi'acudpattern using coa=~egrainingspece~danalysis(CGSA). CGSAhas beandocumentedin simu]atinnstudiesto reliably extract a fractal componentfi'ommixedhannenic and non-harmonicdata (PhysicsD 68: 250-264, 1993). Bremhingparianva~bflitycomistmdyhas s highpetcentoge offfactal power (typicany ;~80% of total spectral power). The slope ([3)of the fi'ac~ c o ~ e n t estimatedflora the log pow~-Iog fi~equoncyrelationship(l/f) has rece=~y been shown to be depondenton inspiredoxygon, with reductionsin slopein hypoxin,and inereeses in hyperoxis. Thu&the panem of breathingis probably fi'actaL Fractalsomefim~but not always, impliesalso chaotic. Rece=~y,computationof the positive Lyapunev exponent was taken M evidence of chaos (Respiration Physiol. 88: 313-321, 1992). Howev~,the algorkhraused(Phys/caD 16:285-317,1985) is sensitiveto noise such as in the re~ng breathing pattern. We recently demonstrated a larger positiveLyapunevexporumtfor ira-spectral surrogate data generated flora orighud breath-by-breath data for f=, Vr, and ~ . An ahernative approach to explore the possibilityof chaos is providedby the combinationof the algorithmof Suglhara and May (Nature 344: 734-741. 1990) with the methodologyof smtogale data analy~ds. Suggestiveors chainsin the or~g~al dataw~ a high initial predictiveabilitythat dropped rapidly. However,theiso-spectralsurrogatedatatiso showedth~spat~'.enr We havenot yet foandconvlneing evidence for the presence of chaos in the pattern of resdng human respiration. A major challenge cor~onesto he the selectinnofthe gppmpdatem=thodohigyto account for the shorldurationcollections ofonisy ptiyslologinalda~

53

56

THE RESPIRATORY OSCILLATOR: PHASE RESETTING & DYSRHYTHMIC STATES David Paydarfar St. Elizabeths Medical Center and Tufts School of Medicine, Boston MA 02135

EFFICACY OF LINEAR MODELING IN RESPIRATORY CONTROL Michael C. K. Khoo Biomedical Engineering Dept., Unlv. of Southem California, Los Angeles, CA 90089.

A component of the respiratory control system is a brainstem oscillator that produces rhythmic output to the respiratory muscles. Episodes of non-obstructive apnea can interrupt otherwise normal respiratory rhythm, most often in sleeping infants, One hypothesis is that such dysrhythmias are initiated by neural perturbations with certain combinations of strength and timing relative to the respiratory r (critical stimuli|. A broad class of dynamical systems exhibit slmilar behavior, in which at*Jactor-cycla trajuctories are dispTaced towards a singular point by critical perturbations. A noisy attractor-cycle system can develop spontaneous dysrhythmla if the singular point is close enough to the attsctor cycle, and the amount of noise is within a certain critical range. In animals, phase resetting experiments identified critical neural stimuli that induced dysrhythmla of neural respiratory activity, and spontaneous dysrhythroias were more prominent st low respiratory drive. In a preterm sleeping infant with episodes of severe nonobstructive apnea, continuous tow-level sudiotactila stimulation transformed the irregular apneic rhythm to highly regular respiratory rhythm without detectable EEG or behavioral change in sleep state. In adult humans, swallowing induced strong respiratory phase resetting. The latency between laryngeal phase of swallowing and reset I-E phase of breathing was least for swallows initiated at the E-I phase, suggesting the latter is a vulnerable phase for aspiration, These findings provide insight into disturbances in control mechanisms that cause apnea and aspiration in clinical disease states, and may lead to novel therapeutic interventions, This work was supported in part by NIH grant HL49848.

Recent studies (eg Donaldson, ResDir. Physiol. 88:313-321, 1992) have suggested that resting ventilatory patterns in humans may display chaotic behavior. Th,s preliminary conclusion, however, remains to be verified due to a number of problems. First, it is unclear to what extent the ~iuirement of stetionadty is satisfied, given the conflict between long experimental times and the possibility of changes in neurobehavlorsl state. Secondly, previous studies have not examined the altemative hypothesis that the "apparent chaos" may have resulted from correlated noise propagating around the chemoreflex and mechanoref{ex loops. Given these limitations, one n~ght question the usefulness of applying current methods of nonlinear dynamics to respirator/measurements in practice, since changes in environmental factors and state of vigilance can affect ventilation considerably. We propose an alternative approach which considers the ventilation time-series to be the output of a closedJcop time-ve~jing linear system with feedback delay that is continual~ perturbed by white noise. Recuraive least-squares and Kalman identification techniques are employed far estimating the time-varyingparemeters of the linear component. We have applied this approach to "data" simulated by nn existing nonlinear model of respiratory control (Khoo et el., J.,~ZOLEbX.,~L78:10641064 1995) as well as expodmental data obtelned from humans dunng the lrenstiion from wake to sleep. In most cases, we found the linear assumption to be adequate and that the model residuals were white. We contend that this modeling approach is useful not only for characterizing nonstationary ventilatory patterns but also for tracking changes in dynamic system stability with time. Supported by NIH Grants RR-01861 & HL-02536, and an Amedcan Lung Association Career Investigator Award.

S-14

Advances in Cardiopulmonary Imaging

57

59

AIRWAY DISTENSIBILITY MEASURED IN VIVO WITH HIGH RESOLUTION CT. W. Mitaner And R. Brown. The Johns Hopkins University, Baltimore, MD 21205, USA.

2D-POWER SPECTRUMANALYSISOF SPATIALHETEROGENEITYIN PET IMAGES OF PULMONARYFUNCTION. JG Venegas and N Aljuri,MassachaseUs General Hospital, MIT and Harvard MedicalSchool. Boslon, MA. Analysisof spatial heteregeneity in pulmonary function has been carried out using the mean-normatised variance (COV2) or the residual COV2 after removingany gravitational gradieat. Such analysis, however, does not provideinformationabout the charac~'istic length scales of the spatial heterogeneity. We have modified the 2-Dimensional Fast Fourier Transform (2D.FF'r) algorithm to obtain the Power (PW) distribution within the spatial freqsency domainof imagesof pulmonary functionmeasured by positronemissiont~nography

Althongh the pressure diameter behavior of relaxed and contracted airways is an important deterndnant of lung function, very little data is available. In the present work we have applied high resolution computed tomography (HRCT) to measure conducting airway size as a function of lung volume m v/re. We studied twelve to 15 Lndividual airways (range: 2 - 15

nun ID at FRC) In each of 5 anesthetized, ventilated mongrel dogs, using HRCT with settings previously published (ARRD 144".208-212). Each airway was measured at 4 airway pressures (up to 30 cmHzO) with and

without M C h challenge,and at each pressure totallu-~ volume was also measured by HRCT as that volume inside the visceral pleural surface. With no baseline tone the curves of diameter vs pressure or flung volume)t/3 were

very nonlinear, rising steeply at low txanspu]monary pressure (Pip) to a. maximal plet~au at a Ptp between 5 and 7cmH20. With muscle tone, me curves were shifted toward higher pressures, and often showed an S-shaped appearance with a very stiff region at low pressures. Thus over small ranges of Ptp range airway tone can actually make ~ airways appear quite distans~le when the =xmoth muscle evanmaay yields. We conclude that airways clearly do not expand isotropicany with the lung paranchyma in v/re. A maximal size can be reached at very low Pip In relaxed airways, but muscle tone often prevents this maximal size from being ever reached at

physiologic pressures. (Supported by ES03819 and HL10342)

(PET). The squareof the magnitudeof the 2D-FFTfrom a PET imageyieldsan estimateof the PW spoclraldensity(PSD)distribution of the localpuimnaryvariablewithin a 2D spatial frequency domain with X and Y coordinates corresponding the spatial frequency of wavele= propagating along the vertical and horizonod directions of the image. Aft~- removingany gravitational gradients from typical pulmonary function PET images PW was found to be propagating in oildir~tions, We tharefom reduced the data to a single spatialfrequency by integrating the 2D PSD over all possible directions. Total PW par voxr summed over all spatial frequencies, was linearly related to the residualCOV2 of these images. To assess the contribution of PET-imaging-noise to the PSD, a cimractotistic PSD of noise was first obtained from PET images from a uniformityphantom. This noisc-PSD was then scaledand subleted from the PSD of each puimona13'PET image hosed on the expected con~bation of noise to the total COV2 of that image. This yielded a noise-free PSD. We conclude that the PSD analysisis a powerful method to assess the characteristic length of spatial hcterogeaeities in pulmonary pcrfasion, ventilation,and ventilation/pcrfusion ratios. (GrantHL38267).

58

60

NON-INVASIVE MEASUREMENTOF REGIONALVENTILATION WITH XENONENHANCEDCT. Brett A. Simon, M.D., Ph.D., Dept. of Anesthesiology and Critical Care Medicine, Johns Hopkins Medical Institutions, Baltimore, MD 21287

A SIMULATION MODEL FOR ASSESSMENT OF PERFUSION FROM DYNAMIC CONTRAST IMAGES Anne V. CIough, John H. Linehan arid Christopher A. Dawson Marquette University. Zablocki VAMC and Medical College of Wisconsin Milwaukee, Wisconsin 53201-1881

Improved technology with increased sensitivity detectors and faster acquisition now make computerized tomography (CT) a practical method for the determination of regional ventilation from the washout of radiodense, non-radioactive gases from the lung. Stable Xe exhibits a linear concentration-CT density enhancement (ACT) relationship, providing a ACT of 2.39 HU/% Xe at 80 kV. and 1.SS HU/% Xe at 120 kV. Since this degree of enhancement is the same magnitude as the CT density changes that occur with normal tidal breathing, it is important to carefully control lung volume during imaging. Regional ventilation is measured by repeatedly imaging the lung at end-expiration during the washin and/or washout of 7056 Xe in O2 CT density data from selected regions of interest (ROI) are fit to a single compartment model to obtain a time constant (~), which is equal to the inverse of the regional ventilation per unit volume (specific ventilation). The specific ventilations obtained are repeatable, change appropriately with changes in ventUatory parameters, and reproduce vertical gradients in ventilation which are comparable to those obtained from the literature. Data quality are influenced by the ROI size, slice thickness, presence of airways and vessels in the ROI, and proximity to the heart (cardiac motion artifact). In order to ultimately extend this methodology for use in patients, optimal protocols must be developed which provide multilevel imaging while minimizing x-ray and inhaled Xe exposure. (Supported by the American Lung Association of Maryland.)

One important application of functional imaging has been the assessment of regional tissue perf~sion using residue detection of a vascular contrast material. But, the accuracy and precision of indicator-dilution methods used to determine regional blood flow and vascular volume from dynamic contrast images arc dependent on the validity of certain assumptions made about the transport of the contrast material to and through the tissue region of interest. Thus, the robustness of the indicatordilution methods was evaluated using a computer-simulated vessel network model that has been developed. Flow through feeding arteries, networks of capillaries and draining veins was simulated with this model. Simulations were used to determine the ability of the indicator-dilution methods to reliably recover estimates of perfasion when the underiy~ng transport assumptions are violated, as they may br in experimental contexts. Practical implementation issues such as the effects of dispersion of the inlet curve, flow heterogeneity within the region of interest, velocity profiles within vessels and random noise were considered. Conclusions derived from the modeling and simulations were considered in the analysis of microfocal contrast angiography images obtained from a perfused dog lung. Estimates of regional pulmonary pcffusion under different physiologic conditions were obtained from these images. Supported by NHLBI HL-19298, NSF BES-9410669, and the Whitaker Foundation.

Respiratory Airway Mechanics II: Poster Session 61

64 A MODEL OF FLOW-INDUCED INSTABILITIES OF MUCUS.

ACUTE T I M E - C O ~ OF INDUCED BRONCHOCONSTRICTION IN RATS: SIGNIFICANCE OF CENTRAL AIRWAY SHUNTING. /.H.T.B.,ees, T,F. ~ , C. Dolman and D.H. Eidelman. Meakins-Cin~tie Laboratories and Dept. Biomedical Engineering, McOill University, Montreal, Quebec, Canada. We invmtigated thz reepirato~ mechsnlc~ changm occurring in four anesthetized, pm'elyzed, o ~ rats doting the 16 9 immediately following i.v. injectionof methacbuline. The rats were bakl qmele thrceghaet this period while ~ mplimde flow (V) osdliatinns.were applied at the tracheni openi~. Tha initial lung ~ preume w ~ 0.4 kPa "rbe V o~illatiom comisted Oftha mperlx~ition of three shmsoids having m u ~ . y prime frequeor of 1.5, 5.5 and 19.THz. T r a s h e e l p r e a u r e ( P } w M s h o ~ . ThaV andPsignels were filtered into three frequency bun~ encompasslng each o f tha i n l ~ freque~cie4. Each of die three filiered lxtirs of PandV weeethmfit tothemedel P = EV + RV + K by recunive least squares using a memmy time-constant of 0.7.S s. The reeulting E aad R signals all increased progressively over the 16 s period, reflecting the bronchoconstriction elicited. The mean fractional incmmm in E were 1.5, 4.1 nod 29.7 for ~ frequencies 1.5, 5.5 and 19.7 Ha, respectively. The c o r r m ~ g mean frar incrmsee m R were 9.3, 18.0 and 18.4. We were sbie to siamisto very shnitar results ruing a two-r lung model ceesisting of an elastic alveolsr c o m ~ connected in F,exie~ to a much stiffer pmximel eltway r merely by having tbu two elrways in th~ model incre~e their resistances without changing the compartmen~ elastanoes. This indlcetm that high ~ ~ impedance in rats during bronchoconstrlction reflects considerable central ail~vay slm~ing. (Supported by MRC Canada, FRS Quebec, RH-NCE and J.T. Costello Mere. Res. Fund,)

J.A. Meriacty & J.B. Grotherg. Depamnonts of Biomedical E ~ g and Anesthesia Northwestern University, Evanston, IL, 60208. Airways are coated with a thin serons lining Oiqnid phase) upon which rests a thin layer of mucus Q~-Iike phase). The present work examines the stability of the mucns/serous system whon to airflow over it- Tbe mucus is modoled ar a viscoelastic anlid, and the serous Isyer is modeled m a Newtonisn fluid. We assonre the alrflow lmperts both a nononl and a shear stress on the nmcns surface, and conduct a linear stability analysis to quantify such dfects as airflow velocity. serous/mucus thiakn~s taxi mucus pmpe~es. Surface tension at the monus/air interface is also incerporated into the model Results indicate a novel coepling betweco the serous and menus layers, wbichc~ndes~tbilizetbesystsm. This study provldes amodel to explore mechanisms rnsponsibie for cough clearance in airways ~ to estimate fluid forces on epithelial ceils induced during cough. Supported by a Whitak'er Foundation Special Oppom~ty Award, NSF grant CTS 9013083, and NIH gram HL41126.

62

65

EXPERIMENTS ON TRANSITION TO TURBULENCE IN OSCILLATORY PIPE FLOWS Donald Wroblewski and Yongehel Choi Boston Unive~'ity, Aerospace/Mechanieel Engineering Dept.

A M O D E L OF N O R M A L AND D Y S F U N C T I O N A L L U N G S U R F A C T A N T

Transition to turbulence in oscillatory air flow through rigid, sWaight pipes of various length has been investigated experimentally. Two component velocity measurements were obtained using a split-film, hot film seosor: The length oftbe pipes were characterized by a form of the Stronbal number, S = r l/U (U is the peak area average velocity), which also represents twice the ratio of the pipe volume to the tidal volume. This study emphasized systems with Strouhal numbers less than 2, which are characteristic of respiratory flows and for which entrance boundary layer effects can be impo~ant. Transition to turbulence was considered to have occmred when significant fluctuations were observed in the radiel component of velocity, osunily a sign of the presence of tmlmlont eddies. For S >1.5, Wansition wns governed by an unsteady Reynolds number based on the Stokes layer thiclme~; Res =/fg/v = 500 to 550, where 8 -(2v/r m is the Stokes layer thickness. This cTitet~, is consistent with u con.~derable body of experimental evidence for larger systems with smaller tidal volumes, i.e., S > 2. For S <1.5, wan~tion to turbulence is coutroUed by u quasi-steedy Reynolds number, Reo= UD/v-2300 to 2800, where U is the time average, area average velocity. This is similar to the widely accepted transition value for steady pipe flows (Re,-2000). At the low velues of Strouhal numbers characlerizlic of respiratol'5' flow in she trachea ( S - 0.2), turbulence arises late in the acceleration phase of the cycle, persisting through most of the deceleration phase. At higher Stronhel numbers, turbulence does not appear until the be~nning of the deceleration phase, and else persists until flow reversal.

63 SURFACTANT SPREADING IN THE LUNG D. HALPERN, O.E./ENSEN AND LB. GROTBERG Univ. of Alabama, Tuscaloosa At. 35487, Univ. of Newcas~e. Newcas'~e upon Tyne NE] 7RU, UK and Northwestern Univ., Evanston IL 60208 Sur factant is an essontlal component of the longs' liquid lining because it prevems airway closure from happening. Exogenous surfactant delivered to the longs of premature babies spreads under the action of gravity, surface-tension grsdlents and surface diffusion. These medumismsm, e ~ m reongaizedar vehicics fordtugtranspnst- Athenretlcalmodelofthis prosess has beco d e v ~ to quantify the effects of ~-factant dilution due to increased long surface area, end of endogenous surfantant that is already prese~t in the lungs. A system Of evolution equations for the film deformation, stafactant and drug concentrations is solved for arbitrary initial surfacumt and drug disu'ibutions. For large doses of exogenous surfa~zat, computations show considembie thinning in the first few airway gcoerations and suhstsntial fluid accnmulatlon in later gencmlinns. As the dose is decreased or a second dose is delivered, the indtt~ed flow becomes smoother but the advecden of new material is diminished. Supported by the Whitsker Foundation, NSF grant CrS 9013083, and NIH grant HL-41126,

S-15

Jo~ I , E~wafd P. Inganito 2 and Roger D. Kamm ! IMechanical Engineering and Conter fns B ~ Engineering, MIT and 2pulmonmy Division, Brigham and Women's Hospital We have previously developed an computational model that allowed us to characterize the dynamic behavior of lung surfactant in a bubble surfactometer. This model was successfully used to characterize the interfaciel behavior Of Surfactant TA, a surfactont analogue, (Otis et el., JAP 77:2681, 1994) and can also reproduce the behavior of bovine lung surfactant and DPPC, the majos phospbolipid constitoent of lung surfactant. Mcee recently, we have examined the dynamic behavior of dysfunctional lung surfactant. We examined (i) long lavage fluid h'om an animal medel of aonte hmg injery pr~xlaced by eedotoxin inhalation and infusion and (ii) lung surfactant oxidized using FeCI 2 + H202. The dynamic surface tension profiles (20 cycles/rain; I mg/ml) showed distinct peltems of dysfunction. Use of the comt~mtioonl model suggests that the profile for oxidized surfactant w ~ ~ n t with an increased solubility of the surfaetant in ~ I~ b l 1 1 4 b I 1 1 the bu~ phase. However, we onuld nor ~ (ramz) the unusual ensl~like behavior seen in the acute lung injury model until we extendsd our model to include the effects of surfactant diffusion in the bulk phase. We then found that the cusp-like behavior was related to the formation of a subsurface depot caused by the squeeze-out of surfactant molecules and their vev/low micolinr diffusion coefficient. Further use of the model might allow a better ondcxstanding of the biophysics of serfactant dysftmctlon. (support provided by NIH grant I-U.,33009)

S-16

Pulmonary Circulation Dynamics and Modeling I

66

68

THE MORPHOME'I~Y OF THE PULMONARYARTERIALTREE. C. A. Dawson, J. H. Linehan, G. S. Krenz. Medical Collegeof Wiscaasin, Marquette University,, ZablockiVetcmoAffairsMedical Center, 53295.

THE SITE OF FLOW LIMITATIONIN THE PULMONARYVASCULATURE S.T. Haworth, D.A. Rickaby, C.A. Dawson, and J.H. Linehan. Marquette University,MedicalCollege of WI. and Zablocki VAMC, Milwaukee, WI.

Commonly, pulmonary arterial mo~hometric data ha~x:been s ~ by groupingthe individual vessel segmente according to generation or ozder and then averaging the dimemious within each generationor order. The most effective criteria for grouping has been a question, and some cd(eria me not applicableto imaging methods having limited tusolufion.We have considered an altcmath'c approach io which we begin with the concept that the bifurcating,volumeIdling chassclcristicsof the tree put constraintson the such tlm the assignment of orders or generations may be superfluous. The indepe~ent, appcarame of the Uoc suggests that one might consider the thxee vessel tcgrmms joined at a bifueation to he the fundamental repenting morpimmetric unit descriptiveof the tser The diameters oftha three vesselsat each bifurcation, D I the vessel diameter, and D2 and D3, the two daughter vessel diameters are used to calculate which is the harmonic mean of a = log2/(lng2D l - log(D2+D3). when: a is the desunptor ofcachbffurcation of the tree. Within the range of resolutionof the imaging modality, a statistical sample of a's can provide an estimate of [3. To put the utility of [~ io perspective, we intmduca1he concept of cumulativevascularvolume, which is Ihe arterial volume upstream from all of the localioas wilhio the arterial Uec that have the mine iotravascoisr pmsaure. The distribution of intravascolurpressure from arredal inlet to capillary inlet us a functionof cumulativevascularvolumecan bc expressed in terms of 1~. Thus. a sample containinga sufficientlylarge number of bifurcationscan be used to relate the murphomettic data to pulmona,'y arterialtree hemodyuamic function. Supported by Department of Veteran Affairsand HL 19298

The condition in the pulmonary vascular bed wherein the artery pressure, 1%, exceeds the ainvay pressure, PA, which exceeds the venouspressure, Pv. (i.e.P. > PA > Pv) is defraud us zone 2. Underthese canditious, the airway pressure must be greater than the intravuscular pressure somewhere within the pulmonary vascular bed. With compliant vessels, uegadve vessel.airway pressure differences can narrow ve.,.,.,.,.,.,.,.,.,~cls, substantially increasing their hemodynamic resistance. Pressures measured in subplcusni veins have been interpreted as indicating that veina greater than 50 pm in diameter are the site of hemodynamic resistance in zone 2 ($hepard et ai. L Appl. Physiol. 64: 874, 1988). This is in contrast to morphometric evidence which has suggested that the capillariesare the site of the nurrowing, In an attempt to better understand this apparent discrepancy, we used the sen,o-null micropuncluse method to mcasore pressures in subpleural vessels ranging from 20 to 60 pm in isofa[ed dog lung lobes. Using this method, we obtained mcasuremenLs for the lobar artery, P., subplenral arteriole. Pma. subplenral venule.Pray, and lobar vein, Pv, for a range of PA pressores (5.2 to 13.1 Ton') with Pv varied between 0 and 24.2 Ton (Pa > Pv 9 PA > Ppl). The ratios ~sPd~v, z~md~sPv, and ~h~176 obtained from the slopes of the linear regressionof the respective pressures on 1%for either P,, 9 P^ or Pv < P^, were compared. For Pv > p^. the mean value of these ratios were 0.81, 0.75. and 0.94, respectively, while for 1% < PA, the values were 0.25% 0.21". and 0.86. respectively(* P < 0.Ol). Thisresult indicates the site of increased vascular resistance was upstream of the micropunctured venules and downslseam from Ihr micropunctured arterioles, These resulls provide hcmodynamic support to compliment the morphometric evidence for a capillary location as the locus of iocrcascd vascular resistance under zone 2 conditions. Supportedby HL-19298 and Dcpa~mcnt of Veteran Affairs.

67

69

PARTITION OF PULMONARY LONGITUDINAL RESISTANCE BY ARTERIAL AND VENOUS OCCLUSIONS IN INTACT DOG. C. M~lot, F. Vermeulen, J. Stephanazzi, M. Dolcroix, R. Nasijo. Laboratory of Cardiorospiratory Physiology, Free University of Brussels, Brussels, Belgium.

NITRIC OXIDE IS NECESSARY FOR THE DELAYED VASODILATION AFTER CONCENTRATED HEMOGLOBIN SOLUTIONS IN THE LUNG CIRCULATION OF DOGS. Rong Z. Gun and Norman C. Staub St. The LoveL~:cInstitutes, Albuquerque, New Mexico 87108.

The vascular pressure transients after occlusion of a small artery or vein with Swon-Ganz catheters were analyzed for the inflection point between rapid and slow components in 6 intact dogs before and after polysWrene beads embolization. Effects of histamine and serotonin (5-HT) were also measured. Digitized curves and double exponential fitting were used to compute arterial (Pao) and venous {Pro) occlusion pressures. Pro calculations were performed on mean and diastolic pressures. Pro calculated from the mean pressure was consistently higher than Pug. We therefore used Pro calculated from diastolic pressure to compute pressure drop in arterial, middle, end venous segments of the pulmonary circulation. At base line histamine increased the pressure drop in the middle and the venous segments, und 5-HT in the arterial segment. Embolizatlon increased the pressure drop in the arterial end middle segments. After ombolizution effects of 5-HT and histamine wore blunted. In conclusion: In intact lung, m t~n~w, al~t~ouE a pertltloning pulmonary vascular S/.SEUmlI, ~ e resistance by arterial and venous s+n occlusions gives similar results than in ~ar*MINS isolated lung in steady flow conditions. I~eou~ s.wr I .+.+.l*~:~a In pulsatile conditions Pro seems to be better evaluated from the diastolic HISTAMINE ~o ~0 1o o 1o pressure. ~EUSURSDROP.mrr~

In the in situ. blood-porfused dog lung, we found that addition ofS-10 g/100 ml dog stroma-frce hemoglobin to the perfusion reservoir caused an immediate (10 rain) rise in vascular resistance followed by a delayed (1 h) repo.vory. To investigate the mechanism of vasoconstriction we blocked nitric oxide (NO) production by adding N~ nitro-L-at#nine methyl ester (L-NAME) during the baseline. The resistance data (mmHg/0itor/mia)) at constant flow (500 ml/min) are summarized in the table. Experiment

Number

Control Experiment

2 5

Baseline

L-NAME

19.5• -19.6:fl9.7 23.6•

Homoglobin-Solutlve early late 32.5• 1.3 13.9-J:I.9 33.I+11.1 32. l:l:lI

After L-NAMe, only one lung showed a significant rise in resistance. A.qcr the hemoglobin solution, the rcsistunen increased in all lungs; mose in those given LNAME. To our suq~risc there was no ddaycd vasodilation. We conclude that scavcaging of NO by hemoglobin is not tlm mechanism of the early constriction but that induocd NO production is ncenssasy for the late dilation.

Pulmonary Circulation Dynamics and Modeling II

S-17

70

72

RECRUITMENT OF VASCULAR GAS EXCHANGE SURFACE AREA IN THE PULMONARY CAPILLARIES Wiltz W. Wagner, Jr. and Robert G. Presson, Jr. Departments of Anesthesiology, Physiology, and Pediatrics, Indiana University School of Medicine, Indianapolis, IN 46202

MICROVASCULAR VOLUME CHANGE AND FILTRATION ARE PRODUCED BY CHANGES IN MICROVASCULAR PRESSURE IN THE CANINE LUNG J. S. Lee and L. P. Lee Dept. Biomed. Eng., Univ. o f Va', Charlottesville, Va 22908

Pulmonary gas exchange reserve can take the form of distension of perfused capilianes, recruitment of unperfused capillaries, or changes in eapillnsy transit time. Using video microscopy of the canine subplanral mierocirculation, we have measured capillary recruitment and transit time. Racnfitment was measured by summing the lengths of perfused capillaries in alveolar walls through a transparent window. Capillary transit times were measured by recording the passage of either a plasma fluorescenl dye bolus or a bolus of fluorescently-labeled red blood cells as they passed through a single pulmonary capillary bed. In pentobarbital anesthetized dogs, measurements were made during baseline cardiac output (mean = 4 L/rain), and after fluid loading and isoproterenol infusion to increase cardiac output (mean = 12 L/min). Red blood cell transit times were faster than plasma transit times by a factor of 1.4. As regional blood flow increased, mean capillary transit times decreased. The changes in capillary volume calcoiamd from changes in regional blood flow and transit time were entirely accounted for by capillary recroitment. The distribution of transit times narrowed with increasing blood flow (relative dispersion decreased from 0.50 to 0.41) making it less likely that the faster transit times would fall below the minimum for complete oxygenation with increasing flow (minimum transit time was 0.4 s at a cardiac output of 12 L/rain). We calculate that minimum transit time would reach 0.25 s at a cardiac output of 20 L/rain in the dog. Supported by NIH HL-36033.

The hematoerit in the capillaries is lower than that in the large veins and 9arteries - the Fahraeus effect. As a result, the observed transient increase in henmtocrit in venous o o ~ o w during the elevation o f perfusinn pressure in isolated canine left lower lobe can be interpreted as that the expansion in pulmonery capillaries sequesters more plasma than RBC and thus leads to a transient increase in hematocrit in venous outflow. A theot3' based on the Fahraeua effect in a d9 and permeable micmvesculaturo and flow dispersion throngh the vaacolaturo is developed. The theory first determines how the capillary volume change and filtration produce the hemaIocrit variation in blood leaving the micrcoirculation. Its convolution with the transfer functinn on flow dispeninn forms the hematocrit variation in venous blood which matches well with the experimental resnlts over the 1 minute course o f an 8 mmHg elevation in both the pulmonary arterial and venus pressure. At 1 minute o f pressure elevation, the lobe as measured gains a weight o f 143 g/kg tissue. The analysis indicates that filtration, micro- and macrovescular expansion accounts for 4, 41 and 55% o f the total weighI gain respectively. These estimates indicate that the pulmonary miorociroulation can redistribute its blood volume to other part o f the oirculation for the maintenance o f cardiac filling. One conseqnence o f this resnlt on transient hematoerit increase suggests that changes in hematocrit can not be used solely to estimate ~msuapillary filtration. Supported in part by NHLBI 40893.

7l

73

A NEW MODEL FOR PULMONARY CAPILLARY BLOOD FLOW. A.S. Dhadwal, R.D. Kamm, B. Wiggst Department of Mechanical Englneexmg& Center for Biomedical Engineering, MIT, Cambridge MA. tPulmonary Research Laboratory, University of British Columbia, Canada.

INTERACTION BETWEEN PLEURAL LIQUID TURNOVER AND THE SYSTEMIC AND PULMONARY MICROVASCULAR EXCHANGES Daniela Negfini Istlteto di Fisiolngia Umaus, Uuiversi~ degli Studi. Via Mangiagalli 32, 1-20133 Miiano, Italy.

Blood flow in the alveolar septum depends on factors such as septal geometry, lung volume, transmoral end transpulmonary pressures, and blood theology. The capillary-matrix model of the septum was developed for a network of interconnected tubes forming a square army (side 75 gin) of 60 segments, and 5x5 cylindrical, distensible 'posts', incorporating the effects of breathing, blood theology, and spatial variability in septal compliance and septal height at zero transmaral pressure (a and he). The lateral septel (and peso dimensions were assumed to change isotropieally with lung volume; post-height was varied assuming a constant 'post-volume'. Equations were developed using a connol-volume method for qunsi-stendy inania-fsee flow. Spatial variability in Iio and a wns imposed by a random selection from a normal distribution, and beueda~ pressures were imposed as boundary conditions. The model was used to explore the effeets of breathing, spatial variability in he and ct, capillary blockage, and the effects of palsatility in boandaty pressures. It was shown that breathing had a significant impact on capillary segment msistunce, with cecillations xs high ~ 45% of the mean value. Neither capilinzy blockage nor spatial variability had much effect on the breathing-induced flow eacilintiom; only the raean values were signiticentiy altered. Pulsatility in boundary pressures caused the junction pressures to oseilintu at the palsatile frequency, bet the flow-rate oscillations accormd at the breathing frequency. Static computetioas demonstrated that spatial variability in Iloand (z resulted in pt~efematiallyIs,fused regions, with variability in he dominating the effect of variability in a; a case with 30% variability in he coupled with 25% variability in a rmulted in a flow-mtudistribution with a ledgecoefficient of variation (.-0.25). The effect of capillary blockage on flow-rates was local,falling to less than 20% for segments more than one segment away from the blockage. Thus breathing, septum morphometry, and capillary blood theology appear to play a key role in determining the transit of blood and its constituents through the pelmona~ microcircolation. Supported by a grant from the NHLBI (HL33009) and the Freeman Foundation.

The pleural space is a fluid compartment delimited by the parietal and visceral mesothelia and placed in series between the parietal interstitium (supplied by systemic circulation) and the pulmonary one (supplied by visceral or systemic circulation). Fluid and solutes filtrate through the parietal pleura, whose low bydmuiic conductivity and low solute protein permeability coefficients guarantee a small filtration [~0.2 ml/(Kg.h)] of hypo-ancotic fluid (~,1 gr/dl). The main route for pleural fluid and solutes removal is the parietal planml lymphatic system that may increase its draining flow by more than one order of magnitude in response to increased filtration rate in order to maintain plearal liquid volume close to its minimum "set point'. Above saturation pleural lymphatics loose control on pleural fluid volume. In adult mammals the pleural liquid turnover is f~nctionally independent upon pulmonary microvaseular exchanges, due to the ve~ low permeability of the visceral mesothelium In site measurements of interstitial and puimana~ hydraulic pressure reveal the existence of a pressure gradient driving fluid from the pulmonary capillaries to surrounding interstitlum, Pulmonaw microvascular exchanges seem to play a role in pleural fluid turnover only during early development of pulmonary edema and in early postnatal life. when pulmonary interstitial pressure becomes positive and fluid filtration into the plenral space may occur.

S - 18

Poster Presentations

74

77

COMPARISON OF IMPEDANCE COMPENSATED VENTILATION (ICV)WITH CONVENTIONAL METHODS; ICV PRESERVES NATURALLOADINGFOR THE RESPIRATORYDRIVE M.A. Berrello Bird Products Corporation, Palm Springs,CA

A METHOD FOR DETECTIONOF HYPEROXIARESPONSIVE PROTEINS IN THE PULMONARYVASCULATURE E,K. I ~ La Fl~Itc, C.A. Dawzon, J.I~ Linchan, and M.P. Mcrker IVhrqueae University,M~lical Collegeof Wiscon~n, and V. A. Medical Center, Milwaukee, W153295

A mathematical comparison is made between conventionalmethods of mechanical ventilation,which presently focus on volumedeliveryand pressure regulation,end Impedance Compensated Ventilation(ICV), which focuses on establishingan effective "nonuel" impedance. A simplelung model is used to demonstrate that conventional methods, although possiblydecreasing imposedwork of breathing,create an unnatural mechanical loud as seen by the patient's respiratory drive. ICV is shownnot only to reduce imposedwork, but to do so in a manner which preserves the structure of natural loading.

To stodythe spectram of proteins exposed to the bloodin intact lungs, wo developeda toehniqne far covalentlabellingof Inmirmllyaccessibleproinias in the isolatedpeffnsed lungs of control rats and rals expesed to hypemxia (>95% O=for 60 hoars). Proteias were labelled by tingle pass bolus injectionof the membrane impermustubiotinylatienreagent, salthmccinimidyl-~(bio6usmido)hr (NHS-LC-biotin),into the polmonmyarterial canusla of the perfusedlung. A membrane fractien was prspared frem the homogenizedluu8 and glyeeprotoinswere partiallype~ied esing lectinaffinitypuriflcatio~ ~ were appliedto 4-12% gradient SDS p o ~ d c gels and biotinylatodproteins were observed on filmsof nitrocelluloseblots using streptavidis-horseradishperoxidase and a chcmiinmiac,cent mbsl~te for the peroxidase. In a tingle passage tluoegh the lung, NH~LC-bietin la~lled a verietyof pmteiss. Tbu amount of biotin assaciated with each w~ qlmnfified by atmnnin~deasltomeu3'and the data flora the ~ms wero analyzed using a curve fitting modelto estimate the proportion of totel bio~inylatedprotein represented by iedividu~ proteim. The prO'des of lecttn a~nity lauificd biotinylatedprotoinsfrom conUollungs were differentfrom those obtained fi'omlungs of rats exposed to hypernsi~ We r that biofinyintionby bolns injectioninto the pulmonasyvusculatorecan be used to identify specificproteinsexposedto the blood that are responsiveto hyperoxin. Supportedby

the CancerCenterof the Mudlcal Collegeof Wiseousis HL24349,I-1L52108,and the Depamncnt of Veteran's Affairs.

75

78

SIMULTANEOUSCONTROLOF FLOWAND PRESSUREFOR MECHANICAL VENTILATIONOF THE LUNG M.A. Borrello Bird Products Corporation, Palm Springs,CA

CALCULATIONOF THE DOUBLELAYER POTENTIAL WITHIN A FIBER MATRIX D.J. Haselton. S. Kimt and J. B. Bassie~thwziehte Ceat~ for BioengineeringWD-12, Universityof Washington.Sentdc, WA 98195 and IDcpartmcnt of CbomlcalEngineering.Universityof Wisconsin,Madison, WI 53706

Feedback methods used to simultaneouslycontrol pressure and flow in a ventilatoroften present stabilityproblems directly linkedto system parameters of resistance, clastance, and loop gain. Both simulationand actual systems have shown the existence of limit cycles caused by the nus-linenrrelationshipbetween pressure and flow. Minimizingthe magnitude of these limit cycles can be accomplished by selectingappropriate controls to increase dampingyet maintainrequiredbandwidth for response.

In biological systems, conveedve transport occurs in electrolytic solutions flowingpast charged molecules. Nevertheless, most calculations of velocity profiles do not include electroHnetic conuibudoas. ~ fleet step in analyzingeffects of eleetrokiocllesis determination of the double layer potential. We calculated this potendal (solution of the linear PoissonBoltzmaun equation) using both a unit cell approximationand a linear soperpositionsolutionin the region within a regular mray of infinite fibers. This geometry repre~ots the histologyfor systems as divc~e as giyencalyxwithin the interendothefialcleft of the transespillaryexchange battier, extrocelinlar matrix in cartilage and the ostenglycan manix in canaliculi.For a Dchyc length typical of extracellidarfluid (8 ~) and a zeta potential typical of exwaccllularmolecules (-60 mV), the potential decays with distance from a fiber surface, but remains nonzero for porosity valuesless than 97% in both approximations.Thus, at everypoint fln'oughoutthe region of this idealized model, there will be enhanced anionic and diminished cationic convective transport compared to enutral mol~ules. The effect wiJl bc more pronounced for lerger moleculeswhose doubleisyersextoad into ~gioas close to the fiber surface where the magnitude of the doable layer potential is greater. This msolt agrees with experimental observationsof transport of charged molecules across the transcapillary exchange barrier. In addition, these solotiousare ganeraland may be applied to variouscharge eansport problems. (Supportediopart by NIH grant HLA-1243 (JBB), the NIH Medical Scientist Training Program (DJH) and the Poocin Scholarship(DJH) at the Universityof Washington)

76

79

A COMPUTER SIMULATION OF THE EFFECTS OF WEIGHTLESSNESS ON THE HUMAN PULMONARY SYSTEM Sharmista Chatterjee and Prof.R.C.Se~q'ave Department of Chemical Engineering, Iowa State University~mes, Iows, 50011.

ROLE OF EXTRATHORACIC HEAT AND WATER VAPOR EXCHANGE IN DETERMINING AIRWAY PARTICLE DEPOSITION. Jonathan W. Kaufman 1, C.-C. Geoffrey Yang2, Peter W. Scherar2 l-Naval Air Warfare Center, Warminster, PA 18974-0591 2-Dept. of Bioengineenng, University of Pennsylvania, Philadelphia, PA 19104

Gravity plays a major role in influencing lung function. Under conditions of short term weightlessness the changes taklng place are homogeneous distribution of ventilatien.perrusion, decrease in the vital capacity of the lung, increase in the abdominal contribution to lung mechanics a~i increase in the diffusing capacity of the blood gun bsrrier. Our objective in this work has been to develop a mathematieal model of the human respiratory system, which can test the various hypotheses which have been put forth to eaplain pulmonm'y behavior i n microgravity. Our model consists o f a distribnted-lumped parameter model o f t b e lung gas exchange apparatus which can take ~ account the impeded mass transfer i n the pulmonary capillaries. InequaLities in ventilation-perrusion are accounted for by dividing the lung into 3 zones apart from a dead space aud a abu~ compartment. The gas exchange apparatus works in a contirawus fashioD. The functioning of the gas exchange apparatus is controlled by peripheral and cerebral chemoreceptors which work i n a discrete fashion. The frequency of breatbtn E is set by a frequency optimizer which takes into account the principles of respiratory mechanics. Non-linear time variation of the lung resistances and compliances is considered. The model is valide~ed by comparing its responses under conditions of hypo~da and hypercapnia in normal gravity with available data. Under conditions of micro-gravity the model predicts a decrement in VO2 which is coneistent with data available in literature.

Hygroscopic 0.2-3.0 mm mean mass aerodynamic diameter particles deposit along the airway as a function of airway humidity profiles. Little attention, however, has been given to the role extrathorecic airway conditions play in particle deposition. Airway particle deposition was predicted by coupling airway heat and water vapor exchange and particle deposition models. Using either experimentally measured oral cavity or pipe heat (h) end mass transfer (ko) coefficients, airway humidity and temperature profiles were predicted for oral breathing in ambient temperatures (T... = -5 ~ 10 ~ 22 ~ 45~ at 40% relative humidity (RH). Humidity profile differences between breathing through the oral cavity or a straight pipe were inversely proportional to T,,.~, with a neady IO0%RH difference at the distal oral pharynx with T~b = -5~ air. Reducing extrathoracir RH causes particle deposition to substantially decline in the conducting airways end increase in the pulmonary airways. These results suggest that oral cavity water vapor exchange has an important role in determining hygroscopic particle deposition in the human airway, particularly in extreme environments.

Poster Presentations 80

83

SPECTRAL CHARACTERIZATION OF CARDIOVASCULAR PARAMETERS FOLLOWING ENDOTRACHEAL INSTILLATION OF BOVINE PULMONARY SURFACTANT IN PREMATURE INFANTS. R, Kundich, M. Daley, and H. Bada; Depts. of Biomedical Engineering, Pediatrics, and OB & GYN, University of Tennessee, Memphis and Dept. of Electrical Engineering, University of Memphis, Memphis TN.

MODULATION OP BLOOD PRESSURE AND M A C R O M O L E C U L A R UPTAKE IN SOLID TUMORS Netti P. A., Baster L. T. Boucher Y., Skalak* R. and Jain R. K. DpL of Radiation Oncology. Mass. General Hospital, Harvard Medical School, Boston. MA 02114, ~ Institute, University of California, San Diego, CA 92093

S- 19

Spectral characterization of heart rate (HR} variability has been previously employed in the assessment of premature infants. Given an interaction exits among respiration, heart rate, and arterial blond pressure (BP) tn neonates, it was the objective of this study to design a methodology to begin to assess the offects of surfactant instillation on the cardiovascular system. Four premature Infants (birth weight, 887-1930 grams) with a diagnosis of respiratory distress ayndremo required surfactant instillation and were the aubjects of this study. Tho ECG end arterial BP signals were acquired et 512 Hz and 128 Hz, respectively using a laptop computer Interfaced with the patiant'a bedside monitor. Data acquisition was done 30 rain.before and after surfactent instillation. Two patients required a single aurfactent inatillation, whereas the other 2 infants had 2 doses. In 4 of B cases of instillation procedures, tho spectral power of HR and BP increased in the low frequoncy (LF) region (0.02 - O.2Hz), with concomitant improvement in arterial oxygenation. In the remaining 2 instances, arterial BP spectral power decreased in the LF region; in neither of these 2 cases did we observe an increase in PaO2. Two factors may be postulated for inducing these HR and blood pressure responses around the LF region, neurological and mechanical, leading to changes in lung compliance and pulmonaw blond flow.

High interstitial fluid pressure(IFP) in solid tumors leads to physiological transport barriers and prevents the uniform dls~hafies of large macromolecules in neoplastic tisssss. A pouible strategy to overcome these physiological barriers b to alter or modulate the vascular pre~um. In the prosunt study the dynamic rcspensu of interstitial fluid and extracolinlar mmzix to penorbations of vascular wessore and tumor blood flow has been mathematicaily simulated. Tbe extrasellular mtarin and the celluinr phase of the tumor have beea desufibed as en eJa=lo macromolecolar network saturstod by fluid. The vascular space has been described as an ensemble of vessois randomly distributed in this antwodr. Tbe medol is shle to lmuilot the IFP and fluid velocity profiles, as well as the tissue hydration end slross distributions, both under steady-state and Wsssient coedltio*'a. Following a change in vascular p~.u,zra or Uueor blood flow ('rBF), thero will be a transient vedistrthutlon of interstitial fluid plsssuro and velocity and solid stress. Model simulations show that following a change of vascular pressure the primm7 mechanism of this redistribution is transcapillary fluid exchange. The predicted time scale for this transient redistribution is tens of seconds. Following a sudden reduction in TEF. however, the predominant mechanism is the percolation of fluid trough the extracefluinr malfix. Under these conditions, the predicted time scale of the transient response is much Iongr162of the ordm" of half ss hour. The modet predictious are in aBreemcot with daia obtained from naperimcata ou isolated tumors. The IFP in the center of an isolated tumor follows any variation of the ~ l a l pressure with a time lag of tens of seconds (I l:t6 see), whereas stoppin S the TEF the decay of ITP occurs in a time scale of half hours (1516i 940 see). The simulations suggest that chronic vascular h ~ o n should not lead to a signifr..ant iecw.aso of n ~ l r uptakc in the central area of solid tumors. Instead a more favourable effect would be expected by periodic modulation of the vascular p ~ .

81

84

EVALUATION OF A COMPOBITE HOLLOW FIBER MEMBRANE FOR AN INTRAVENOUB MEMBRANE OXYGENATOR L. Lund, W.J. Fedorsplel, M. Wallace, F. Walter=;, H. Borovetz, B. G. HaUler Bioengineering Program & Department of Surgery, University of Pittsburgh

VOLUNTARY CONTROL OF BREATHING: EFFECTS ON HEART RATE VARIABILITY A.R. Patwardhan. S. Vallorupalli, J.M. Evans, D.L Eckberg, E.N. Brace, C.F. Knapp. Center for Biomedical Engineering, University of KentuckT, Lexington, KY 40506-0070

Designing an effective intravenous membrane oxygenator requires selecting hollow fiber membranes (HFMs) which present minimal resistance to gas exchange over extended periods of time. Porous fiber membranes, as used in extraccrporeal oxygenators, offer e minimal exchange resistance, but one that diminishes with time due to fiber wetting and subsequent serum leakage. Potentially attractive alternatives are ccmpealte HFMs, which inhibit fiber wetting and serum leakage by Incorporating a true membrane layer within their porous walls. To evaluate composite and other HFMs, we developed a simple apparatus and methodology for measuring HFM permeability In 9 gas-liquid system under conditions relevant to Intravenous oxygenation. "the testing chamber houses a bank of fibers immersed In a small volume of stirred liquid. Nitrogen flow through the fibers washes out O= and CO= preequilibrated within the liquid volume, and the rate of the exponential washout is used to calculate the exchange permeability of the fiber-liquid system. Actual membrane permeability (k,,) is estimated by eppropdately extrapolating the measured permeabilities to infinite atirdng rates and negligible liquid baundary layer resistance. Using this methodology, we measured the O= and CO= exchange permeabllitles of Mitsublahl (MHF 2001.) composite HFMs. We found that the value of k,. obtained for O= was 6.8 x 10"* ml/s/cm=/cm Hg, which is less than the overall exchange permeability ultimately required of our intravenous oxygenation device (k 9 1 x 10.= ml/s/cmZ/cmHg). Accordingly, the MHF 200L composite fiber may not be suitable for intravenous oxygenation devices like ours, and therefore other composite and aitemative HFM technologies need to be explored.

Analysis of heart rate variability provides insight into the autonomic function in cardiovascular regulation. We investigated the r162 of voluntary control of brnaOdag on beast rate variabih'ty, because during vohmtery control of breatldng, the breathing pattesu and the mechanlams that generate the bPetthin~ pattern arc different than spontaneous breathing. We computed two ~ e~ heart rate variability, pawer specmm~ and fi'actal sinpcs d u ~ two stsdiss designed to address the following issuss: I) To inves~gatc tbu effects of altered breathing pattsm, we compared heart rats variability fzom 8 volontscrs during three brsuthta8 trial= of 10 minutes each: i) sponmncotm breathing (meau rate of 14.4 breathgmin), ii) breathing tu a metronome et the rate oflS, 18 a n d 2 l bmetht/min for 2, 6 and 2 minutes, and lli) breathing to a metronome at the rate of 18 brcethgmin for 10 m/nut=. 2) To investigate the e f f e ~ of d/ffcre=;es in the way the breathing pattc~a is generated, we compased heart ran= vaxinbility from lO volunteers between five minat=s of spontanr and controlled breathing. During o0atrolled breathing, for five mln.t~c subjects voluntarily reproduced their own spontaneous breathing pattern (both rate and volume on a breath by breath bad=). Eesulta from the t i n t study showed that heart rate pow~ in the re~iratoly frequency region was smaller during metronomic brssthin 8 than du.,ing spontaneous breathiag. Results from the second study showed that during volnatm7 control of breathing, fractel and spectral pawcrs in heart rate in the r-..spiratery frequency region decreased. Our results indicate that contrary to what has been suggested in the llteraturr the act of veluetsry control of breathing decreases the parasympathetic influence in cardiovascular vegulstioe.

82

85

A NONINVASIVE ESTIMATION METHOD FOR THE MEASUREMENT OF RESPIRATORY RESISTANCE AND INERTANCE Narciso F. Macla! , William J. Dorson 2 , and Walter T. Higgins Jr3 I Departmcot of Electroaics and Computer Technology, 2 Department of Chemical, Bio, and Materials Engineering, 3Department of Electrical Engineering Arizona State University, Tempe, AZ 85287

AUTOMATED DETECTION OF THE INTRATHORACIC AIRWAY TREE FROM VOLUMETRIC X-RAY CT IMAGES Milan SonkaI, Wonkyu Park~, Eric A. Hoffman2 Dept. ~Elest.& Comp. Eng., and 2Radiology, The University of Iowa, Iowa City, IA 52242

A noainvasive technique for estimating lung model parameters was developed, implemcoted, and tested with in vitro and in vivo expefimouts. Called the Quick Obstruction Method (QOM), it estimates respiratory resistance and iuertaece. The test procedure required obstructing the airflow while simultaneously recording flow and mouth pressure. The portion of the recorded signal d u n g the mommtm 7 obstruction (differsuce betwesu the response and the prr excitation extrapolation) was extracted and fitted to the lung model using least gluares curve fitting. Data from the 45 ms duration obstruction was processed using 9 compliance value obtained using the weighted spirometer method to cohsuce the accuracy of the technique. The iuocedure produced respiratory resistance and inertance. The test was performed on fifteen male, nonamoklng subjects. Values of measured respirator T resistance were in agrecmeat with a derived expression for predicted respiratory msistmxce obtained by modifying publisbed airway resistaece ioformation. ~ t a b i l i t y of re~istence and iecrtenco values was 4.5% and 9.59G oa a single subject, respectively. This method promises a practical and clinically usable test for measuring resistance and inertance.

Global pulmonary function tests am often too insensitive to assess heterogeneous or peripheral abnormalities in structure and function. Accurate quantitation of the a ~ requires volumetric CT imaging. 3-D airway detection can provide landmarks for tracking lung perenchyma and airways in the regional evaluation of pulmonary physiology. We report an automated knowledge-based method for segmentation of intrathoracic airway trees from 3-D CT image data sets. The methodis bassd on a combination of 1) 3-D seeded region growing that is used to identify large airways, 2) rule-besed 2-D segmentation of individual CT dices to identify probable locations of smaller diameter airways, and 3) merging abNsy regions across the three-dimensional set of slices resulting in a trae-like akway structure. The approach utilizes a pnon" anatomical knowledge about pulmonary airway and vascular trees end their inter-ralationshipe. Our new rule-based method end a conventional 3-D region growing method for akway tree detection were.applied to five volumetric CT data sets of canine lungs. Each data set consisted of 40 slices and was acquired/n vivo via electron beam CT using 3 mm slice thickness with lung volume held at a constant pressure. In 8 dices per set (40 slices total), we identified correctly detected, undetected, and falsely detected a ~ by comparison to the independent standard defined by an experienced observer. I Method Airways: Correctly detected Undetected Falsely detected] Conventional 1211255 (total) 3.6 + 2.9 (per slice) 0.2 + 0.4 (per slice)l Rule.Based 194/258 (total) 1.6 • 1.6 (per slice) 4.9 • 4.7 (per slice)l Rule-based airway tree detection substantially decreases the numbers of undetected airways in CT data sets. Current effort is directed to eliminate faisely identified airways in a pest-processing step utilizing a kncwledge-based tree growing strategy.

S-20

Poster Presentations

86 FLUID AND SOLUTES EXCHANGES ACROSS THE PLEURAL SPACE DESCRIBED BY A THREE=COMPARTMENT MODEL. D.Vuaturoli. D.Negrini, M.Del Fabbro, and G. Miserocchi Istilum di Fisiologia Umana, Universi~* degli Stndi, Via Mangiagalli 32, 1-20133 Milano, Italy. We developed a mathematical model for a three compartment biological system based on the mass conservationprinciple.The compartments are arranged as sketchedin figure:they can

be identified with the systemic micrccircniafion (I), the s~bpleural interstitium (2) and the planral space (3). The membrane separating compartment 1 and 2 is the capillar/ endothelium, that between compartment 2 and 3 i$ the plcoral mcsothelium. The drainage flows are generated by interstitial (2--+I) and pleural (3 ] ~1) lymphatics that can increase the flow rates . proportional to the i ~ in liquid pressure. Jml and Jm2 represent flows through sieving membranes whereas Jdl and Jd2 are nonsieving drainage flows. Using the balance equations for fluid and solutes one can perform a sensitivity analysis obtaining the influence of the parameters appearing in the equations (namely, the permeability properties of membranes and the operative features of lymphatics) on the state variables (hych'aulic pressures and protein concentrations in compartments 2 and 3). The analysis shows that compartment 2 (the subpleural interstitium) acts as a buffer to prevent an increase in pleursl liquid volume and protein concenWation. Normal pleural lymphatics f~nctina assures a minimum pleural liquid volume: pleural effusion may result by reducing lymphatic drainage efficiency.

2 1 2 [ z [

Current

and Future

Trends

in Mechanical

Circulatory

Support

I

87

90

RELIABILITY ASSESSMENTOf= MECHANICAL CI~CUJ..ATORYSUPPORT SYSTEMS Tofy Muasivsud CardiovaSen'lat Deyices Divhi0n, :Department of Sui'sery, Faculty of Medicine, Uni;aei~ity Of Ottawa Heart Institute,"OttaWa,Ontario, Canada.

DEVICES T O ASSIST AND REPLACE T H E H E A R T R T V Kung, ABIOMED, Inc.

Due to the lll'e supporting nature of mechanical~i~Ulstot7 sYSter~, regulato~ ~genci~, users and patients, expect Imd demand that the safety and relinblli6/of these devices be dem0ustratod. The.s, these devices must be evaluated in a seties of in vitro, in vivo and clinical experiments. Reliability issues should be addressed from the beginning of the coueapt and throughout the device design, manufacturing and utilization cycle. Through the use of reliability analysis tools early in the development cycle, device dzaigu can be euhaneed and failure risk reduced. A hierarchal order system and failure mode matrix were devehipad. These reliability anal)sis tools were utilized to identify critical componenU; is the totally implantable ventricular assist device being developed by this laboratory. Subsequent to this anal)sis, in vitro and in vi~'o tests were conducted to evaluate the integrity of the most critical parts and subsystems. Besed on the results of the reliability analysis and etitical component evaluation, design modifications were implemented, for example: staini~s steel bearings in the energy convertor were repiseed with eeramic-hybrld beanngs, the pressure transducer in the blood pump was replaced with an infi'ared sensor, a four-layer blood diaphragm was replaced with a single layer diaphragm, and the axial flow pump originally a reversing turbine was converted to a uni-directional design. These desigu modifications have improved device reliability and prototypes at various stages of optimizatiou have been operating failure free for up to 2.5 yeats in vitro. In conclusion, the methodologies used have proven to be practical, effective and affordable tools for reliability analysis. Early reliability analysis has lead to improved device reliability and reduced development costs. This preseutstion will focus on the analysis process and procedures, iu the hope that it will be of benefit to other investigators.

So21

The use o f mechanical circulatory support device for patients with recoverable hearts is a clinical reality, while the realization o f a reliable replacement heart is still a number o f years away, although there is little doubt that that day will arrive. The two ends o f the spectrum o f cardiac support devices are exemplified by the BVS 5000 and the total artificial heart. The former is in clinical use for post-cardiotomy support, while the latter is in the final stages o f engineering development in preparation for pre-clinlcel testing. These two devices not only represent different treatments for different patient groups, the design criteria are derived fi'om different premises. In addition to safety and effectiveness, features o f a temporary cardiac support device are based on simplicity, low cost, and compatibility With existing equipment while for the total artificial heart, size, weight, anatomic fit, hemocompatibility, biocompatibifity, and reliability are the key design guidelines. We will contrast the development efforts of two such devices, a tubular pump for temporary support, and a totally implantable artificial heart for permanent support.

88

91

EVALUATIONOF THE NOVACOR LEFT VENTIUCULARASSIST SYSTEM Phillip J. Miller, T. Juergen Billich, David H. LaForge, James Lee, Andrew Naegeli, Narayunan Ramasamy, Jal S. Jassawalla, Peer M. Portaer Novacor Division, Baxter neslthcare Corporation

THE DEVELOPMENT OF TEMPORARY AND PERMANENT VENTRICULAR ASSIST DEVICES (VAD3) Robert L Whalen, Ph.D. Whelen Biomedical Incorporated, Cambridge, Massachusetts

The Novacor LeR Ventdcular Assist System (LVAS) is an implantoble, electrically powered system intonded to provide chronic circulatory suppan to patients with terminni heart disease. Under development siuee 1970, it provides effective cardiac rehabilitation and a high quality of life. The system utilizes a unique, dual pusher-plate blood pump, actuated by a balancedsolenoid energy converter. The autonomous, self-regulating s)~'tem operates in s3~chrooens rounterpuisation with the natural heart, unloading the left ventricle while supporting the cotin: systemic circulation. The current configuration us~ a wearable, micTeprocessor-based controller and rechargeable battery packs. Ergunomically designed for portability, convenience and appearance, the system offers the recipient uniimited mobility and a high quality of life.

We have been developing a family of vantricular assist devices (VAD's) intended for either short or long term cardiac supporL From an engineering standpoint. the design requiromunts for these devices are quite distinct. Most recently, we have focused our attention on devices intended for permanent implantation. Permanent devices present unique challenges. In this situation, the cost of the device is not a limiting problem, as the most important factors are long term reliability and the quality of life provided by the device. In our approach, a skeletal muscle 0afisuimos dorei) is dynamically conditioned to perform continuous work. The muscle is first rate conditioned at low load pressure using an implanted myuntimulator in conjunction with an intrathoracic, externally controlled conditioning appliance. It is then load conditioned with increasing pressure& During this period of muscle conditioning, the drculation is supported with an implanted prosthetic blood pump powered by an external pneumatic drive systmn. When the muscle performance is deemed adequate, the conditioned skeletal muscle is then employed to hydraulically actuate the blood pump. The resultant VAD is then essentially a metabolically powered device, as the only source of electrical power needed to operate the VAD is the battery in the myoedmulator, a component with a typical 5 year life. if proven workable, such a system has the potential to offer a significantly improved quality of life compared with either other mechanical circulatory support systems or with cardiac transplantation and it's medical complications associated with immunosuppreanion.

The LVAS has been exhaustively tested to verify its safety, raggedness and reliability. Companants have been tested for electrical and functional safety, electromagnetic compatibility (EMC) and resistance to shock, vibration, crush, temperature and humidity. Durability of the system has been demonstrated through multi-year, multi-system In vitro endurance tests and animal implants to 10 months. A human factors assessment was performed to evaluate system usability. A multi-center clinical study of the device as a bridge to cardiac transplantation has been ongoing since 1984. More than 300 cases have been done to date, including the world's tint successful bridge to transplant and subsequeut successful implants to 370 days. Several ufthe 31 recipients currently on LVAS support are living at home and are back at work while awaiting heart transplant. A clinical trial evaluating the device as an alternative to transplant (i.e., for chronic support) is currently underway. The LVAS received the European CE mark in 1994.

89 VENTRICULARASSIST DEVICESFOR AT HOME USE E. 1. Burke, MSME :rCLThermo Cardios)stems,Inc., Wobum. MA Recogeizingthst the modish waitinglime for e denor beart entainues to grow, there is a need to de~.~I~Ve~icdm"AssistDevices(VAD) thatwill allow patientsto undergocardiactreatment outside the ho~iud setting. FCI is ~m'antly developing two Left Ventri~dar Assist Systems (LVAS)for mating pati~ts at home: the Hem-tMate*IF'LVAS w/HeertPak* and the ~ t o * VE LVAS. Both HcmlMste* LVAS's allow the patient n high degree of freedom while nmintslnlng cirenistray anppad. The IP LVAS enmists of an implantoble blood pump thnt is actustod by a pambis pneumaticthiv~ ( HesrtMato*l-le.~ak" ) powered by recbargeablc han~es. The VE LVAS is ~m~prisedof an imphmtsbleblood pump and eleetnc~ which is eoenectad to an extmud enntroller and batte~ park The VE LVAS ls ~nre~lly andergeing tes~ng in a nlinical study to evaluato its use o u ~ ~ hospital setting. Fourteen pstimts from six clinical teeters have participated in the study. Tm pade~ bave besurdeased fi'omthe hospitci en me to three dey passus. Four of the patients bave been discharsed frem'the hospital for lreehnent en an outpatient besis. Two of the patients rived st hamefor six and sevm moeths, cheeking into the hospital en a weekly basis, before anderguing transplantation. No dgnificent malfunctionshavebeenobservedoutsidethe hospital. Two of the fourteenpatients expired, six were transplanted,and six remain on ~ An eppliesfionhas bern mbmiaed to the FDA to be0n n clinical studyfor home use of the HeertMste| IP LVAS w/HeenPak*. The results to ttata show that patients with end stage heart disease supported with a 7"C1 HesrtMato* VE LVAS out of the hospital envirenment as a bridge to transplantatien have been highly sucoo~ful. The HcartMate* VE LVAS has beenshownto provide adequateeirouistory SUplXXtinsideaswell as outsidethehospitalenvironment.

S-22

Flow, Surfaces and Reactions/Interactions

92

95

TISSUE FACTOR ACTIVITY IN RAT VASCULAR SMOOTH MUSCLE CELLS EXPOSED TO SERUM STIMULATION IN A RECIRCULATING FLOW SYSTEM C.L Hall, *MB. Tanbman. *Y. Nan~:~c~ V.T. T~'ino. Dept of Biomedieal Engineering, The Uaivendiy of Memphis; *Mr Sinai School of Medidne

FLUID-STRUCTURE INTEP,.ACTIONS O F T H E ENDOTHELIAL-CELL GLYCOCALYX ANn BLOOD IN THE MICKOCIRCULATION E. R. Damlarm, B. R. Duiing, and T. C. Skaink, Departments of Biomedical Enginnertng and Physiology, University of V i r ~ i n , CharIotteevill~ VA 229O8

Quicam~ nr vamdar amenth mascle ceBs (VSMCs) exposed to growth media o~taiaing 1 0 % c ~ smart synfiz~c tissee factor (TF) md demcostrat~ TF activity boginaing at'~ 60-90 minutes. To determine ff flow pmueeters affected this response, monoisyers of culls were exposed to stimulatnty med/a in a r ~ flow Icop for 90 minutes. Ceils were cultored on Permanox brand chamber tildes and then placed in a perallel plate chamber canuented to a roller pamp supplying media at flow rates cutresponding to wail sha~ cates of 26,000,13,000 and 538 see". A indic (z~"o flow) cou/rol was paired with each cell isyer tested und~ applied flow. Following the applicationof media, cellswece rinsedwith Hank's Balanced Salt Solution + 0.1% bovine serom aibmain (I~SS/BSA) and TF antivity wes mesmred. Factor Xa (Xa) enncentration at the ~ outlst in the presence of faotoc Vlis and cadaium indieated TF ectivity and was measured at swell shesrrateof80 ser foe20 miantes. Effluent umplesobtsiuedat 1 . 2 , 3 , 4 , 5 , 6 , 7 , 1 0 , 12, 15 and 20 mijmtes ~ stesdy stateaider5 minutes.~ o r e , the last6 values were avecaged to cumpa~ staticelly troated culls and culis stimulated ander flow. In all of the paira examined at the two highest rates activity was stro~y diminldsed in ceils exposed to flowing media. FoUowing shcar at a rste of 26,000 sec't (n=4) the mesn steady stntc Xa coaceatrstion of 3,9 _+4,6 nM wss 65% ieas than Q~ ststic cuntrols at 11.3 + 2.4 nI~ CCI~ stimalsted and~ flow at 13,000 sr162 4 (n-3) d,owcd 82% less activity than those stimulated statically (1.7 + 0.14 nM v~ws9.3 +3.7 nM). T h r statisticallysigsff~:ant withp values of 0.01 andil,047 Rspe~vely (paired t-test). Thae was no diifeveueebetwrea ststic e~arols and the lower shesr rate of 538 ~r (n=4) which resulted in mean concentrations of 5.9 + 5.1 nM versus 5.5 + 6.1 nM (o=0.38), respe~valy. Tberefare, the TF reslxmse of rat VSMCs to sedan is demcesed or ahered at high wall shear rates, but similar to static stimulation at moderate shear rates,

The luminal surface of the anduthelial culls comprialng the capillary wall is ouated with a glycocalyx which consists of glyenproteins and proteoglyeans. The poeslble effect, of the glyenealyx on the rhoulogtcal behavior of the blued/capillary system is investigated using o0ntinuum medmnice of h e - - u s media. As a first apprmlmation we consider the flow of a dingle-fits fluid suspendon of axi-cymmetflc, done]y-fitting rigid pellets in a cylindrical tube lined with g porous, deformable wall layer which shnuintes the glycoealyx. The fluid is modeled an Newtonian using axi-cynunetrie lubrication theory and the wall layes Is modeled es a biphanie nutterinl using mixture theory. The two phases ennsint of a linearly elastic solid constituent and a Newtonian viseous fluid cunetituant. Considering easse both with and without peflets, analytic anlutinue are derived for the velocity profiles over the croewueetion of the tube inciuding the distribution in the wall inye~. Forthe case with pellets, the Rcynolds equation is derived for the preesure gradient along the pellet. Results are presented in terme of wall layer thieknem and permeability. E~timat~ of these ~ fc* the glycucalyx ure taken ~ experimental data for fibrinogen gels formed/n-e/fro (Blomb&ck & Okada, Throm$. Rea. (25) 51-70, 1982). For a wal.l-lltyer t h i ~ which is 10% of the tube radlue, IA times the pressure drop is required to achieve the same volume flow as wouid occur without a wall layer. If, in the presence of this same wail layer, a rigid epherical pellet in Introdaned which is 89.0% of the tube radius, 2.5 times the driving pvee~lre is roquLred to produce the mm~e volume flow as in an eqnivelent system without a wall layer. Supported by NIH gr~mts HI, 12792, HI, 07284 and HI, 49145.

93

96

T H R O M B O L Y T I C THERAPY UNDER A R T E R I A L CONDITIONS Srirara Aannd, Jung-Hr Wu. and Scott L. Diamond Bionngineedng Laboratory. Dept. ChE. SUNY, Buffalo. NY 14260.

LARGE SCALEMODELINGOF THE VASCULARBLOODFLOW RESISTANCEGENERATEDBY AN ADHERENTLEUKOCYTE G.B. Chapman,J. Gessertand G.R. Cokelet Degattmentof Biophysics,Univem~ oi Rochester, Rochester,NY 14642

Emergency ur,armont within the first hours of myocardial infarction commonly includes i.v. infusion of pinsminogon activators for the dissolution of an arterial thrombas. Using in vitro kinetic parameters, we have conducted dynamic simulations of the systemic circulation during dirombolytic therapy. The simulation of the reacting plasma agreed well with tPA and plasminogen levels measured in humans dmlng therapy. The results of this simulation were then used as the inlet conditions for transient simulation e r a dissolving fibrin clot where plasma species pormented and diffused under arterial conditions (hP/L = 60 mmHg/cm). Solution of coupled PDEs for convective-dispersion of reactive species (tPA, uPA, plasminogan, pissmin. ct2-antiplasmla. PAI-I. macroglobulin, and several inhibited complexes) in dissolving fibrin allowed predlntion of thiombolyticrates for a given therapeutic regimen. We have validated these: simulations for/n v/trosystems involving diffusion or pressorr permeation of enzymes into fibrin gels and recalcifred platelet poor plasma clots. We predicted that the position of al lysis front can move about 1 to 2 mm in 15 min during artedal thrombolysis. At the lysis front, fibrin bound plasminogen was rapidly depleted by the bound tPA. generating very high levels of plasmin bound to the fibrin. The zone that contained the peak amounts of plasmin and tPA bound to fibrin was at the active lysis front since both of these species adsorb to fibrin with high affinity. In this reaction anne. the antiplasmin and PAI-I in the free phase we~ completely depleted due to compisxadon events during the pr~eding 15 min. Since active tPA and pissmin could survive at the iysls front, thear two species were rapidly adsorbed by the dense fibrin strocturr as they permeated forward and then were subsequently released during lysis, This lad to a dramatic concentrating and continual accumulation of tPA and plasmin at the lysis front The fluid permeation front was actually ahead of the lysls zone, and this was seca with the irddbited Complexes that do not bind the fibrin and hence permeated through the lysis fronL These studies are dircete, d toward establishinguseful guidelines for drug and therapy desig~ based on ~ in vitro properties of the activator.

Leukocyte adherence to the venular endothetiumcan result in increased resistance to flow, especially as the vessel diameter D approaches the diameter of Ihe heukocyte(d). We have shown, using a large scale model, that the resistance to flow of a Newtonian fluid past a sphere, adherent to the wall of a cytinddcal tube, does not increase significantly until the sphere to tube diameter retio (d/D~,is approximately0.50. The flow resistanceincreases rap~y for d/O> 0.50. This study addresses the possible effects of red blood cells on the flow resistance generated by an adherent leukocyte. The large scale model consists of: (iI rchher disks suspended in high viscosity sgicane oil representing red blood cells and plasma, (it) ~id spheres representing lsukocytes and (iii) rigid cylindrical tubes reprasenflng the blood vessel. Espenmantswere concluctedat d/O = 0.33 ahd 0.50. Tube hematocnt (hct) was varied between 0 and 20*/, by adding disks to a stirred feed reservoir. The flow rate was controlled by a sydnge pump. Pressuredrops across an edhetent sphere, in the region o| disturbed flow, were measured for vadous flow rates. The effective wscosity was calcdatod by using the measuredpressure drop (AP) tot vandus hct suspensionswithoul a sphere fixed to the wall. , The Ap across the disturbed region of flow was convededto a Fanning fdcfian factor (0 and the flow rate was cenveded to a Reynolds number (Re). Withoutdisks f*,Re was leund to be a hxctien of anly d/D. As long as the effectiveviscosity ol the oil and disk s u s p e ~ was used, f.Re did not signiticantiy change for d/D = 0.33 or 0.50 e~jardless of hct. This indicates that the continuum model of blood is vaTidfor d/O < 0.50. This d/D ratio is n conservativeesl~malebecause our model disks are not as flexible as actual ted blend cefls.Thus, the continuummodel could be velid in vivo for d/D 9 0.50 and hct 9 15%. Using relationshipscbtained for the force an the sphere, the force or the wall for various hct can be calculated. Supportedby NIH 18208

94

97

IMAGING METHODS AND EARLY RESULTS FOR FREEZE-CAPTURE-SECTION STUDIES OF PLATELET CONCENTRATION PROFILES Xianhai Chan, Eugene Eckstein University of Teonessen0 Memphis. 899 Madison Ave. Suite 801, Memphis. "IN 38163

M U S C L E S T I M U L A T I O N A N D L E U K O C Y T E - -ENDOTIIELIAL C E L L INTERACTIONS. William P. Selent and IogridlI.Sarelius DCl~t'Unent of Biophysics,Universityof Rochester, Rochcster N Y 14642.

The frenze-capturo-ssction technique (FCST) is relatively simple to accomplish and yields useful results when fluorescent beads are used as pintoiet analogs; such studies demoustrated many facts about the flow-asu~inted development of uear-wali excess of plateiet-sized objects. In rive studies have demonstrated that plateiet concentration in the microcimulahon is elevated, indicnting that the event is of thcoiogtoai and not biological origin. Extending this technique to platelets has proven more difficult than anticipated, despite valuable work by several groups investigetot's who have used f l ~ t i y labeled platelats to study of thrombotic depo6ition onto surfaces. In retrospect, there seem to be two aspects of the FCST that make it more difficult than fluoresenuce studies of thmmbuais ,t. platelat deposition: the first is the need to distinctly Rsolve and identify individual pisteiets - neither misses or double counts are tolerable, which differs from prior work where the goal is to resolve amounts of material or residence times, and second, r162 need to identify these platelm in images with a significant amount of focused background fluerescenee (thin sections.are nee possible), While flow chamber studies also involve large levels of background light, the marginal layer with its relative deficiency of cells in combination with the thin focal zone of high NA objectives provides a more uniform background, a situation n:medied well by collecting imag~ with a relatively large number of grey levels. Recent work, using all expected tools (high reanintion 12-bit C'CD. image analysis, monocloual staining for specificity, etc.) shows that we am able to image platelnts sufficiently well to use the FCST with ounfi-denue. Particolars of the image processing technique and the demonstration of profiles for pinte-lels in flows of whole blood will ha shown. Current effort is to measure p]atelnt concentration profiles in flows in tube with albumin (to model a non-reactive surface) or fibriuegen (to model a reactive surface) coatings and to fit the profiles to a convective diffusion equation with drift.

Our laboratoryhas been concerned with describing the in vivo behavior of ]eukocyles as well as attempting to discern the relative importance of the microvascular bemodynamic conditions and molecular mechanisms governing leukocyte behavior. W e have observed highly variable leukocyte flux distributionsand ieukocyte-endotheliaIcell interactionsin differentvenules within the same tissuepreparation. This variabilitycan occur in vesselsof similar diameter, class,prevailingflow conditions,anatoralcallocationand locationalong a given vessel or vascular tree. To furtherundersUmd the conVibutJon of shear conditions we physiologicallyalteredblood flow using fullfieldelectricalstimulation.Electricalstimulation is known to produce changes in arreriolardiameter,blood flow and blood distributionwithina vascular bed. In anesthetizedmice (Sodimn Pemobarbital 75 mg/kg [P) we sdmulated the cremastermuscle for pednds of fiveminutes at I and 2 Hz frequencies(25 ms pulselength,515 Volts) The number of rollingand adhering Icukocytesin venuies of diameter 11-27 l.tm was quantitated. Flanrescentlylabeledred ceilswere used to measure red cellvelocitywhich was then used to estimate the Newtonian wall shear rate, ~/w. No currelationwas ohaerved between the numher of roilingor adhering cellsand the Newreaian wall sheas rate. W e did observe an increasein post exerciseleukocyte interactions(292• of controltollingceils, 161:E31% of control adhering cells) and this increase occurred in both the p .r .r~mce and absence of higher Newronisn wall shear rates. Group 1 (increased shear rate): A Yw = 179,2:19% of control, /,rolling = 396+70%, ~ndherlng ffi 105• Group 2 (No change in shear rate) A ~/w = ]02:k10,7% of control, Arolllng ffi 314_+67~, ~adhering = 162• These data suggest that either leukocyte flow redistribution occurs with exercise and/or that moiecaiar even~ associated with exercise and their influence on adhesion molecule expression are the principle factors governing lenkocy te-endo thelial cell interactions during exemlse for the range of A ~ w observed in this study.

Heart Valve Prosthesis

S-23

98

101

FLUID MECHANICS OF PROSTHETIC HEART VALVES AT CLOSURE J C Maymlr, R S Meyer*, S Deutsch', D B Geselowitz, J M Tarbell Bioengineering Program and Applied Research Laboratory', The Pennsylvania State University

Comparison of Catheter and Doppler Derived Pressure Gradients Aecress St. Jude Valve Prosthesis in the Mitrnl Position: An Intraoperative Study Pieter M. Vandervoort. Dominic Y. Leung, Neil L, Greenberg, Delos M. Cosgrove III, James D. Thomas. Cleveland Clinic Foundation, Cleveland, OH.

Most studies of the fluid mechanics of prusthctlc heart valve* have f ~ on forward flow through the open valve where turbulent (Reynolds) stresses on the order of a few thousand dyneaora2 have been observed, Tilting disk valves, however, usny a significant volume of fluid in a retrograde or regurghant dtrectlan at valve elusure, and there is a sustained leakage flow betwoen the disk and the support ring while the valve is closed. We have investigated the turbulent s~re~ns generated by these flowsusing 2-dimenslonal laser-Doppler anemomew/(IDA) in an anifielal heart environment and 3-dimensional IDA in 9 natural he.an envtronmenL Bjork-,Shileymonostrut valvea with Delrin dbks were employed and me,~urements were obtained in four near-wall profiles extending axiallyfrom a puslilon only 0.5 mm from the valve disk out to a distance of 9 mm in the major and minor orifice region of both the mitral and aortic valve. High time resolution (1 ms time windows) was achieved In all measurements. In the sustained leakage flows turbutent wall jets are formed which have Reynoldsstressus an order of magnitude higher than observed in forward flow. Very near the valve, short temporal spikes (a few milliseconds) of emremely elevated Reynolds stress are detected right at elusure. These elevated stresses may contribute to blood damage. Acknowledgement: This work was supported by NIH Grant HL 48652.

We have previously shown using in vitro studies mad numerical simulations the importance of pressure recovery in the pressure-flow relationship across bileaflet heart valves, We presently evaluated (he impact of pressuse recovery across bileaflet mitral valve prosthesis in the clinical setting. We studied 5 patients undergoing mitral valve replacement with a SL Jude prosthesis. Simultaneous leR atrialpressttre,leftveatrlcular pressure and transvalx~alarDoppler velocities were obtained. The central and side orifices of the prosthesis were interogated separately. Doppler derived gradients were calculated across central (APo~c) and side orifices (APo,~s) using the simplified Bernoulli equation and compared with the involve transvalvular gradients (APc,~. Subsequently a previously validated pressure loss coefficient (K=O.Ot) was incorporated for gradient calculations through the central orifice (cAPt~). Results: APcm ranged from 4 to 9 mmHg (6.2~l.2mnfl-Ig). AP~o# ranged from 7 to I I mmHg (9.1• mmHg) and overestimated APc4,~by 47+90%. APoo# was 6.9r mmHg and agreed closely with APc~ (15r difference). When K=0.64 was incorporated in the Bernoulli equation, c.APc~c ranged from 5.2 to 7 mmHg (5.9~0.7mmHg) and differed from APc~ by only -6• Conclusions: Differences between velocities across central and side orifices are present in miwal St. Jade valves in patients. Pressure recovery hi the central orifice of the St Jude valve is present in the clinlcel situation and can be corrected by the experimentally determined pressure loss coefficient K=0.64.

99

102

RE~URGITANTJETS THROUGHBILEAFLET~ C A L HEARTVALVES Jeffrey T. Ellis, Tanothy M. Healy, Arnold A. Fontstne, Ajit P. Yoganathan School of CbemicalEngineering,Georgia Instituteof Technology, Alhinta.GA

BIAXIALMECHANICALBEHAVIOROF FATIGUEDB I O P R O b ~ C LEAFLETS M.S. Sacks and D.B. Smith Department of BiomedicalEnginenring,Universityof Miami, Coral Gables,FL 33124

Fluid stteas~ occtwrleg in retrograde flow fields during valve elouue may play a significantrole in thrombegenesis.The closingflow and regorgitsntjets present during diast~e can cansc damage to foaned blood elements due to high levels of shear stress. The turbulent nstete of thesejets may imposeshear stresses on abe blcodelemcots which exceed Iowor limits for bloodelementdamage. Lethald~ma,,e to red bloodcells can occur with fluid shear sucsscs of about 1500 dyoes/cm2; even less for blood flow over a foreig~ smface. There t ~ flow fields md tbeir clinical implioailoeshave ant beco previouslysmd/ed hi great detail In vitro experiments wore ~nth~oed to investigate the llme-vm'y~g ~ flow ~ of 27 mm bileafletanrdc veives. Thten-amlpcl~t, cuiacideat Laser Doppkf Anemomeay velocity mensmements wan obtained facilitatingthe determinatioo of the 3-D princilmls ~ io the valveflowfields, The expatmmts wan porfmmedin abe Gent'gtaTech aortic valve model trader physiologicpolsatileflow ccodliicosof 5 l.Jmin nteen amhac Output m d 30 L/mln peak systolicflow rate. A reseeableclork wus empioyed to gate imisatiledeta ecqutsifion throughout cod systole trod dias~e over ~ ~ cycles for phase w~dow averagingand the geheratice of mean velocityand turbulence statistiraover 20 ms intorvals. A ttgim ~lrimato~y 0.l hi2 W&q ~ l mm ul~tffanmof ooe hinge with an incremental resolutintt of 0.005-0.010 in. Animation of the data allowed Ibe lavestigailotlof the full temporaLcharact~ of abe flow mid ~ that Im'tmlentshear s ~ s e r ranged from 500t000 dyneaem2 in abe closing flow md 1500-3500 dymaem2 (Ix*k) in the region of the regurgitantjet High ~ were observed to lmmistin the jets throughout4lastoh,. This t'esem~ should yield a better nademtmdingof dm signil'tceacoof transvuivuiar fluid ~ c s to the thrombegenicityof btlceflet mechanical heart valves m~l ~ in the develop~t of row d~gns.

Porcine bioprnstheficheart valve(PBHV) leafletsare essentiallymft-tls~ membranes that consist primarilyof a dense netwc~ of collagen fibers. PBHVsgenerally have excellent shooterm dmability, trot poor Iong4ormp e r ~ . This poor durabilityis tumallythe rnsalt ~ collagen fiber disruptionI.~,,41.~to leaflet tearing at the flee edge or at points of leaflet attachment to the supportingsmut, along with calcification.Althoughthe exact mechanism is stallunknown, it is alm~t certain that mechanical factms play a signLficantrole in leaflet tearing and calcification.Ualaxlal testing of thin tlramestrips ent titan PBHV leallats have been mmd to r thalrmachaui~ ~etmrtiet Although adequate fc*qualimive comparian~ this meth~ cmmotbe nsed m Rdlyc h ~ c t e ~ tim meelmniuslbehoof of leaflettissueswhich normallyexperience mnltiaxialloading Binsial meehamcal testing fully characterizes the PBHVanitmtropicmechanical ~,xsT~~des and resolvesthe limitatinm of unlaxi~ tuning. In thi, study wa performed blaxia mschaniceltesta an lenflblsf~umbotli virgin (i.e. untested) and aorehrated testod(fatigued)PBHVvalvular m . The trot preteco/ wm begin by geeoeditiot~ug t ~ Wecimanandor equi-t~m.t mal~ Next, tern Mra l~rform~l in whieh tim m o bet~en We m;c~mumaxuil strains wmakel~ r o t , usi~ r~insofl:3, 1:7.,l:l, 2:l, nad3:l. Acceler~ted~weral~t~'m~~1200beat~mim~ up to a to~ of 50 to ~500millioncycha. In tbe v i ~ ~ the le~et cucomfemnhal ~ t i ~ achlaved~.~en tim~ tim puk IoszLcompwed to the ~ directi~ under equi-blaxlalsmscM Fatigued lealletscowm,'teutlybecame ~ in both axial directions. "rnus, si~ni~mt e l m ~ th fixed PBH3/mt~hatficalpropetsies can occk~by In-vitro Irm,4in~I~_ t ~ ' ~ alone. Tbe~ changes in mechanical rmpon~ may ennm'buteto poor PBHV durability,and need to be enmpe-f~__e,J for in new PBHV designs

100 LASER OPTIC METHODS

IN V A L V U L A R

FLOW

DYNAMIC

RESEARCH

Ned H. C. Hweng, Ph.D. Department o f Biomedical Engineering, University o f Miami In vitro measurement o f catalina valve flow dynamics are benefitted by recent development in laser optic techniques. Coherent laser light offers many edvamages in pulsatile flow dynamic measurement, including: no physical probes (or uan.ulucen) present in the flow field; high frequency response; no cafibrafion for most applications; and unaffected by the temperature, density, or other changes o f physical properties o f the fluid media. Recent development in laser optic methods for valvular flow dynamic measurements discussed at this presentation include: 1) the optic-electro-hybrid feedback LDA (OEHF-LDA) for back scattering and near surface measurements; 2) the Scanning I D A for instantaneous mapping o f velocity profiles; 3) the Laser sweeping technique (LST) for measuring occluder/housing impact velocity, and 4) expanded laser beams for measuring tissue valve opening area and conduit compliance.

S-24

Cardiovascular Biomaterials

103

106

FLOW cY'rOMETRIC EVALUATION OF BI.J3OD MATERIALS INTERACTION: PLATELEr. LEUKOCYTE AND COMPLEMENT ACTIVATION Cynthia H. Gemmell, John P. Black, Erik L. Yen, Michael V. genoa Department of Medlcinc and Chemical Engineering, Universityof Totonto, Canada

CHARACTERIZATION OF INTERFACIALLY PHOTOPOLYMERIZED THIN HYDROGELS IN CONTACT WITH ARTERIAL TISSUE Lyman, M,D,, Melanson, D., Jarrett, P.K. Chan, L., Sawhnev. A.S. Focal, Inc. Lexington, MA 02173

Vascular biomate~als continue to present significant medical problems through failure and incompatibility. A better understanding of the mechanisms involved in biomaterials failureand development of effe~ve means to qoanlitate blood activation will fu.qher understanding of the role of surface chemistry in thrombngealcity and lead to effective evahtetion of binmatesiah. We are using flow eytometry to identify material induced cellular activation events. In our In vitro test, hepariniznd whole blond (WB) wus gently rncknd within 25 cm lung" 1.57 mm ID tubes (polyethykam PE, polyvlnyl alcohol PVA hydragel, SllusticTM)for oac hour at 37~ Aliquote of WB were analyzed by fluore~ent ustivated flow cytomegy (FAFC) for platalat activation (micropardcle formation) and lenkncy~ activation (up-regaiation of CD 1Ib). Mioroparticles (MP) were identified as those positive for GPIlb/IIIahot ~oaller than intact pintele*s based on forward light scatter. One hour contact with Silarti~ polyethylene and poly (vinyl alcohol) hydrngel susfacus lead to 30~1, 33~:4, and 43:t:4, x 109 microl~tleles/L respectively whereas h:sfing blood samples cootalncd only 10il x 109/L. In conttust, only WB cootact with PVA surfac~ led to leukocyte activation; 300% increuse in CDllb m=eptor density as rdlected by fluore~ent intensity. Given the known impact that complement activation has on cellular activation we used Enzyme Linked Immuncsothant Assay (ELISA) kits from Quidel to uss~s sCSb-9 (membrane attack complex) and Bb (breakdown of C3 converUtses) levels following material contact. Only WB contact with the surface that led to appresinble leukocyte activation. PVA hydrogel, led to detectable levels of sCSb-9 (6.58 :t:0.4I ng/ml). The Bb levels were very low although above background for all three surfaces.

INTRODUCTION: interfacial photopolymerization has been described by Sawhaey et al.l wherein the chromophore (Eosin Y) is immobilized on a substrate and can initiate polymerization from the surface of the subs*rate outwards, into the macromer solutions upon illumination with visible light. Variables, such as macromer concentration, chromophore (Eosin Y) concentration, and accelerator concentration (N-vinyl pyrrolidinoan) and illumination conditions, that affect the process of interfacial photopolymerization and the resulting hydrogel barrier thickness were investigated. Degree of conversion of the resulting interfacially polymerized hydrogel barrier by way of compressive stiffness and end group analysis methods was also characterized. RESULTS: A linear relationship between barrier thickness and Eosin Y concentration was observed for beth a 10% & 23% w/w macromer concentration at a 30 second illumination time & 1 [.tl/ml VP concentration. An increase in hydrogel barrier thickness was observed with an increase in VP concentration at any given illumination time. A linear relationship between barrier thickness and macromer concentration at a 30 second illumination time was observed in the range of 0-2 BI/nd. End group analysis by IC of the interracial hydrogel barrier indicated 75% conversion which corresponds to a stiffness of 40 kPa. DISCUSSION: It can be seen that each of these variables has a directly proportional effect on the photopolymerization of the hydrogel barrier, The IC method can be used to determine degree of conversion of an interfacially formed hydrogel barrier on tissue, from which gel stiffness can be calculated. This allows a great measure of flexibility and control in the engineering of such barriers for various biomedical applications. REFERENCES 1. Sawhney AS, Pathak CP and Hubbell JA. Modification of islet of Langethans surfaces with immunoprotective poly(ethylene glycol) coating via interfacial polymerization. Biotech. and Bioeng. 1994. 44:383-386.

104

107

EFFECT OF NITRIC OXIDE DONORS ON EARLY PLATELET ATTACHMENT TO BIOMATERIALS UNDER PHYSIOLOGIC FLOW. J Worse),, K Kalia, K Lahadie, C Hong and P Johnson. U~versity of Pittsbusgh Medical Center, Pittsburgh, PA. Prevention of early eeclusinn of smaB diameter prosthetic grafts remains a major challenge as platelst adhesion to the biomaterial and the development ofa platelet rich thrombus initiates early occlusion. The effect of nitric oxide, an endogenous platolet inhi'biter, on this interaction was studied in a customized perfusion chamber using whole blood with Indium''t labeled pintelets at low shear and 37C. Two well characterized nitric oxide donors DEA-HO (EhN[N(O)NOlNa) and MAHMA-HO (CHvNH2*(CHD6N[N(O)HO']CH~) were used as a scor~ of NO (1-2 11101NO released/tool catTier). Polyethylece (PE, 0.86ram internal diameter) and expanded polytetraflanrcethylenc (PTF~ lmm ID) conduits were used and pintelet attachment measured by gamma coenting of perfused specimem. Both DEA-NO and MAHMA-NO inhibited ADP induced platelet aggregation at cortcentratious between 1 and 10urn. However, DEA-NO (0.01-10uM) paradoxically increased platelet adhesion to PE by 100-150% following peffusion with 5nil of blcod, an effect reproduced by DEA alone. MAHMA-NO reduced platelet depusitioo to PE by 20-25% (Smi perfusinn, 10-100uM, P
105 DELIVERY OF CELL-MATRIX MODULATORS VIA ENDOLUMINAL HYDROGELS IS A VIABLE S T R A T E G Y FOR LIMITING NEOINTIMAL HYPERPLASIA Stephen P. Massia and Marvin J. Slepian University Hcan Center / University of Arizona Collogu of Medicine, Tucson. AZ Mechanical injury to the vessel wall such as su-ctch injury during angioplssty or compliance mismatch in the anastomotic sites of synthetic vascular grafts triggers a healing process known as neointlmal hyperplasia (NIH). In response to injury and the release of blood-berne growth factors" medial smooth muscle calls (SMCs) become activated, deposit excessive amounts of cxtrscellular matrix (ECM), and migrate into the subintima to form a neointima. This neointlmal inyr progressively increases in volume ~ulting in wall thickening, luminal anrmwiag and decreased vessel or graft patency. While migration of SMCs into zones of NIH contributes greater than 50% of the cell mass, to date limited studies have examined anti. migratory drug strategies for reducing post-injury NIH. We previously demonstrated that interference with SMC integrin-matrix interactions, through local edventitial delivery of a cyclic RGD peptide (cRGD) via a polymeric hydrngel vehicle, resulted in a 90% rednctlen of NIH at 1~ days post-injtm/in rat carotid arteries. To further develop this local delivery strategy for clinical applications, we employed a previously described PEG-lactide hydrngel polymer which may be applied via a catheter as an endoinminal coating for sustained local delivery (PNAS 91:5967, 1994). We examined the effects of local delivery of cRGD to ballooa-injused rat carotid arteries via the endoluminallv applied PEG-lactide hydrogel polymer. At 14 days post-injary we observed a significant reduction in hyperplesia in r treated rats (compared to untreated and drag-free gel treatedballoon injured controls). These studies indicate that catheter-based endoluminal application of hydrogcls loaded with antimigratory agents such as cRGD is an effective and clinically applicable approach for limiting post-injary NIH. In conclusion, SMC-matrix interactious may be an additional target for pharmacologic manipulation aimed at limiting NIH following angioplusty and ana.stomotic sites in synthetic vascular grafts.

COLLAGEN/POLYMER COMPOSITES R. A. Berg and W. Rhce Collagen Corporation, 2500 Fabcr Place, Palo Alto, CA 94303, USA Collagen has been successfully used as a biomaterial for decades. Its use has been limited to specific applications which exploit its unique properties including its hemostatic properties, its tensile strength, or its unique gelation properties. Recently, hydrated collagen composites have been prepared by crosslinking fibrillar collagen with difunctionally activated succinimidyl polyethylene glycols (di-SPEG). (US patent 5,292,801.) Hydrogels produced by combining collagen and PEG arc more elastic and have higher tensile strength than collagen gels alone. Strings were prepared by polymerizing 35 mg/ml Zyderm| I (ZI) with 10 mg/ml di-SPEG i n TFE tubing with an I.D. 1.2 ram. The strings were air dried and tested using an Instron model 4202. Collagen composites crosslioked with di-SPEG were also used to prepare small diameter tubes and stems that maintain their shape and strength in a vascular environment. Mechanical Stren[th o f Colla[en Strings Crosslinked by Diameter Length Tensile Strain Stress Sample* (mm) (ram) ( N / m m 2) ( d L / L ) Z-I + di-SPEG 0.46 20 206.00 0.243 Z-I 0.46 20 110.67 0.086

di-SPEG Young's Modulus ( N / m m 2) 847 1225

Current

and Future

Trends

in Mechanical

Circulatory

Support

II

S-25

108

111

ROTARY B L O O D PUMP DEVELOPMENT AT NIMBUS K. Butler, D. Thomas, T. Rintoul, H. Harry, L. Taylor;, Nimbus, Inc. - Rancho Cordova, CA H. Borovetz, J. Antaki, 1L Kormos, B. Griffith, P. Litwak, University of Pittsburgh - Pittsburgh, PA

EFFECT OF PERFLUOROCHEMICAL MECHANICAL TRAUMA

Nimbus and its development partner, the University of Pittsburgh Medical Center COOP), believe that technology involved with rotary blood pumps will play an important role in shaping future trends in mechanical circulatory support. Accordingly, the Nimbus/UOP tenm has actively been engaged in several projects whereby both centrifugal and axial flow blood pumps have been designed, built, and tesfed. In the case of the axial pump, a long-terra implanted device that can be used either for left or right ventricular support is under development. The pump itself is sufficiently small (76 gin/66 en) such that it could be applied Jto both male and female patients. In this regard, the centrifugal blood pump is a miniature extracorporeal version (13 cc prime volume) intended for postoardlotomy cardiac support of neonatal and pediatric aged patients. To date this pump has performed successfully in seven animals for periods approaching 7days.

Blood mechanical trauma is one of rite major problems in the development of heart assist devices. We found that a 20% replacement of pinsma volume with perfluorocarben emulsion (Fluosol) remarkably reduced hemolysis compared to control during pumping of animal blood with a centrifugal pump. We hypothesized th~ Fluosol particles compete with RBCe for the near-wall space and thus decrease the time of RBC contact with the foreign surface. To test this hypothesis we performed experiments wherein blood samples from the very near wall space were obtained directly into mlcrohemamcrlt capillaries. These experiments showed that near wall hematocrit of flowing blood with Fluoanl was significandy lower thin corm,el despite the bulk I-h being the same. TO investigate another possible mechanism of reduced blood tra.m.; namely the ability of Flunsol to enhance viscoelastic response to applied stress, we studied the flow behavior of diluted FIuosol emulsion, which we found to be similar m solutions of drag reducing polymers. Such polymers axe known (H.L.Greene and S.R. Madun, 1975) to reduce turbulent frequencies and elongadonal strains with aCcompanying reduaien in the level of waU shear stress.

109

112

LONG-TERM TOTALARTIFICIALHEARTS: DESIGN CONSIDERATIONSFOR TRANSITIONTO CLINICAL USE AJ. Snyder, G. Rnsenborg, WJ. Weiss.W.S. Pierce The Pennsylvania State University

COMPUTERIZED MULTIDISCIPLINARY DESIGN OPTIMIZATION OF CIRCULATORY SUPPORT SYSTEMS

Long-term heart replacement devices cunendy in development are designed for their eventual clinical use, but are ~ted for years in mock circulatory loops and animals. This paper outlines some of the challenges we face in predicting requirements for human use and interpreting laboratory results as they may apply to clinical situations. Clinical artificial hearts are typically designed to provide cardiac outputs of three to nine liters per minute. The experimental models in which the devices fit grow very quickly to require the full capability of the device. This makes the extrapolation of data relating to Ihe control of cardiac output difficult,leaving important questions about control methods un.answered. Bovine eryfl~ecytes are smaller thun are those in humans, and there is some evidence that bovine red cells are accordingly less fragile. This raises questions about devices that cause borderline high hemolysis in calves, as the results in humans may be more signilicanL Determining the relationship between bovine and human hemolysis rates, and bow this relationship may be afftcted by specific hemolysis mechanismsis a challenge for the hematology and cell mechanics communities. In laboratory systems or systems for in-house use. emphasis is on biocompatibility and longevity. In the design of products, additional factors apply, including manufacturability,product labeling, padent and care-giver training, regulatory compliance, and cost. Consideration of these factors affects fundamental features of a system's design. For example, the size of an implanted battery depends not only upon the pump's energy requirements, but also upon our estimates of patients' requirements and preferred behaviors. Similarly. any alarms and indicators designed into the system must take into account the abilities and expectations of a device recipient who is trained in the use of the system, but is not an expert in its workings.

110

DAVID B. MELVIN, M.D. UNIVERSITY OF CINCINNATI Mechanloal Reactuation of the Cardiac Ventricles

Myocardial power failure causes cardiac death. Blood-contacting surface failure is the prime limitation of device trea~nent. "Re-powering" of a heart could circumvent these difflcu Itles---cormcting only the true deficit, power. There are, however, difficult requiremenls: (1) preservation of valve competence, coronary flow, rapid low-impedance refilling and independent left and right pressures, (2) avoidance of wall ceaptation, (3) hardware fitting available space, and (4) adaptability to deliverable power. Despite earlier interest in acute devices and later study of muscle wraps, little systematic work has been done toward elucidating these obstacles. Concerted efforts towards (1) developing research techniques for their study and then (2) finding mechanisms to meet them could well yield one or more effective modalities which circumvent a ms~er obstacle to the indefinite and cc6t-effective mechanical treatment of heart failure. Modeling of the passive behavior of failing_human myorardium, both by digital finite element and physical simulations could facilitate choice and screening of surgicel and engineering approaches. Prosthetic models, excised transplant recipient hearts, unaltered pig hearts, existing/11 v/re animal failure models, and failing cadaver hearts all have serious limitations as physical analogues; we have develope(J a usablepreparatlon by altering excised pig hearts. The coob~nuing protocol is (1) refinement of the analogue and development of complementary finite element modeling, (2) in vitro observation of specific deformation patterns and heart anchoring modes without regard to size, biocompatibility, or durability of power mechanism and delivery, and (3) systematic consideration of the practical issues for promising systems.

EMULSION

ON

BLOOD

M.V. Kameneva, H.S. Bomvetz, and J.F. Antaki University of Pinsburgh, Medical Center, Pittsburgh, PA 15219, USA

James F. Antaki (*), Omar Ghattas (t), Gceg W. Burgreon (*), Beichang He (t), Jonathan Cagan (t), and Harvey S. Borovetz (*) (*) University of Pittsburgh and (*) Carnegie Mellon University, Pittsburgh, PA. The development of mechanical pomps to replace or support the natural heart is a formidable undertaking and requires muitidinciplinary expertise from a wide range of scientific and engineering disciplines. As theoretical models relating form and function become more aophlatieated, and with the continuing advances in computer hardware, the viability of computerized optlnfization to assist, and indeed partially automate, the design process is increasing. The physical phenomena governing the performance and biocompntibility of these devices are nonlinearly coupled. Accordingly, to achieve an optimum design, competing r e q ~ o n t o must be compromised through iterative refinement of the design parameters. The computer is an ideal platform for accomplishing this task. Numerical optimization is being closely integrated with computational fluid dynamics analysis, blood damage models, and electromagnetic analysis to optimize the design of a rotary artificial heart. Zereorder algorithms, such as genetic algorithms and simulated annealing, have been explored as well as higher order methods such as sequsutial quadratic programming. The progress to date demonstrates encouraging premise for further application of this technology in artificial organs development.

S-26

Atherogenesis and Vascular Endothelium

113

116

VASCULAR ENDOTHELIUM IN ATHEROGENES|S: AN INTEGRATOR OF DIVERSE PATHOPHYSIOLOGIC STIMULI (OVERVIEW) Michael A. Gimbrene, Jr., M.D. Vascular Research Division, Dept. of Pathology, Brigham and Women's Hospital and Harvard Medical School, Boston MA

LOCAL ELASTIC MODULI IN BOVINE AND PORCINE AORTIC WALLS MEASURED BY PIPETTE ASPIRATION M. Sato, I". Ohashi, T. Matsumoto, and T. Aoki* Tohoku University, Sendal, Japan * University of Tokyo, Tol~o, Japan

Vascular endothelium forms a dynamic interface between circulating blood and the vessel wall. Its ability to respond to locally generated and blood-borne endogenous mediators, such as cytokines, hormones and growth factors, as well as exogenous products, such as bacterial endctoxins, represents an important aspect of its physiological and pathophysiological adaptations. In addition to these biochemical stimuli, there is ample evidence that biomochanical forces, generated by the pulaatile flow of blood through the branched tubular geometry of the circulatory system, can act on the endcthelium to modity its structure and lunction. Several of these alterations, which involve increases (or decreases) in the production of key biological effector molecules, including vasoactive mediators, adhesion molecules, chemoattractants, growth factors and pmcoagulant/fibdnolytic substances, are dependent on germ expression. Current in vitro studies of endothelial activation by these diverse biochemical and bJ.omechanicalstimuli hopefully will lead to new insights into the role of endothelial dysfunction in the generation of athereeclamtic lesions, in vivo. References: 1) GlmbreneMA Jr., CybulskyMI, KumeN, CollinsT, and RannickN. Vascularendofilstium:Aft integratorof pathophyalologicststimuli in atherogennsis.Annals N.Y. Acad. Sei..VOI.748:1995, t 22-132. 2) Re.snickN, GimbroneMA Jr. Hemodynamicfomes are complex regulatorsof undcthnlialgene regulation (a Ruview) FASEBJ. fin Dress,July 1995). 3) GimbreneMA Jr. Athemgunesis:CurrentConcepts,In: Schcon& Gimbrenu,Cardiovascular Patholoov:Clinical PatholoQicCorrelationsand PathooeneticMechanisms.Baltimore,MD, Williams & Wilkins, 1995.

Local elastic moduli of bovine and porcine aortic walls were measured with a pipette aepiration technique. Histological observation was also performed to examine the relationship between the local mechanical properties and wall structures. Specimens were excised in three different directions (radial, circumferential and longitudinal) fi'om each aortic segment. Deformation process into the pipette was monitored through a video mioroseope. Local elastic modulus was calculated by comparing the slopes of the pre~ure-defonamtion curve at zero aspiration pressure obtained from the experiment with that of finite element analysis. Bvins aortae elastic moduli in both the circumferential and longitudinal directions decreased significantly from luminal to adventitious sides with a linear relationship. In contrast, the modull of porcine aortae were independent on the locations in the wall. Histological observations suggested that the ratio of collagen fiber to elastic fiber and smooth muscle cell in bovine aortae was higher in sub-intimal region than in the mid-walL The nonuniform distribution of elastic moduil observed in bovine aortae might be attributable to this histological nonuniformity. In case of porcine aortae, such histological nonunlformity was not observed. Supported in part by the Ministry of Education, Japau (Nce. 05221101and 06213101). We also thank Messrs. Mroaobu Abe and Takaya Uno.

114

117

3-DIMENSIONALVELOCITY~ S AND SHEAR STRESS DISTRIBUTIONSIN THE LOCALIZATIONOF INTIMALTHICKENINGIN EXPERIMENTALARTERIO-VENOUS FISTULAS. Lucio Florez, James McKiusey, Denise Polacek, and Nalacha DePaola. Rensselaer Pulythechine Institute,Trey, NY12180, and Univet~ty of Chicago, IL60637.

MECHANICAL PROPERTIES OF ACTIN CYTOSKELETON ENDOTHELIAL CELLS R.L. Satcher, Jr., C. Forbes Dewey, Jr. Massachusetts Institute of Technology, Cambridge, M A 02139, USA

"Fnelocalizationof athernsclerelic lesionscorrelates with the presence of alteredflow patterns in the arterial circulation.Detailed numerical simulations of the flow field in experimental aneriovenons fistulas were performed to identify the specific flow chaructea'isticsthat may be associated withlesiondevelopment.The three-dimcosioanlNavier-Stokesequationswere solved using the spectral element code NEKTON. Genmetncal features measured from 15 arterioveaeas fistulas created in normechulesterelemic (NC) and hypercholest~olemic0tC) rabbits, were used to build the l'mite element mesh. The geometry is based on the hamodynsmically-ndapted vessel dimensions (2, 4, 6, and 8 weeks after fistula creation). Velocitiesmeasured in the afferent arte~ (Doppler flowmeter)ere used as boundary conditions in the numerical model. Blood flow is considered Newtonian and steady, and Ihe vessel walls rigid. After 8 weeks, hemodynamic evalustius of the fistulas revealed a 5 fold increase in the Reynolds nombur(Re=UD/v)in the aft=cut artery and a l0 fold increase in the effel~t veins limb. Velocityfields and ~eer slreasdisuibuti~ in the vessel sut~..es have bee~ olXaiued for ftsmla models with a Reynoldseumbur range from 150 to 300 in the afferent arte~ and 18 to 70 in the efferent vein. Resultsdemonstrated the eccarrenco of well defined flow recircelafion in the dorsal, ventral and medial walls of the vein, and in the distal wall of the artery in the fistula. Areas of flow staguatian (with no renireniaflon)associated withhigh spatialshear stress gradients were predicted in the ftstula on the distal watt of the vein and on the medial wall of the artery, intimal thickeningin the NC rabbits and intimal thickeningwith foam cell lesions in the HC rabbits coincide with the predicted areas of recircdiation,low surface shear stresses, and high shear stress gradients. These data indicate that altered sheer forces play an important role in intimalthickeningand foam cell lesionassociated with atburogenesis.

We used three-dimensional electron microscopy to study the adjustments which occur in the ultrastructttre of bovine aortic endothefial cells as they become aligned with the flow direction by the aetiou of fluid shear stress. Laminar shear stresses of 10-12 dynes/cm2 are applied for 36 hours ~, and the cell membranes are extracted 2 leaving an insoluble cell fraction which includes eytoskeletal aetin and intermediate filameuts. By rotary shadowing with metal, a replica is created which preserves the 3-dimensional arrangement of the eytoskeletou. It is viewed on a tilting stage in order to obtain stereoscopic photographs. A cortical F-actin network is visible in both aligned and nonoriented cells which is not observed with light micro3copie techniques. Previous investigators have assumed that the cytoskeleton helps endothelial celIs to withstand lumenal shearing forces based on the appearance o f prondnent aligned 'stress fibers' in oriented cells. The cortical actin network was not considered. Modelling o f cortical network mechanical properties for small deformations agrees with experimental measurements of cell elastic moduli reported in the literature. This implies that the F-actin cortical network bears surface shearing forces. The creation of 'stress fibers' should reinforce against deformations caused by shear stress. [1] Dewey, C.F. et. al. (1981), J. Blomech. Eng., 103: 177-185. [2] Hartwig, I.H. and DeSisto, M. (1991), J. Cell Biol., 97: 407-419.

115

118

HEMODYNAMIC FORCESARE COMPLEX REGULATORSOF ENDOTHELIALGENE EXPRESSION. IResniek N,, IKhaehigianL., INagel T., lAnderson KR. lAtkinson WA. ICollins T., 2Dewey FC. Jr. and 1GimbroneMA. Jr. IDept. of Pathology. Vascular Research Division, Brigham and Women's Hospital,Boston MA and 2Labumtory of Fhiid Mechanics, MIT, CambridgeMA.

A MECHANISMFOR HETEROGENEOUSENDOTHELIALRESPONSES TO FLOW IN VIVO AND IN VITRO Peter F. Davies,Trever Mundel,and Kenneth A. Barhee Depur~cots of Pathologyand Mathematics,~ Unive~ty of Chicago, USA

Vascular endothelialcells, by virtueof their uniqueanatomicalpnsition,are constantly exposed re fluid mechanical forces generated by the flowingblood. These forc~ have been implicated, both in-vivoand in-vitro,in affectingendothelialcell structure and function. Certainof these effects appear to result from transcriptinnalregulationof endothelialgenes by biomechanical forces. The ann-aniferm pattern and timekineticsof the response of different endothelialgenes to the same bemndynamic forces suggestscomplex mechanismsof transcriptionalregulation. We previouslyhave identifieda "ShearStress Response Element" (SSRE}in the PDGF-Bchain promoter which appears to be necessary and sufficientfor gene inductionby laminarsheer stress, Recently, we have shown that p65 and p50 members of the NFIcBtranscriptionfactor family, can bind to the SSREand activatetranscription. However, the complexityof transcriptional regulationof endothelialgenes by laminarshear stress is illustratedby the followingobservations: 1)More than one transcriptionfactor is capable of bindingto the SSRE; 2) Genes whichcontain NFkB consensus sequences in theirpromoters are mK.uecussmily induced by shear stress; 3) Transcriptionalregulationof a given guue (e,g., ICAM-1) by laminar shear stress occurs throughthe SSRE, as well as, additionalelements; and 4) Other SSRE's can he definedin genes that are shear stress responsive(e.g., PDGF-A) but do not encode the PDGF-B/SSRE.Better understandingof the mechanismsthrough which hamodynaminferces regulate endothelialgenes will 13r valuable,as these forces have been shown to be important stimuli for endothelialcells in physiologicand pathophysioingicconditions.

Expostns of endothulinmto a nominallyuniformflow field in ~ivo and in ~itro fiequentiy results in a heterogeneous distributionof individualcell responses. Eauemes in response levelsme often noted in neighboringcell~ Such variationsare imporJantfor the spatial intertretationof vascularresponses to flowand for an understandingof mechanotransdanfion mechanisms at the levelof singlecells. We propose that vm'iationsof local forces defined by the cef ~'face geometrycontributeto these differences. Atomicfore microscopy menserements of cell smface topographyin livingendotbeliumboth in vitro and in situ combinedwithcomputationalfluiddynamicsdemonstrated large enll-to-cellva~infionsin the dism'bctienof flow-generatedshear ~ at the endothelialluminalsurface. The distribaticoof forces throughoutthe surfaceof individualcells of the monelayerwas also foetal to vat/cousiderablyami to be definedby the surfacegeometr/. We ceeclnde thattbe endothelial3-dimensionulurn'facegeomeWdefinesthe detaileddistributionof shesr slr~es and gradientsat the singlecelllevel,and that there are huge variationsin force magnitude and distributionbetween neighboring cells.The mcaserements suppe~ a topographicbasisfor differential endothefial responses to flow observed in vivo and in vitro. Includedin these studies ale the first preliminarymeasurements of the livingendothelialcell smface in an intact Briery,an approach that shouldallowdetailedanalysisof endothelialtopogaphiesin differentregionslflow fieldsin vivo. Supportedby NHLBL

OF

Electrophysiology 119

122

MULTIPLE-SITE OPTICAL RECORDING OF TRANSMEMBRANE VOLTAGE (MSORTV) IN DESIGNER-BUILT 2-DIMENSIONAL HEARTS: ACTION POTENTIAL PROPAGATION CHARACTERISTICS MONITORED ON A SUBCELLULAR SCALE WITH MICROSECOND TIME RESOLUTION. B,M. Salzburg and S. Rohr Departments of Neuroscietwe and Physiology, University of P ~ l v a n i n Seheol of Medicine, Philadelphia, PA 19104-6074 and the Marine Biological Laboratory, Woods Hole, MA 02543.

OPTICAL ANALYSIS OF MICROVASCULAR M E M B R A N E P O T E N T I A L CHANGES USING A SPECTRAL-SHIFT FLUORESCENT DYE Beach, L M . , Xia, J., and Duling, B.R. Departments o f Biomedical Engineering and Molecular Physiology and Biological Physics, University o f Virginia Health Sciences Center, Charlottesville, V A 22901

Impulse propagation acro~ abrupt expansions of excitabletissuemay exhibita vafieW of sondattian d i m an u macroscopic salt ranging from small delays to unidirectional or complete conduction block Nevur.4tbo-le~ in the normal hea.q, corrent-to-lond mismatches such as those found at the Purkinjefiber-vcotricuinrjunction, stillpermit smooth propagation of the action potential despite the unfavorable geometry. In an effort to understand normal and pathological conduction patterns in excitable tisme, we have used photolithographically patterned cultures of vcotricular myocytos and a system for multiple site optical recording of transmembrane voltage to examine the effects of geometry per se on impulse propagation. Jj$1ng the voltsge-senaltivo melocular probe dl-8-ANEPPS, together with a fast data a~lUialtion sy~em and a XI00 objective, w~ co~d resolve impulse propagation on a microscopic scale (15 ~n) with high temporal rtsolmion (uncertainty of +5 Vs). While action potential propagation was smooth in the case of fuaneled expaasinns, delays of variable magnitude ocoumd during propagation into ~'tangular or incised expansions. Distortions in the shape of the cm'diae action potential could be explained on the basis of bidirectional electrotoulc interactions across the transition region, and changes in conduction velocity could be andaratood in terms of the micro-amhitectt~ of the tissue. Supported by USFHS grant NSI6824 (BMS) and Swiss National Science Foundation grant 823A-028424 (SR).

S-27

A new fluorescenon method based on measurement o f emission spectral shifts from the voltage-sensitive dye di-8-ANEPPS has been developed for studies o f membrane potential changes m microvessds. We show that in situ d~'r pcrfusion o f 20-40 pm arterioles results in selective label o f vascular endotheltal cells. Fluorescence spectra imaged with a prism spectrograph and camera mounted on a mtcroseopr showed that dye emission from arterioles shifted toward shorter and longer wavelengths respectively, when vessels were either depolarized by potassmm or hyper.polanzod by acetylcholme. Tem.P?ral patterns o f simultaneous change in vessel dtametar and in membrane potcnttal obtained from imaged wal emission spectra show the electrical change in endothelial cells in situ is correlated wtth change in tone. Arterioles in isolated tissue patches were also dye-labeled and membrane potential changes were monitorecl with a recording mierceleclrode. Stmullaneous with electrode recording, dye fluorescence was recorded with photomultipliers at two voltage-sensitive wavelengths, 560 nm and 620, and the ratio F620 / F560 was determined with an analog divider circuit. The change in the fluorescence ratto correlated with amplitudes o f depolarization, with a voltage sensitivity o f 9.7 % ratio ehan~e per 100 inV. Vessel motion during vasomotor responses is shown not to significantly affect the ratio. This is the first analysis o f voltageMepondant shtfls o f di-g-ANEPPS emission specfra. Supported by grants VA-91-F-5 l, HL49593 and LH 12792.

120

123

OPTICAL MAPPING OF ACTIVATION, 1L~POLARIZATIOH AND REENTRY PATTERNS IN ARTER/ALLF PEP.FUSED CANINE VENTRICLES. Guy Sainma, Mark Restivo and Bom-Rak ChoL University of Pittsburgh, School of Medicine, Departmcot of Cell Biology and Physiology, Pittsburgh, PA 15261.

PATTERN OF CARDIAC TRANSMEMBRANE POTENTIAL RESPONSES AROUND AN EXTRACELLULAR, STIMULATING POINT ELECTRODE Michel Neunlist** and Leslie T u n g * * I N S E R M U381, HSpital Civil, Strasbourg, France, and ~;Dept. of Biomedical Engineering, The Johns Hopkins University, Baltimore, MD, U S A

The right or left coronary artery of 9 dog heart was peffused with 9 KrebsRingur's solution gassed with 9S% 02 plus 5 ~ CO= truing a pulsatile pomp to control diastofic and systolic perf~on pr--~sure. The right or left vcotrienlas free wall was dissected lad major coronary vessels were sutured to maintain peffualon at - 7 nd/mia/g of the I~.nmlnlng sheet. The voltage-sm~tive dye RB421 (10 pM) was injected in the coronary to ~ the ventricle and optical action poteeti~s (APs) wure JimulUmenusly recorded from 124 sites o f the eadoem'dinm or eplcardktm with a t2x 12 phor army. A "roving" light guide was used to simultanoonsly record APt from the mldwall of the veatricular sheet. During spentaucous or paced electrical activity, APs were recorded from various regions of the p e ~ veatriclet to map spatial hetorogcoeities o f A P dumfiom, "spike and dome* APs from the epicardium, shorter duration enducasditl APs, the spread of activation and repolarbmtlen and their conduction velocities. These mapa revealed that repolarizstion spreads anlso4mpicully with dower "conduction" velocities than autivation and with orientations that am indepcodmt of activation pathways. Premature e x ~ induced veatriculas tachycasdin which were mapped on the epicardial sut'face end M cubs of the midwalL Stable reantry patterns were generated on the epictrdinm exhibiting 2-D circus m o ~ n e m or more complex breakthrough patterns from deeper cells. In contrast to s u p e r ~ e d tissue slices, maps of APs from perfused heart muscle exhibited faster conduction velocities, systewmie heterogeneities of APDs and R pattoms, and made it possible to map reentry pathways in a 3-D syncytium.

Recent theoretical models of cardiac electrical stimulation or defibrillation predict a complex spatial pattern of transmembrane potential (o m) around a stimulating electrode, resulting from the formation of virtual electroden of reversed polarity. The pattern nf membrane polarization has been attributed to t h 9 anieotropie s t r u c t u r e of the tissue, To experimentally verify nueh model predictions, an optical technique using a fluorescent voltage sennitive dye was used to map the spatial distribution of o m around a 150 B m radius extraeellular unipolar electrode, on the endocardial surface of bullfrog atrium in directionn parallel and perpendicular to the cardiac fibers. With eathodal pulsex the membrane depolarized in the longitudinal fiber direction only in a regi9 within 0.5 m m from the center of the electrode (n=9) and reversed polarity at greater distances. In contrast, the membrane depolarised in the transverse fiber direction at all distances from the electrode (9 The results were qualitatively s i m i l a r in both fiber directions when the atrium was bathed in a solution c o n t a i n i n g ionic channel blockers. A two-dimensional c o m p u t e r model was formulated for the case of highly aniantropie, passive cardiac tissue, and qualitatively accounts for nearly all of the observed spatial and temporal behavior of o m in the two fiber directions.

121

124

DETERMINATION OF LOCAL MYOCARDIAL ACTIVATION: A STATISTICAL APPROACH Kelley P. Anderson. R. Walker, P.R. Ershler. M. Fuller, T. Dustman. RL. Lux University of Pittsburgh, Pittsburgh, PA, University of Utah, Salt Lake City. UT Electrical activation sequence mapping requires accurate identification of local activation, but because extracellular recordings do not exclusively reflect local events, complex electrograms may be difficult to interpret. In such cases, the assignment of local activation is subject to error that could affect interpretation of the resulting activation maps. The purpose of this investigation was to develop an approach that would provide quantitative indexes of errur in the determination of local activation, An electrode array with 64 closely spaced u 9 electrodes was used to record from the left ventricular surface during open heart surger/. Elactrograms with multiple defleetloi~s were recorded from four patients with scarred myocardinm; two other patients with normal myocardial function served as controls. Each of 784 deflections was scored on the basis of three features: evidence fur propagation, the configuration of the bipolar signal, and the effect of changing from the chest to an average reference, Local activation was co9 probable if evideece for all three features was present and improbable if none of the three features was present. Deflections that were ambiguous with respect to this standard were excluded. Of over 30 test variables analyzed, the three with the greatest power to discriminate signals due to local activation from those due to distant activity were 1) a linear combination of the extracellular potential plus the ratio of the second derivative and the extracellular potential. 2) the second derivative, and 3) the minimum (greatest negative) first derivative. For each of these variables, the threshold value providing the greatest performance was identified by the maximum quality of efficiency, an index of agreement. This statistical approach provides aa objective basis for determining local activation and provides a quantitative assessment of error that could enhance interpretation of electrical activation sequence maps.

V O R T E X W A V E S T A B I L I T Y IN H O M O G E N E O U S E X C I T A B L E M E D I A : SIMULATIONS ON A RANDOMIZED DISCRETE LATrICE A.B. Feldmaa, Y.B. Chemyak, RJ. Cohen H a r v a r d University - Massachusetts Institute o f T e c h n o l o g y Division of Health Sciences and Technology The breakup of a single rotating vortex wave into a turbulent pattern of electrical activity is believed to underly the transition from vo9 tachycardia to ventricular fibrillation in the hes.~. While the precise mechanism of this transition is unknown, current theories generally rely on the notion of condoctin9 block al 9 regions of the vortex wavefront due to inhomogeneities in the medium. If these regions of block are large enough, new pairs of vortex waves are created that can interact with remnants of the original wave leading to turbulence. Such a process has been observed in recent cellular automaton models of the heart and in experime9 studies of otber excitable media. In the heart, the required inhomogeneities are generally the manti of disease (isabemia) or result from tegio 9 varistioas in the properties of the myocardial tissue. However, it is also possible that the regions of conduction block arise spontaneously due to the spstial pattern of excitation by the vortex, which results in turbulence in a completely homogeneous medium. Several groups have observed such a phenomenon in discrete state ceUular automaten models of homogeneous media on a square lattice. However, these ~odels ganerally suffer from the drawback that the geomet~/of the underlying lattice (squares) manifests itself at macroscopic scales (square wavefr 9149 introducing an unpbysical numerical artifact into the simulations. We studied, the stability of rotating vortex waves th 2D homogeneous excitable media using a cellular automat9 model on a randomized lattice. We performed two different sets of simulations using both square and circular regions fur the interaction 9 In both cases, the spentaneous breaking of the vo~ex waves was eliminated by randomizing the positions of the lattice nodes beyond a minimum length scale Xc. This scale was an order 9 magnitude smaller for the circular interaction region.

S-28

Poster Presentations 128

125 AN INVASION PERCOLATION MODEL OF ARCHITECTURAL OBSTACLES TO TRANSPORT IN TUMORS. J.W. Baish, Y. Gasit, D.A. Bark, M. Nozue, L.T. Baxter, and R.K. Jain, DepartmentOf Radiation Oecoicgy, M a s s s c h ~ GeneralHospital,HarvardMedical School,Boston, MA 02114. Transport of diffusiblesubstances in tissuesis limited in imrt by the distance over which substances must diffuse between the vascular space and the sunoundlng extravnscuinrtissues and by ~ icsses prior to local deliveryby the blood. By examiningthe fractulbehavior of two dimensionalvascular networks in the murine dorsal skinfoMchamberprepara~ we have identified architectural features of normal and tumor vssculat networks that lead to fundamentally different wansport behavior. Healthy capillary beds have fractul dimension 1~99"z0.01 and minlmampath dimensiondminffiO.99"~O.02, consistentwith the widely-used, h cylinder model of vesssls in a regulasly-spacod array. In contrast, tumor vascular networks have df=!.88iO.04 and dm/n=l.iO~O.04, nearly identical to invasion percolation, a wall-known growth process emphasizing random local heterogeneity. Based on these observations,we have consu~cmd a pr compulexsimulationof tumor vascular grewsh which enables us to predict the effects of vascular tortaosity and the widely-vlu'iable avescala~q~ces on transport. We EodJ~t e_~fra~lionof t~ue in avsscalarspecss largerese linonrdimensionLsealesss I - ( L I Lo ) utr~ l whece Lo is tbc uppcr boend of fractal seali~g and dti#sue is the dimensionof the embeddingtissue. This suggeststhat them will exist a few large avascular spaces and many smaller spaces between vessels. We also fled that the tortuosity of the vessels, as reflected by the elevated minimum path dimension, produces regions of locally flow-limited transport and raduees flow to the tumor as a whole. A comparison to oxygenation measurements in normal and tumor tissues shows that the percolation model predicts the architectural obstacles to transport in tumors more accurately thse the Krogh cylindermodeL This work was supported by CA-56591, CT5-9057412 and a Howard Hughes Predccto~l Fellowship to Yuval Gazit.

126

TRANSCRANIALDOFFLER SONOGRAPHY:A NEWNONINVAStV~MErHO~ FOaMEASURING TOglCANT-INDUCrDALTERATIONSIN CERBRALBLOODFLOW Michael E. Drues, t Ph.D., David L. Hopper, Ph,D. & Douglas N. Lange, D.V.M. Waseular Sciences. 11 Crescent Street, Natick, MA 01760-2506. The purpose of this experiment was to use transcranlal Doppler sonography {TCD) in the dog to noninvasively measure changcs in cerebral blood flow (CllF) resulting from a low-level exposure to the pyrethroid insecticide deltamethrln. Although the use of TCD in animals has been limited, pilot work indicated that CBF can be used in dogs and that manipulations of CllF osing CO~ can be measured. Deltamethrin has previously been shown to increase CBF in rats as determined by other techniques. In this experiment, ddtamethrin was dissolved in glycerol-formal (GF} and administered LV. to greyhound dogs. It was expected that vascular changes resulting from these ddtamethrln exposures would be subtle and occur over several minutes. Mean blood flow valociW (MSFV) in the middle cerebral artery, mean arterial blood pressure (MAP), end tidal CO~ (pCO~), mean heart rate (MHR) and concentration of deltamethrin in .the blood were measured. Systolic-to-diastolic ratio (SDR), pourcdout pulsatflily index (PPI),Gosling pulsatihty index (GPI) and systolic upstroke (SU) were calculated using the MBFV waveform. An ANOVA revealed that GF alone did not effect the physiological parameters measured. However, there was a decrease in SDR,PPI and GPl. This suggested that the shape of the MBFVwaveform changed in the presence of GF without any corresponding changes in the measured physiological parameters. Foliowing deltamethrin exposure, there was an increase in MBFV,MAP and pCOz and a de~ease in ~ The increase in MBFVcorroborates the increase in CBF observed in rats. SDR, PPI and GPI returned to control levels following deltamethrin exposure. This experiment has shown that TCD is sensitive to alterations in CBF caused by deltamethrin in the greyhound dog.

129

MONITORINGCEREBRALOXYGENATIONAND BLOODVOLUMECHANGESDURING CARDIOPULMONARYBYPASS(CPB)USINGNEAR-INFRAREDSPECTROSCOPY PautB, Be~ MS,"8o ChartMS,"DavidAmoe/MD,PhD,and JdnnK-J.U PhO. Dept.of BIomedcsiEngineering,Rutg~ University, and'Dept,of Anesthesia,RobertWoodJohnsen Medcsi School,New8nJnswlek,NewJersey08903,USA. Nelselogi~d dw|t=~!iccis om of the su~ouseom~icatic~d hypothem~cCPB due to 02 u.q~y and demandde~dt du~ngthe p ~ r e . We havedssigneda novelNIRSsystemto monRor mglonal u~tral hemoglobin u~can~ation ~m0es (A~, AHbO~, ~HI~A~) in patisuis undargdnghypothendcCPB. AHb sed AHbOzwerecaleuinledaccor~ngto a modfledB e s f - ~ Law. IOtov~ tim hemato~tlevis, d~nges in mginml ccrsixal binod~lume (ACrBV)weredndvnd tram AHb*AHbOz. 0.~ ~ r 1.5 "'"

,..

v

/ ~ J ~V-k~ < -oo2 J

I

.. . . .

~ / ~ ' - - " ~

cooito 2li'c

II R~am.in~

I

~.

9

-'% .o.s

t.ts

<

~ e figure ~hows the mspoase d s typical patlonL The sharp ~ e n t d'up d hemoglobin cor~tral~ns at the beginningd CPB is due to hemodlution. AC~V decmssasduring hypothemtia (29~C)due to the decreasede s ~ metabdicrate, and then incrensssto aboutthe pm.CPB level during rewa,'1~. A transientdop of ACrBVoccursfollowing a temporarylow pedusicc pressure dJdngthe rewsu~ngphase. Decrssseof AHb02and it~-ssss of &Hb dudnglatemwan~ingsuggests a reducedC~supplyI demandratiowhlch ccc~Jrsin sumeCPB patienis. The mst~tsshowthat our NIRSsystemla a p(oenisingmo~tor of c e ~ oxygenationdudngCPB.

QUANTITATIVESTUDYOF THE ~ OF RISK FAC'I~RS ON CORONARYARTERyATHEROS~OSIS AjcclhomarGnddipati1, J.FredfickCornhill1,2,Edward Herdr 0 The PathobiologicalDctcnninsutsof Athereselerosisin Youth(PDAY)Rcsustch Group IBiomadicalEaginccringCenter, The Ohio State University 2Dept. of BiomedicalEngineering.Research Institute.The ClevelandClinicFoundation Atherosclerosis is the thickcaingof the tuner anesial wall caused by accumulationof lipids, cellularproliferationand migration,and the synthesis of exWacclinlarmatrix which results in luminalnarrowing. In advanced cases, atiu~osclerosiscan lead to myecm'dialinfarctionand st.,oke. "fherisk of aderosclevosiscan be ~uced by attenuating factorscontributingto this process. A study of risk factors effccfingatbetosclemsishas been conducted by Laborato~ of Vasculm" Diseasesin enllabeeationvlith11other institutions.Quantitativerelationshipsbetweenrisk factors and atherosclefosiswere obtainedas a p ~ of this study. Lipidacc~nuintionin ~rteflalwalls,one of the major conuibutiogfactors to athcrosclerosis,has been qr investigated.Left AnteriorDescending Coronary ~,tery histologicalcross sectionstak~ just distal to the origin of left circumflexcorenszy efte~ywere ataiccd for lipid. The slideswere segmented intoregionsof inera-celluinrlipid ([CL), extra-cellularlipid (EC'L)and normaltisst~ usingan automateddassitierdeveloped by fae authors.Following classification,conmmry~ imagesweregenmelrically transformed into a standard circularcross section.Each image was then farth~ divided into 36 equal radialsegments to study the distributionof lipidin athe~sclefodc lesions. The amount of lipidwas found to increasesignificantiywith ~acre~e in thicknessof lesion.It was found that the layer of intima neascst to inn~n bad lesslipidcompm~l to middle and outer layers of the in~ma. Lipid sccumalatiun increased sliginlywith age, Smokers had higber lipid accumulation than nonsmokers. Persons with high levels of LDL in filch"blood had more lipid accumulation in theirarterialwalls.ECL content was significantiygreaterthan ICL contest.

127

130

MORPHOLOGICAL AND PROLIFERATIVE RESPONSES OF VASCULAR SMOOTH MUSCLE CELLS TO SUSTAINED HYDROSTATIC PRESSURE L. A. Detr~s, P. J. Del Vecchio*, and R. Bizies Dept. of BiomedicalEngineering, Renssalaer Polytechnic Institute, Troy, NY 12180 *Dept. of Physiologyand Dept. of Ceil Biology,Albany Medical College,Albany, NY 12208

FLUORESCENCE RATIO IMAGING IN THICK TISSUES: APPLICATION TO THE NONINVASIVE DETERMINATION OF TUMOR pH IN VIVO Dellian M., Helmlinger G., YUSUF., and Jaln R.K., Department of Radiation Oncology, Massachusetts General Hospital,Harvard Medical School, Boston, MA 02114, USA

The effects of sustained hydrostatic pressure on the morphology and proliferation of vascular smooth muscle cells were investigated /n vitro. Bovine pulmonary artery smooth muscle cells were ianlated using the expinnt method (Clin. Sei.. 85: 501-513). Smooth muscle cells of passage 1-10 were seeded onto rigid glass substrates, placed in sterile pressure chambers where hydrostatic pressure heads (2.5 - 15 cm H20) were established under appropriate volumes of Dulbecco's modified Eagles medium (containing 10% calf serum), and maintained in a 37~ humidified 5% C02/95% air environment for up to seven days; controls were cells exposed to 0.3 cm H20 for the same time periods under standard cell culture conditions. Ligh~ microscopy wan used for the analysis of smooth muscle cell growth patterns, while fluorescent microscopy was used for the assessment of cellular morphology. Compared to control cells, smooth muscle cells exposed to hydr~tatic pressure exhibited eytoskeletal rearrangement and increased proliferation in a pressure-dspendent manner. Those results provide evidence that cellular models could be used successfully in investigatingthe effect~ of mechanical forces on smooth muscle cell functions that have important implications in vascular pathophysiology.

Ruorescence ratio imaging(FRI) has been used successfullyto measure dynamicsignaling events (e.g., Ca2+, Nu§ H+, membrane potential) in living cells, both in ~itro and in aitu. To apply FRI noninvasivelyin an ~ vlvo setting, a suitable tissue preparation, an appo)prime mute foe loadingof the fluorescentlX'Obe0and a method for quantificationinsensitiveto light absorption of hemoglobinand te light scattering of thick ~ . axe required. Our goal was to measure locally the interstitialpH of haman tumor xea~grefts (l.,S174'r) transplanted in the dmssl skinfoldchamber of the severe combined immuoodeficiont(SCID) mouse. To this end, we used the pH-sensitive probe BCECF, and applied FRI combined with a pinhole illumination,object scanning acquisitionscheme and an ex vivo calibrationmethod. Intenfifi~ tumor pH was 6.85:1:0.07(menn~SEM, Nffi6),while pH of normal subcutaneous tissue was 7.37:1:0.09(Nffi6).Imagingof diffe~P.,nttumor regionsrevealed spatial b e t ~ t y in pH, and steep gradientscould be foundbetween tumorbloodv ~ l s . Tcmpccal changes were measmcd during induction of hyperglycemia (6 g/kg i.v.), which selectively decreased tumor pH to 6.61:L-'0.15at 100 mln followinginjection (P<0.05). These data indicate that FRI is a unique approach which allows non-invasive,prelonged observation of microscopic changes - both spatially and tempm'ally- in tumor and normal tissue pH. Combinedwith other ion-selective probes and optical techniques, it will permit the investigation of tumor signaling and metabolism in an in vivo setting. (Suppmted by NIH R35-CA-56591; M. Dellisu is a recipientors Feodor LymmFellowshipby the HumboldtFoundation 1993-1995)

Poster Presentations

S-29

131

133

THE TRANSITIONIN MECHANICALPROPERTIESFROM CHORDAETENDINEAETO MITRAL LEAFLET K. May-Newraan and F. C. P. Yin Johns Hopkins Medical Institutions,Depts. of Medicineand BIDE.Baltimore,MD.

EVIDENCE FOR THE FUNCTIONOF THE "SHEAR STRESS RESPONSE ELEMENT' IN TRANSCRIPTIONAL ACTIVATION OF ICAM-I IN E N D O T H E L I A L CELLS EXPOSED TO LAMINARSHEARSTRESS. T. Nagel, N. Resniek, IC.F. Dewey, Jr., ~,I.E. Gerritsen,M.A. Gimbrone,Jr. Vascular Research Division, Departments of Pathology, Brigham & Women's Hospital, and Harvard Medical School, Boston,IdA 02115; tFluid Mechanics Laborato~, MIT, Cambridge, MA 02139; and =Bayer,W~t Haven, CT 06516.

Chordae tendineae arisingfrom the miwalvalveleafletmay be a si~ of local stress concentrations in the deforming valve since these sites are involved in the majority of chronic destrocfive processes. The cleat" structural transition from leaflet to chordae suggests a transition in mechanical properties, which mustbe considered in any stress analysisof the valve. Our aim was to quantify the regionalvariationsin mechanical lxoperties in the chordee-lceflet transitionzone. Re~tanguinrsectionsof Ixxcineantexiurmitral valveleafletscontaininga large strut cherdae were isolated. 15-20markers extendingfrom leafletbody to cherdes were placed 1-2 mm apa~ in a grid pattern on the veetricularside (A). The specimenwas tethered to a biaxialsWetehingapparatus by suturingthree sides along their lengthand wing the chordns to the fourthcon~age, The spenimen was stretched along the axial directionwhile holdingthe transverse length constant at 0, 5 and 10% stretch. The marker field was recorded on videotapeand digitized offliee. Groupsof four markers were analyzed and an isoparametric interpolationmethod used to obtain Lagrangian sWain.The cross-sectional area for each quadrilateralwas determined from regionalthicknessand width measurements,and Cauchy stress calculated. All regions from chordae to leaflet exhibit nonlinear, large deformation mechanical behavior (B). Extenaibilitywas tabulated at a stress value of 25 kPa and is shown as a functionof normalizedannuiur--chordne distance, NAD (r A distance of 0% correspondsto the annulo-valvularjunction, and 100% to the separation point of the chordac from the leafletbody. A decrease in extensibilityof 6%/ram was observedover 100% of NAD, but subsequently reached a steady value in the chordae. From these results there is evidence that a steep transition in ..tJ.U.~_l ~.eo~ . t4*" I mechaalcalpropaniesoccursprier tothe :[:'~:":~ ~ i ,~|_~~ [ 1 f " " 1 ~ chordna separation point.The variationis ~:.~~-:f:~ ~ -Ic~*~,,t'~t~,e~ ~ 1 ~ - ~ , _ ~ . ~ _ likely related. to. collagna / I ~ ~/w / , . ~ , "~o~ . . fiber orientation l:l: ~::/ :1:11~ o~ o I ~~ , ~* and tortuostty, wh|ch exhthlt . . . . p l e x I % ~ / l l ..... . . . . m-rangement in this tissue structure. IA" "11' " [ [s Sn~ [ C m~ot~l

/I

Hemodynamic shear stress induces various functionalchanges in vascularendothelinm, many of which reflect alterations in geae expression. We have previously identifieda transcriptional regulatory element, the "shear stress response element" (SSRE), which is present in the promoters of several shear stress-responsive genes and may represent a common pathway by which biomochanicalforces influencegnae expression. The SSRE was shown to be mqu~d for the transcriptionalupregulationof plamlet~lerivedgrowth factor (PDGF-B), and I~eat work in our laberatoW has established that nuclear faetor-r,.B fNF-r,.B) binds the SSRE within the PDGF-Bpromoter and may participatein the inductionof this geue by laminar shear stress (Khaehi~an et al, J Clin Invest, 1995, in press). In the present study, we have examined whether the SSRE is functional in the induction of another endothelial expressed gene, Intercellular AdhesiOn Molecule-I (ICAM-I), which we have previously demonstreted is upregulated by laminar shear stress at the level of beth steady state treaseript and coil s~face protein. An ICAM-I promoter construct, containing the SSRE (GAGACC) as well as several consensus NF-r.B sites, coupled to the Iocifmese repor~ gene. exhihited transcriptional activation when transfueted into bovine aortic endothelial cells which wc~ then exposed for 3 hours to laminar shear stress (10 dynes/cm~). This shear stress activation was ~laced, but not completely abolished, in a promoter construct lacking the SSRE but retaining the NF-KB sites. Thus, these data implicate the SSRE in the transcriptional upregulation of ICAM-I by laminar shear stress and suggest that it may be functioningin conjunction with additional regulatuW elements yet to be defined.

132

134

MARIAN: A DATA VISUALIZATIONTOOL FOR TItE ANALYSISOF CARDIACFLOW MEASUREDBY VELOCITYENCODINGMRI Sylvaln F. Miler, Peter G. Walker, *Kim Hcolind, *W. Yong Kim, *Erik M. Pedetr~n, Ajit P. Yoganathan - Georgia Institute of Technology, Atlanta, GA, USA and *Aarhns UniversiW Hospital, Aarhus, Denmm'k

DEVELOPMENT OF A REALISTIC ARTERIAL FLOW SYSTEM FOR MRI VELOCITY VALIDATION Meera S. N a i l B.E., Stanley Rittgers, Ph.D., Jeffrey Duerk*, Ph.D. Department of Biomedical Engg., The University of Akron, Akron. *Department of Red. & Biomed. Engg., CWRU and University Hospitals, Cleveland.

Velocity encodingMRI is a new non-invasivetechnique for measuring three-dimeasinnalblood flow velocities. However, the large amotmt of data acquired with this method requires specialized analysis softwure. For"this pmpese we developed a data visualization tool named Magnetic Resonance Imaging Analyzer (MARIAN). Methods: MARIAN contains vilumliT~rion,animr|o~ and c~mpnlattonalfools fur the analysis of caldiac VetOCityeucod~g MRI dot& Tha toois ere acnessed through e graphies user interfane (GUI, Rg. I) besed on graphins widget.sand developedee a Silic~ ~ e s Onyx system. This software was used fer the analysis of blood flow patterns in the let~atrium fer two groups of human subjects: 10 nurmal volenter~ and 8 patients. The patiests wet'e all diagnosed with prnvious left ventrinular myocardial ~ . With MARIAN,the ttmperal velocityprofdes wca'e sampled at the tip of the mitral leaflets and at me ~lmonary vein. The isovolumic relaxation time (IVRT) was estimated. The E ned A waven, end the ecceleratien and deceleratinn times were mcasemJ en the miwal profile. The S, D and R waves were measured en the pulmoemy venous pmrfle. The difference in averaged values were compared between the 2 groups (t-test, I)<0.05). Results: The nssultsobtsined with our analysis method wm'e ~nsi.stent with previouslypublished data. The patient group had significuetiyhigher IVRT, lower and later E-wave, lower D-wave, whea compm~ to the nmmal group. Conclusions: MARIANprovided fast, convenient and reliable tools for analyzing cardiac MRI dam. The flow patterns in the patient greep were characteristicof reduced leB vea~cuiar fillingin early diastole.

!,t!}=,,:...

The purpose of this study was to develop an in-vitro flow system capable of producing a known, repeatable physiologic waveform to evaluate Magnetic Resonance Imaging (MILl) flow quantification and error-compeesatinn techniques. A programmable stepper motor controlled flow generator was used to generate reproducible flow waveforms in compliant tube models (I.D. : 20 mm, wall thickness : 1mm) of the deaconding abdominal aorta. A mixture of Dextran and distilled water with a dynamic viscosity of 3.5 cP was used as the blood analog fluid. A dual crystal, l 0 MHz ultrasound transducer was fabricated for making 2-D velocity measurements. Velocities were mea.~ured at 11 radial positions at each of the 5 axial positions under steady (Re=600, 90~, 1200) and pulsatile flow (Re=560) conditions. For steady flow, at three and a half tube diameters, the velocity profile was nearly parabolic (r = 0.99, Re = 6(X)). For pulsatile flow, a close correspondence was also observed between the mean velocity profile and the theoretical profile derived from the flew'meter recordings (r = 0.95), with the radial velocity component being greater towards the wall. A physiologic model with a realistic flow waveform has been developed in a flexible system capable of simulating different test conditions and with 2-D velocity collection capabilities. This will provide a comparison for MILl flow quantification and compensation methods.

132a

135

ENDOTIIELIAL-MESENCItYMAL TRANSFORMATION IN ATHEROGENESIS: A RECAPITULATION OF EMBRYONIC HEART CUSHION TISSUE MORPHOGENESIS V. A. Mironov, C. H. Hill**, B. Starcher**, J. C. H. Shih**, S. Hoffman*, R. R. Markwald. Dept. of Cell Biol. & Anat., *Dept of Medicine, Med. Univ. of South Carolina, Charleston, SC 29425; **Dept. of Poultry Sciences, North Carolina State University, Raleigh, NC 27695.

ON THE VASOMOTION OF THE CONDUIT ARTERIES O F THE HUMAN LIMBS: AN IN V I V O STUDY C.-A. Porter, N. Stergiopuina,D. Hayoz*, H. R. Broonur*, J.-J,Muistur Biomedical EngineeringLaberatesy, Swiss Federal lestitute of Techn~ogy, and *Division d'Hypettensinn, Cent~ HnapitslinrUuiversitsiesVandois, Lausanne, Switzerland

Endothelial-mesenchymal transformation (EMT) occurs during cushion tissue morphogenesis in the embryonic heart development. Gradually downregulation of N-CAM and endothelial specific QHI antigen during EMT in vice and in vitro was accompanied with the expression mesenchymal markers such as a-smooth muscle actin, proeollagen type 1 and tenascin and increased gelatinolytic activity. Is the EMT during heart cushion tissue morphogenesis a unique phenomenon or can this process be recapitulated in vascular endothelium during atherosclerotic intimal thickening? To answer on this question we investigated aortas from male Japanese quail with susceptibility to cholesterol-induced atherosclerosis with atherosclerotic lesions using endothelial specific QHI mab. The presence of subendothelial QHI positive and QHI negative cells in intimal thickening was demonstrated. QHI negative cells may have a nonendothelial origin or they may lose completely their endothelial markers during EMT. These data give direct evidence for EMT during morphogenesis of intimal thickening.

We investigatedthe patterns of vasomotlonin various conduit arteries of the human a~a. The internal diameter of the brachial, radial,ulnar, and digital trtetles was measured nouinvasivelyin healthy volunteers,usinga high .wecisionultrasonic echo-trackin 8 device. Under restin 8 conditions, the radial, alnar, and digital internal diameter exhibited spontaneous oscillations(vasomo~on) with a relativeamplituden u ~ from I to S percent of mean diameter and a fundamental frequency ranging from 0.01 to 0.05 Hz. The lowfrequency mode (f< 0.05Hz) poesem in the diameter sigml was identified neithes in the heart rote nor in the blcod ~ signal. In e~lur to ~ whether the nscigato~y activity was propagative, simultaneous measurements of diamet~ at two dtna on the right redial aneff wece performed and reveskd no significaatconsistoat phase.drift, llndlatenl radial and ulnar diameters, mnasnsed at the wrist level, exhibited ~ and synchronous For all subjects, conmdatersi ~ , performed at two vesomotion ~ symmenical siteg of the radial atte~es, showed similar nacillatm7 patteam.swith a seroag correlation(0.85 < r < 0.99, n = 12).These n~ults suggested the existenceof a global regulatorymechanism which coordinatesvaanmotion in the large conduit arteriesof the huma~

S-30

Poster Presentations

136

139 A ROBUSTNONINVASIVE ESTIMATIONOF CARDIACOUTPUT

AN IN-VITRO STUDY OF THE IMPACT OF CARDIAC OUTPUT, TEMPERATURE, AND L E F r VENTRICULAR dP/dt ON MICROBUBBLES FORMATION ASSOCIATED WITH MITRAL MECHANICAL HEART VALVE (]~1~.. CLOSURE Edmond Rambod. George .~ Gunt, Den[.~J. Le~, $imcha MIIo , Mory Gharlb California Institute of Technology, Pasadena, CA, Baxter Healthcare Corporatiun-CVS Divisions, Rambam Medical Center and the Techninn-Schonl of Medinine, Haifa, ISRAEL~ Purpose: Spontaneous Echocardingraphic Conla~st (SEC) is a well known, although poorly understood phenomenon associated mainly with MHVs in the mitral position. It has occasionally been observed as bright high velocity echoes passing through the milral valve into the left ventricle (LV). Recent reports have described them as microbubbles and have indicated that the intensities of these gas bubbles in elderly patients are less than those in younger patients. These facts may indicate a correlation between formation of the gas bubbles and the LV functions. Our goal wns tn utitise the new pulse duplication system to create the corresponding physiologic conditions and to understand the impact of these parameters on SEC formation. Metheds: Our new system with a flexible LV model simulated physiological conditions, (LAP)~,~ffig, (LVP)p,~,ffiI30, (dP/dt~.~lg00-3000, (SP)=m-100, HR=tflbpm, CO=3-5Ftoin, SR=35%. A 27ram Edwards-Duromedics valve was used in the mflral position. For the experiments, the lio~id was de-ionizod, de-bubbled. The liquid temperature was carefully monitored and gradually increased from the room temperature to 37~ M-Mode Doppler and 2D-echo of long-axis and short-axis planes were performed using a Vingmed CFM7JO machine. An Argon-Ion laser beam was used for planar illumination and high-spend video flow imaging. Results: High-speed videography of the milral valve depicted little bright bubbles formed in the immediate vicinity of mechanical prosthesis upon closure at CO=3Fmin. As the CO was incrensed, we observed an increase in their intensities and a significant difference in their nature. The effect of dP/dt was even more substantial. At dP/d~3000, we observed a major increase in the size of these bubbles and in their life-span. Increasing temperature from 22*(2 to 37"C caused an immediate increase in the bubbles intensities end size. Conclusion: CO, LV dP/dt, and Temperature have major impact on SEC formation in MHV.

137 AN IN-VITRO STUDY OF THE IMPACT OF MECHANICAL HEART VALVE (MHV) CLOSURE ON THE RELEASE OF BUBBLES IN GAS-SATURATED SOLUTIONS Edmond Rambod, George 3(. Gunf, Denis J. Levy, Shncha MiloW, Mary Gharlb California Institute of Technology, Pasadena, CA, Baxter Healthcare Corporation-CVS Division s, Ramhnm Medical Center and the Techninn-Schodi of Medicine, Haifa, ISRAELe Purpose: Recent clinical reports have indicated eehocardiographie observations of bright mobile echoes (Spontaneous Echocurdiograpbic Contrast-SEC) in the left ventricle (LV) and atrium (LA) during post--operetive studies in patients with MHV. Depending on their Doppler spoettum signal intensities, some of these echoes have been categorized as gaseous emboli (GE). Our goal was to utilize the best innovations in pulse duplication system design combined with ultrasound and laser video flow imaging methods to fundamentally investigate and understand the release of gas content and formation of bubbles associated with a closure mechanism. Methods: Our system with a flexible LV model simulated physiological conditions, CO=51/min, (LAP)t~=8, (LVP)~ffiI30, (dP/dt)Lvffilg00, (SP)~m=100, HR=60bpm, SRffi35%,. Mitral and aortic valves were: a 27rnm bilenflet mechanical and a 23ram tissue, respectively. M-Mode Doppler and 2D-eehocardingraphy of inng-axis and short-axis planes ware performed using a Vingmed CFMT$O machine. An Argon-Ion laser beam was used for video flow imaging. Results: High-speed" videography of the miffal valve from the atrial side indicated bright bubbles formed in the immediate vicinity of the mitral valve upon valve Olosere, lraveling in the LA before passing through the mitral valve during diastole to the LV. M-Mode, color M-Mode and spectral Doppler ultrasound images best detected these gas bubbles and their intensities. Conelusiuns: Our in-vitro observations left no doubts that the dissolved gas in a saturated solution can be extracted from the solution at the valve closure. Such a closure process creates squeezing flow between the leaflet and the housing followed by a sU'ong deceleration of the leaflet and sudden pressure drop which generates vortical slructores on the atrial side. The large gas concentration gradient then promotes the rapid process of "rectified diffusion" of the dissolved gas from the narrowing gap to the low pressure vortex core which ultimately, results in formation of these gas bubbles on the atrial side.

Judith D. Redling, Metin Akay, W a i S t We|kowltl, Alan J. Spotnits Dept. of Biomedical Engineering, Rutgere Ualvereity Dept. of Surgery, Robert Wood Johnson Medical School, UMDNJ A robust method for calculating cardiac output noninvaeivaly and automatically has been developed. The achievement of a simple-to-use, automatic measurement system is imperative, given the clinical demand for a method of nonlnvnsive flow estimation acceptable over the standanl procedure of therraodilutinn. The method relies on fast Fourier transform (FFT)anMysis of pulsee measured externally at the cexotid and femoral pressure points. A transfer function of the aorta is computed from digitally filtered pulse measurements, and a tapered model of the aorta is parewnatrically adapted using a simplex optimization algorithm eo that its transfer function matches that derived experimentally. An aortic input imped~ce term in obtainod from the optimized model and utilized along with the c ~ o t l d pulse (analogans to input.voltage) to compute aortic flow. The method in robust in that it is automatic, requires relatively few pulses fur analysis, and exhibits considerable resistance to noise artefact. For 100 data records collected from 82 cardiac surgery patients~ the average flow measurements computed over several pulses compare well with the standard invasive method of thermodiintinn. In addition, aortic parametere (aortic taper~ input area and hoop elasticity) derived from the optimized model and computed for 82 patients were all within physiological ranges.

140 ANALYSIS OF EQUIVALENT FORWARD FLOW RATE FOR DURABILITY TESTING OF BIOPROSTHETIC HEART VALVES Romaro C. M., Guo G. X.. Kingsbury C. Baxter Hcelthca~, Edwards CVS Division, Irvioe. CA. In the FDA's gnidaocr documcntatiou, in vitro accelerated wear testing (AWT) has been required to evaluate the long term durability of prosthetic heart valves. In addition to the specification of back pressme load, complete opening, which has not yet been defined, is also required for each valve. Since most of current tissue valve designs show flow rate dependent opening, a larger opening degree due to a higher flow rate may create worse wear doe to leaflet contact with the fabric covered stunt. During AWT, therefore, quantifying the forward flow rate for complete opening of the valve is as important as quantifying the back pressure loud. The following experimental method has been developed to ensure complete valve opening in AWl': 1) Prior to any AWl" study, test aortic Waives at a 30 liters per minute (lpm) and mitral valves at 25 Ipm in a steady forward flow test (SFFr) system. Using high speed video OISV), images of valve opening ( I , ) can be captured from both the inflow and outflow sides. 2) The AWl" study should be initiated by adjusting the valves as per test protocol. Use printed I., to gauge for the complete opening of the Waiveand then adjust the peak differential pressure during valve closure ( & P , ) so that it is at least 90 mmHg for the nutrias or 120 mmHg for the mila'nis. Captured HSV images showing valve maximum opening and closing can be seen from either the inflow or outflow side once the tester is adjusted. 3) Compare the valve opening shown on images printed from AWT ( 1 . ) to those printed for IN, If the valve opening area shown on I . is comparable or larger to the area shown on I., the forward flow rates in AWl" (F.t) can be considered equivalent to their corresponding 30 or 25 lpm SFFT flow rates (F.).

138

141

GASEOUS EMBOLI FORMATION AND PROSTHETIC MITRAL VALVE CLOSURE: AN IN-VITRO COMPARISON BETWEEN TWO MECHANICAL BILEAFLET MITRAL VALVES AND A BIOPROSTHETIC MITRAL VALVE Edmond Rambod, George X. Guoi, Denis J. Levy, Slmcha Milan, Mary Gharlb California Institute of Technology, Pasadena, CA, Baxter Healthcare Corporatinn-CVS Division s, Rambam Medical Center and the Technion-School of Medicine, Haifa, ISRAEL*

NONINVASIVEMEASUREMENT OFF~IDOTHELIUM-DEPENDENT VASOREACTIVITY:

Purpose: Spontaneous Echocardingraphic Contrast (SEC) is a poorly understood phenomenon with recent reports of a variety of echoes seen in patients with prosthetic mechanical mitral valves (MHV). Doe of the most visible types of SEC in the left ventricle (LV) and atrium (LA) has been described as microbubbles or gaseous embolL By utilizing a new pulse duplication system at Caltech, we attempted to compare the fonnatinn of gas bubbles associated with valve closure using two different mechanical bilenflet and one tissue valve in the initial position. Methods: The new system with a flexible LV model simulated physiological conditions, CO=5, (LAP)pe.~ffig, (LVP)te~=I30, (dP/dt)t.v=lg00, (SP)mm=100, HR=60bpm, SR=35%,, Bileaflat mitral MHVs were: n 27ram Edwards-Duromedies (ED) and a 3 lmm St. Jude Medical (SJM). Tissue mitral valve was a 27ram porcine Carpentier-Edwards (CE). For the experiments, the liquid was de-ionized, de-bubbled end kept at room temperatore. M-Mode Doppler and 2D-echo of long-axis and sh0rt-axis planes were performed using a Vingmed CFM750 machine. An Argnn-lon laser beam was used for planar illumination and high-~poed video flow imaging. Results: High-speed videography of the mitral valve from the atrial view-peint depicted bright bubbles formed in the immediate vicinity of hnth mechanical prostheses upon closure. However, the nature of the formed bubbles was different in each valve. While in the ED, the bubbles were originating from the circmnference and had a spherical shape, in the SJM, the majority of bubbles were originating from the gap hntwcen the leaflets and had a stretched flume-like appearance. M-Mode and spectral Doppler ultrasound images best detected these bubbles. No bubbles were detected by any of our imaging techniques when the CE valve was used. Conclusions: These observations clearly indicated an association nf SEC with MHV. The lack of such association with tissue valves may further clarify the understanding of SEC.

IMPLICATIONS FOR THE ASSESSMENT OF ATHEROSCLEROSIS R. W, Stetiler, W. Clem Karl, S. F. Ibrehim, R. S. ~ Boston Heart Foundation. Harvard/M~ Division of Health Sciences and Technology, and Boston University, Department of Electrical, Computer and Systems Engineenng Early assessment of ather~clemsis and monitoring of therapeutic efficacy may be possible through me~m'eroeets of andothellte'n-depeodent vns~enctivity (EDV). A new system based upon duplex ultrasound was coe.~eted for the noalnvasive measurement of EDV. Peripheral flow velocities we~'cmmanred with staadasd pol.s~-Doppl~ p ~ . ~ was measured from end-diastolic B-mode images in skew planes (i.e., planes rotated about 10* off-anis of the after), asing n new diameter-estimafim algo~thm. Tbe alg~thm exlr~l~l arte~M wall edge infonoutico by gradin~t-based edge dctectinn, and combined the exlracted edge ioformafion into a dinmeter nslimatc by a least-Ulouras Ixuabulic fit to the edge points. The autonmted diameter estimator was shown to have similar bias but only 20% of the variance of staedard ealiper-based ti.e. mmual) diameter estimates. ~ t s Of armial dizneter mul flow were condeeted on the brachlal artary in31 subjects, from 10 seconds before to about 2 minutes after the release of 5 minutes of foresrm waanalat ccolasinn. The role of EDV in tbu observed physiology'has been previously reported. The variability of the parameters Of EDV was estimated from repeated meamreman~ The post-o~Iosinn time of initiadou of Inehinl urtesini diiatlon, the maximunt dilation, and fae rata of decay Of the hyperemla wm'e ~ as flarametera Of EDV. Of thena, the maxlmum dilation was the most repeatable parameter wlum the peak hyperemla (representing the stimulus stsength) was e~isWained. The variability of the pammetera of EDV was dependent upo~ the nqyroduc~itity of the raantive ~ The tempemta~e of the forearm and the rate of onset Oftbu vascular ccolosion significantly affectod the hyperemia, but short-tenTh r~pmdanibillty ~ d i e s yielded c o ~ t results. measurements of EDV may be seffiolentiy accurate to allow the evaluation of antiatherosclemtic therapy on an individual besls,

Poster Presentations 142

145

DETERMINING CROSS SECTIONS OF VOLUME USING THE INTENSITY PROFILE TAKEN ACROSS THE CENTRAL AXIS OF FLUORESCENCE-FILLEDCONTAINERS. B. Thipakorn and V.H. Huxley Dept. of Physiology, U of Missouri, Columbia, MO 65212

A MODEL FOR THE INITIATION AND GROWTH OF EXTRACELLULAR LIPID LIPOSOMES IN ARTERIAL INTIMA. Yongyi Yin s, K w a n g - H e e L l m I, Sheldon W e l n b a n m 2, Shu Clden s a n d David S. Rumschitzld ~ D e p t of Chem.Z & Mech.• E n g , City College of CUNY, New Y o r k , NY 10031 a n d D e p k of Bioeng?, Univ. of California, San Diego, L a JoUa, CA 92093 There is considerable evidence that lmnen lipopmtaln cholesterol, after crossing the ar~rial endothdium and en~rlng the intima S u m the vascular lumen, lodges in exUacelhflar lipid packets 0alxdod "liposomes") bound to extraco]lular matrix. These liposumns appear to form by occnsiona[ attsclnncut o f aa L D L to infimal malxix and to grow in situ mainly by appending available free LDL. The liposume size distributions observed experimentally in obronically hypercholesteremic (WHHL) and in short-term cholesterol-fed rabbits are quite different. We propoje a hierarchy of simple nucleation-polymerization models to doscribr liposome formation and growth. Most of these models admit analytic solutions. Even th~ simple,st of these (with only one adjnstablc parameter) agrees extremely well with the W H H L data. In contrast, the cholnsterol-fod rabbit data seem to rcsuk from the shun-term, non-uniform indmal history of L D L supply which is a consequence of the focal nature of the transendothelial L D L flow through isolated, transient leaky junctions. The same models used for the W H H L data, together with this indmal nonuniformity, superimposed upon a slow, uniform transrndothelial seepage also account for this data very well.

The most widely accepted assumption in studies of transcapillary exchange is that exchange occurs from a volume described by a right cylinder volume. Our in vivo evidence suggesLsthat this assumption is in error and that the volume of vessels are better described as elliptical. Therefore. we hypothesized that the video intensity profile taken across the central axis of fluorescence-filled containers differ with shape; significant differences between intensity profiles generated from the image of elliptical and circular containturs would be observed at -245~ around the central axis. Glass capillary tubes, round (50 gin), elliptical (48 x 20), and square (50 x 50), were filled with 20 mg.ml "1 Frrc-FICOL in Ringer's and viewed under a Fiuovert FS Leitz microscope with 20x long working distance objective lens. Their 2-D images were recorded on a Panasonic AG 6300 VCR through a CCD-72 Dage MTI camera, at fixed value of gain and black level setup, eofipled to a Geniisys Dage MTI intensifier, at fixed value of gain setup. Average 2-D video fluorescent intensity profiles (over a 20 x 71 gm area) were generated durlpg video tape playback for each tube. Tbe video intensity profiles were nomalized with to maximum intensity value. Then, the areas under the intensity profiles were calculated and compared. Post hoc tasting (oneway ANOVA) indicated differences between all 3 type shapes (round and elliptical tube, P<0.003; round and square tube, P<0.02; and elliptical and square, P
143

146

EX-VIVOMECHANICALPROPERTIESOF THE NORMAl.AND ANEURYSMALABDOMINALAORTA ML Raghavan,M.W.Webster,D.LSteed,M.ll. Mekaroun,S.C.Muluk,R. Shapiro,D.A.Vorp Departmentof Surgeff,Universityof Rttsb~gh,PiRsburgh,PA 15213

MEASUREMENT OF MECHANICAL HEART VALVE REGURGITATION IN PHYSIOLOGICAL MOCK LOOP Alex A.Yu. Kishore Puppala . Ned H.C. Hwang Depamnent of Biomedical Engineering, University of Miami

developmentof an aneurysmis believedto be assodatedwith changesin the mechanicalpopertias of the abdominalaorta, in particular, aiteraiiom in stiffness and tans~ sttung~ may ar abdominalaorticaneu~m (AAA)locrnaii0nand/orprecede AAArupture. WNlo In-vlvoimagingyields useM Informalonon the mechanicalpropertiesof AAA, it is very limited. Ex-vivotesting ~ o~er usefulWonnalionnot obteinabloby imaging. We haveperformedunla.,dalmedw'~al tesitngof excised aneu~/smaland non-aneutysmaiabdumkmfa o ~ specimens.Spedmeesodentedbothlongitudinallyand ckcomfomnltolZy wereobtainedfrom pabenlswithAAA undergoingsurgicalrepairend from organdono~ ('norm,Is'). ~ specimenswerecut into flat s~pe. and subjectedto continuousunlaxldextrusionin a tensile test~ appetatos at a conste~ s~im rate of 8 . 5 ~ . ~ ~ was receded at g4z u~l spedmenle~lure. Tim recordedforue-extatwiondate was convat~dto sbess.shain data usb9 Immm oeomethc,~mmlons of the spedraens and by assumingI~:an~mssibmiyfor the ~sue. From sWss-stedndate~ maxJmuma~as~cmodulus(E~ ~ndyblcl stress O'S)~ r o determined. Spacimans wero testedfor ~ groups: 1) ~ IongRudh~lorlenta~on(N,,4S);2) ~ drsumfi~ren~alo r ~ (N=16);and 3) normalaorta.~-,giludina~oflentaiten(N=8). The resultsfor eachgroupImear~SEh~ate ehov,r~in~ tableendexpressedin N/c~, EMwas significanitylowerb GroupI thanGxoup2 (p<0.06). BothEMendYSweresignificanb'ybwor (.o<0.05)fct Group1 thanfor Group& Themwereno significant ~lleronussIn othercomparisons. ~ lowerv a ~ of EM and YS In C,m~p1 (A/~,) as comparedwith Group3 (nommi)may be due to obnormali6esof the aor~cwell end sugoeslsthat the suacept~l~yof Group EM YS AAA to rupture is largely due to alterafions in mechanical 351• 65*9 properitus."nle differencesIn EM betweenthe longitudinalend 1 c~ferential direc~ons is inolcativeof the anlsultopyof the 2 581• 6 8 = 1 2 anet~mal wall. lhuse dateata impodantIn the devdopmenlof 3 585=121 13~• cli~ca]lyusefulbiome~tanicalmodelsfor

144 EFFECT OF HOUSING COMPLIANCE ON MECHANICAL HEART VALVE (MHV) TRANSIEN f PRESSURE AT CLOSING Z. Jon Wu, Bruce Z. Gan, Christian Ritt~ and Ned H.C. Hwang Department of Biomedical Engineering, University of Miami, Coral Gables, FL Detailed r study was performed on the effect of the valve housing compliance ou the ocduder closing velocity (CV) and the transient pressure (TP) at MHV do~CVmxlTPofaMedtrmicHallMHV w~simultanenusly measured inapulsatile mock flow loop with geometrical and dynamic similarities using a la.~" sweeping technique and a high froquency pressure transducer. The ~ t a weze conducted e n ~ physiological ventricalar and aortic pressures at heart rates of 70, 90, 120, and 140 benta/minute with cardiac outputs of 5.0, 6.0, 7.5, and 8.5 litors/minute, respectively. The simultanenue n~,asm'en~ots of the diac dosing velocity and the transient lXassure showed that TP generated ou the inflow side dropped balow the vapor pressta'e of liquid in both rigid and flexible mmmtings und~ physiological operational conditions. The amplitade of the transient pressure spike was proportional to the disc approaching velocity. Howcver the c h a r ~ ofthc t r ~ l~casurcs were significantly different in the rigid and flexible motmtings. In the rigid mounting, the pressure drop occurred immediately following the disc dccclcration. High ffcquency pressure oscillations wcxe observed about 450 ps followingthe prussuro drop. The preasure oscillations may be caused hy the violent collapse of the cavitation bubbles, in the flexible mounting, the preasm~ reduction pattern was different from that in the rigid mounting. ARer the transinat pressure dropped to the lowest, it quickly recovered end followed by a positive peak. High frequency pressure oscillaticm did not appear in the pressure trace. It was clear that the mountLngmechanism played a significant role in the MedW0nio Hall MI-W cavitation inception.

In mechanical heart valve prosthes~ (MHV), codudet closing is always associated with certain amount of retrograde flow know as the "closure volume". This is usually added by the amount of valvalar leakage due to manufacture tolerance. The sum, commonly rofered to as the "valvular regurgitatinn (VR)" is found to be a function of MHV dasi~. Queatificafiou of VR is mede in a pulrafile mock flow loop which ceesisted of a cemputer controlled linear DC motor, a disumsible valve chamber, and an afl~lcad system induded an elsstic acrtir rout and a series of two complianue and thnm resist- unita. The drive unit consisted of a solenoid linear displacement motor and a I~-~md computer serving as tie ~otrol unit A proper vantricular volume cm~e wee used to provide the hgcet sil~nalwhida is amplified to drive a piston which moved to piston oontrast and dilate the Silaatic ventrg'alar sas day0ouically. The presstare and flow wavefonm adjusted to mimic physiologic conditions were monitor~I in real time. In the physiologyical mock loop, the fluids diacharged for one minutc was onflccted, and thatwes purnpedhythe piston displacement was caloulated, Thcdiffrre~eofthasrtwo vaines ~ e s e n t s the n:gusgitatot volume of the MHV. A 23ram St. Jude medical biolcaflet mechanical valve was tested in the aortic position with a Carp~tir Edwards tissue biopmtsthrsis valve mounted at the venticular inlet. The expeting'nt was ounductod at 70 bcaLVmin with CO of 5.0 litas/min. The results indicated the regurgitant volume was 0.71itedmin end that the relative regurgitation is about 13%.

S-31

S-32

Modeling and Measurements of Cardiovascular Function II

147

149

THE USE OF FORWARD RUNNING WAVES IN THE ANALYSIS OF ARTERIAL WAVE PROPAGATION N Stergiopulos, F Pythoud, C. Frei, C. Bertram*, JJ Meister Biomedical Englncefing Laboratory, EPFL, Switzerland *Graduate School of Biomedical Engineedng, University of New South Wales. Australia

INTERPRETATION OF WINDKESSEL COMPLIANCE C. M. Quick, J. K-J. Li, D. A. O'Harat, end A. Noordergnmft Cardiovascular Research Lab, Dept. Biomedical Engineering, Rutgers University tDept, of Anosthesi~ UMDNJ-RW.I Medical School tCardiovascular Studies Unit, Dept. Bioengineering, University of Pennsylvania

This study is concerned with the application of wave separation principles to the study of wave propagation phenomena. A newly developed wave separation method that takes into ar,cotmt the elastic nonlinearity of the arterial wall is used. Data were obtained from in vitro pulsatilo flow tests on nonlinuarly elastic tubes simulating the geometry end properties of the human aorta. Pressure and flow waves were measured simultaneously at two locations and a recently developed nonlinear wave separation method was used to obtain the forward nmalag waves. Fourier analysis performed on the forward running waves showed that the true phase velocity exhibits minimal dispersion and Womersley's linear theory predicts well the frequency dependent of the true phase, despite the presence of strong elastic nonlinearities. Furthermore, Ixensli-time analysisof the two forward tanning waves showed that it is possible to obtain a good approximation of the pulse wave velocity-pressure curve. It is concluded that additional information on different aspects of wave propagation may he ohialn~:l by the appropriate analysis of the forward running waves.

As envisioned by Otto Frank a century ago, the systemic arteries form a oumpliem chamber filled with blood. The relationsiaip of the change in blood volume in the chamber (V) to the change in blood pressure (P) was dermod by a constant, Cw, termed the systemic arterial compliance (Cw= dV/dP). From mass balance and this definition of Cw, Frank introduced the famous Wlndkessel to descdbe the pressrun-flow relationship of a linear arterial system. Implicit in this descdption is the assumption of inEnite pulse wave velocity. The rapid propagation of pulse waves led Frank to believe that the value of Cw approximated the sum of all the compliances of the individual arteries in the arterial system (Cr). This interpretation of C w dominates conventional opinion. To test the validity of this interpretation, we estimated the value of Cw from a model of the arterial system that includes a distributed compliance and/'mite pulse wave velocity. Results showed large errors when using C r to predict Cp This error can be attributed to the spatial dismi~udon of arterial compliance, which is not a consideration in the traditional inte~retation of Cw. A more general inmpretatlco of Cw is provided by transmission line theory. Wherever there is a wensition in compliance from one position tu enoth~, wave reflection may arise end affect the estimate of Cw. It can be shown theoretically that when the windkeseel is fit to data from a linear dislributed system, Cw is no longer s constant, instead, it is a function of heart rate, pulse wave reflection, characteristic impedance, end peripheral resistance, in the special case that pulse wave velocity approaches infinity, Cw degenerates into Cr in accordance with Frenk~s original inteq~retation. Thus. a more general interpretatioo of the WindkesseI compliance is presented that includes the traditional interpretutioo as a special case.

148

150

SIMULATION OF CARDIOVASCULAR RESPONSE TO UPRIGHT POSTURE USING A MATHEMATICAL MODEL R. Sdal Sdnivasan. KRUO Life Sciences, Houston,"lX 77058

A NUMERICAL MODEL OF THE CARDIOVASCULAR SYSTEM TO ASSESS STRATEGIES FOR CLINICAL INTERVENTION. E.T. Ozawa, R.D. Kamm. Fluid Mechanics ~ , ~ttsetts lnsfitote of Technology, Cambridge MA 02139.

The cardiovascular response of normsi human subjects to upright pasture (free standing) is similar in many respects to that observod with the application of lower body negativepr~s~re (LBNP) at ..40 to -50 mmHg [I |. However, there are some differences. For example, the changes in hesn rate sod diastolic arterial pre~uro are higher in the upright position compared to those seen during LBNP. We ~amined these differences using a simple mathematical model of the htraan s a r d i s _ _~ system develol~d to simulate response to orthostatie str~ses. The model uses gatie reistiandai~ on venom pr~mm~-v~un~ changes to determine the redistributiou of blood volume and the ~ change in senlral venous l~.seure caused by the orthostafic i'a'eu. The dynamic portion of the model, which allows calculation of ateody-state chs~es in pressure and flow pulses, condsts of time-varyiag els.stance ~ f i o n of the left ventricle coupled to the vascular beds. The central ~mous and alledal preeswe changes provide the requisite sisnaL~;for controlling the heart rate the pe~/ph~al resistances. The model has previnusly beeo used to simulate and analyze LBNP response, tn the Ineu~t stody, It waS ~ to s'tmulate the response to updght postt~'e taking into ancotm~the ~ in presmm: fdmull to carotid sinus and amlic barareceptm's. Simulations were performed to study the changes in re.~pcore due to blood volume loss and due to r in p;wameters inch as banxeflex gsin grst result from an exposure te weightinas conditious. This pape~ will prese~ the results dthese shouisdou studies and discuss the sensitivity of the response to upright pes~n'e to changes in c~dioveseulw Farametcrs. Preliminary simulation results indicate the respeuse to uprigl~ prattle to be more sensitive to blood volume less than the response to LBNP at -40 mmHg.

A mbuat and comprehensive numerical model of the cardiova_toular system has been developed to r tic effectivenessof vetoes mechanicaland pharmaceutical Inte~cotinns on a specific paticnL Such a model would provide the clinician with a more re,f'med intervention ixotocol, thus eliminating the time intenalve "timltion" proeedur~ that are performed in practice. The model is based upon a num~ical solution of the 1-D equations of motion in a geometrically accurate brooching netweck of the alledal system including energy losses at bifurcations, a von~oular model i n c ~ g specified ~nedependent wall compliances and onldirectioonl valves, end lumped parameter vono~ and pulmonary cireulatoW systems. Also incoqxxaled ate damping mechanism.s related to the phase dependence of the wall shear during periodic flows, and the viscoclantic behavior of the arterial walls, both of which have been shown to be important in the reproduction of realistic arterial wnvefeems. The model is ~ l c of rcfwoducing the conlplex wnveforms observed l~ysiologically. Key featur~ such as the increase in wave amplitude with distance from the heart due to vessel taper, major wave reflections, and higher mode oantimtions observed in the upper extsemities are all captured by the simeL~on. The model will be fmthes modified to inclode hammcept~ feedback, and system identification tochMques will be applied to "train" the model to mimic more toulisdcally the behavior obtainedhorn UtecaUm:and direct I~..asu0rements Erom1 ~ SuPlxxt for this project has been peovided through a fellowship from the Naficoal Scienoe Foundation, and the John Stepbens FeUowship fee Health Sciences.

1. Smith j r and Ebett TI. General response to 0rthostatic stress. In: Circulatory Response to the Upright Posture, edited by YJ. Smith, Boca Raton, FL: CRC Pre.~, 1990, pp. 1- 46.

150a COMPUTER SIMULATION OF THE DYNAMIC INTERRELATIONSHIP BETWEEN REGIONAL MYOCARDIAL FUNCTION AND METABOLISM Karen L. Naim, William P. Santamore 1, and Dov daron Biomed. Eng. & Sci. Inst., Drexel Univ. *Dept. Of Surgery, Univ. of Louisville A model was developed to study the dynamic relationship between regional myocardial function and local coronary perfusion. The LV hemodynamic model consists of five parallel compartments. Each compartment is represented as a time-varying elastanca, allowing convenient estimation of regional myocardial O= consumption from compartmental peessure-volume-area. The coronary model contains five corresponding compartments, each representing the vessels that supply this myocardial region. The time-dependent effect of regional LV mechanics on local coronary flow is represented as intramyocardial pressure. The level of regional LV function is calculated on a beat-by-beat basis by adjusting regional contractility according to the status of the local 0 2 supplydemand balance and its effect on tissue energy stores (phosphocreatine content). The model was used to simulate both partial and complete coronary artery occlusion affecting various areas at risk. Results from the simulation of complete coronary occlusion compared favorably with corresponding experimental data fr the literature. For partial regional flow reduction, the model tended to overestimate dysfunction due to the exclusion of a representation of local coronary autoregulat[on. The dynamic multicompartmental model developed in this work successfully coupled regional myocardial function and local perfusion. In doing so, the model overcame a common limitation of many previous models that were unable to predict the dynamic behavior of the heart in response to coronary occlusion.

Oxygen Transport 151

154

THE DUAL ROLE OF THE RED BLOOD CELL (RBC) IN TISSUEOXYGENATION M.L.Ellsworth, T. Forrester, D.M. Collins, W.T. McCullough Department of Pharmacologicaland PhysiologicalScience, Saint Louis University School of Medicine, St. Louis, MO 63104

P R E D I C T I N G END-CAPILLARY PO= IN ATHLETIC AND NON-ATHLETIC ANIMALS AT VOsm~ T K Roy & A S Pepsi Dept of Biomedical Engineering, Johns Hopkins University School of Medicine

Local regulation of microvascularblood flow is a complex process in which the needs of the tissue must be communicated to the vasculatureenabling the appropriate matching of oxygen (O~) supply to demand. Previous studies by Stein and Ellsworth (J. Appl. Physiol.75:1601-1607) suggested that O~ content is more important for maintaining 02 supply than PO~ during severe hypoxia. The only component of the O=transport pathway which is directly influencedby O~ content is the R.BC, the primary O2 carrier. We propose that the R.BC is involved in the regulation of Ch supply, serving as both an O 2 sensor and affector of changes in (3= delivery via release of ATP. The ATP sabsequendy binds to P~, receptors nn the endothdinm inducing an increase in O2 supply. To test this mechanism, we evaluatedrelease of ATP from hamster and rabbit RBCs exposed to normoxic, hypoxic and low pH conditionsin v/t,-o.We found that hypoxia (PO2-17 4- 6 m m ~ and low pH (pH- 7.06 + 0.07) each siguificandyineseasedthe amount of ATP released compared with normoxia (PO~ - 87 4. 4; pH 7.38:1: 0.04). In the hamster retractor musde microdrcnlation, imraluminal application of 10~M ATP to 1" and 2* arterioles using a micropressure system inducedan 8-10% increase in diameter 140 ttm upstream. Application of ATP near capillariesincreased KBC supply rate by 31% in arserinlar capillariesand 81% in venular capillaries,due to an increase in lineal density. Adenosine (10~M) and comrol solution had no effect on vessd size. The administrationof L-NAME (0.4 mg/100g body wt.) diminated the vasndilatory response to A'I'P indicating a role for NO in the ATP response. The idea that the RBC is more than just an 02 delivery system is intriguing and worthy of timber investigation,ffJL-39226 and RCDA HL-02602 (M/E))

Identifying the resistances to O2 transport from the erythrocyte to the mltoc.hondrion is crucial to efforts to enhance ~ygen delivery. A steady-state homogeneous model of oxygen supply and delivery was developed to predict the minimum (critical) end-capillary P02 required at ~/OT,m~ to prevent hypoxla in s cylindrical region of skeletal muscle surrounding a capillary. Capillary density was used as a measure of ~xygen supply and mitochondrlal density was ~ to predict oxygen consumption. The effects of kinetics and hemoglobin facilitation in the red blood cells and myoglobin facilitation in the timue were taken into account. Calculations done for selected skeletal muscles in three adaptive animal pairs yielded values of end-capillary PO2 and SPa consistent with measured values of mixed venous Pea in maximally working animals, and were found to be independent of predicted muscle V O s ~ . No s~'stematic difference in end-capillary POs was observed for similar muscles in athletic animals as compared to non-athletic animals. Calculated intracapillary Oz transport resistance fractions were found to range from 30 to 70%. Further investigation is required to explore the extent to which heterogeneity plays a role in c~ygen transport. [Supported by NIH Grant HL 18292.]

S-33

152

155

E X P E ~ N T A L AND THEORETICAL SIMULATION OF OXYGEN TRANSPORT w r r l l ERYTRROCYTIgSAND E,XTRACELLULAR HEMOGLOBIN SOLUTIONS T.C. Page, C.B. M c ~ , WJL LigM* and J.D. Helinms. Cox Laberatoty f0r Biomedical En~neexing. Rice University,Houston, 'IX 77251 and *BinpumCorporation, Boston, MA.

ENHANCEMENTOF TISSUEOXYGENDELIVERYBY A PERFLUOROCARBONBLOOD SUBSTITUTE R.A.Uneanmeler,T.K. Goidstlck, R.D. Braun,J. Wu and S.N. Vaslef Departments of Biomedical and Chemical Engineering,Neuroblology& Physiology,and Surgery, NorthwesternUniversity, Evanstonand Chicago, IL

Estracelininr hemoglobinsolutions have zeod'veda glest deal of attention as potential blood substitutes. Scveral~,d;,~,~ ate cun~ntly urxlsrgning clinical trials. Resesrda on these products has ban onguins, but much is unimownabout their c~ccts on oxygen transpoR to tis~c. This is pdmarily due to the r associated with In rive oxyg~ b ' a ~ z t expefimestation. Quantitative data have apt been available for Immmeter or doniug dsterminafian. Because of the complexitiesof In rive studies, an e~ct~lcmAI system for In v/ire s~aelatica of oxygen Uanspmtto tissue has been created, Tiw system includesa 25 lain diameter"capilin~" embedded in a 120 gur fleck siliconefilm which is fmnrallatedat asch r and Ixafessdwith axl blood cell suspensions, hemoglebinsolutions, or mixmt~ ~ . By m~umin~_ the capillmy under a micmscape a dual wavelength microspeclmpbotou~'a'ie is esedto detemlizethefrac~ml s~maion of the utmplcat variousaxial positions. The oxygen tension of the sample and of the cxlmcapiU~ spece is ceat~lled to simulate uptake in the l u ~ or rele~e to the tissue. Expemnc~ havebe=npcffonzed on washed h.,r.An ma ~ s~..n=o~, ~ bovine hemoglobin, a ~ mo~ed ~o~merin~ bovi= hemoglobin. P.e~ts show th= effect on traus~t of alteration ~ the slmpe of the oxygca dissociafioucur~candoxyhemoglobindiffmicecocft~cicm. Wnskceutinuesonmistures ~ n:d blood celia and emaceIluinrhemoglobin to identify and quantify O~ transport cahssunnsnt The experime~ willbe ~m,,!=tedwith a pn:dictivemathematical model which has previously been shown to be in good agreement with experimental tesnits on ted ceU suspcndoes and on hemoglobinsslu~om. Snppo~d by NIH gemt HL19824, TexasA'rP 8rent 003604-051, and Bi0pem Co~., Bosto~ Mum

Oxygent (Alliance Pharmaceutical Corp, San Diego, CA) is a second generation fluorocarbon ernuislon(pedluorooctylbromide. 47% by volume) wlzh high 0 2 solubility. In our expedmentaon anesthetizedcats, microe~eotrodsswere used to record PO2 in the vitreous humor dose to the retina, providing an essentiallynontnvaslvemeasureof tissue PO2 in a portion of the CNS. BaselinePO2 and the increase In PO2 dudng two minute episodes of hyperoxia (60% or 100% 02 Inspired) were measured before and after I.V. Infusionof each of anverel0.5 to 1 g/kg dcass of Oxygen or the emulslf34ng vahlcJe alone. The veh~e had no effect, but as little as 0.5 g Oxygeot per kg body weight Increased tissue PO2 substantially. After 1 g/kg, the increase In PO2 dudng hyperoxia was enhanced by about 60%. Subsequent doses resulted In further enhancement, w;th 3 g/kg Wov4dlng a maximum cumulative effect of about 150%. Theseeffectsere muchlarger thanwo~d be expectedon the basis of the Inc/eassd02 contentof attedal b~codend suggestthat smalldoses of Oxygentmight havea clinically beneficial effect. Separateexpeflmeotawith extracorporssl blood oxygenatorsshowed that Oxygentalso Increased02 deliveryto blood when blood flowed within the capillary tubes of the device, but not when It flowed outside the tubes. We speculate that the enhancementin vitro and In vivo may I ~ a near-wallexcessof the small (0.25/Jrn) padluorocarbon particles. This may enhancemesstransport both by the location of the partic~ss and the infinite raleasskinetlcaof the berfluorocarbon. Currentwork focuses on a theoretical model of the effect and an experimental test of the prediction that tissues besides the retinaare slmnadyaffected. (Supported by NIH EY05034nnd Alliance Pharmaceutical.)

153

156

Pea HETEROGENEITY IN SINGLE CAPILLARIES USING A NEW MODEL TO ANALYZE PHOSPHORESCENCE DECAY CURVES R N Pittman, L Zheng and A S Golub Dept of Physiology, Medical College of Virginia, Richmond, VA 23298-0551

QUANTITATION OF RETINAL VESSEL OXYGEN SATURATION USING DIGITAL IMAGING Sdnivas, S.V., Tiedeman, J,S., Beach, J,M. Departments of Biomedical Engineering and Ophthalmology University of Virginia Health Sciences Center, Charlottesville, VA 22903

We have used phosphorescence lifetime (J. Biol. Chem. 262:5476, 1987) to determine Oa tension in single capillaries. Palladium meso-tetra (4carboxypbenyl) porphine (10 mg/ml, pH 7.40, bound to bovine serum albumin) was used as the phosphorescent O= sensor. Our measurement system consisted of a microscope configured for epi-illumination, a strobe flash lamp, a 430 nm bandpnss excitation filter and a 630 nm cut-on emission filter. A rectangular diaphragm was used to limit the illumination field to 10 lain x 10 #m0 and an end window photomultiplier tube was used to detect the phosphorescence signal, that was then input to an A/D board in a personal computer. In vitro calibrations were carried out on red blood cell suspensions and plasma placed into rectangular mlcrocuvettes (light path length 20 pro) at four different O= concentrations (in %): 0, 2.5, 5 and 7.5. In vivo tests were carried out on capillaries of the retractor muscle of anesthetized hamsters. Phosphorescence decay curves were analyzed by a new model of heterogeneous Oxygen distribution in the excitation volume. Mean PO2 and a measure of PO= variability within the excitation volume were calculated from phosphorescence lifetimes obtained from individual decay curves. The time course of PC)= obtained during 0.5 see measurement periods (5 decay curves) at a given site along a capillary indicated the presence of a gradient in Pea within the plasma space between and near red blood cells. The mean POa for capillaries over the 0.5-see observation periods was 26 y. 4 mm Hg. The magnitude of the POz gradient in the capillaries, expressed as the coefficient of variation, was 68%.

Non-invasive measurements of blood oxygen saturation (O=SA'I~ have been obtained using a fundus imaging system employing a 16-bit digital camera. Values of O~SAT were determined by measurement of vessel optical densities (OD) at two isosbestic~ wavelengths (569nm and 586nm) insensitive to OsSAT, and at a third ($58nm) where absothance of hemoglobin (I-re) and oxyhemoglobin (HbOs) differ maximally. The retina was illuminated through the pupil with an annulus of light from a flash lamp (520-600 am) and reflected light was filtered using loam band pass filters mounted on a wheel. In order to determine vessel optical densities, digital images were scanned using an algorithm besed on the method of Delori (1). Briefly this method detects vessel edges and calculates vessel OD based on light reflection at the vessel center and surrounding tissue using the relationship OD = I n g j ~ ) . We solvedlinear equations, based on the Lambert-Beer relationship, OD~ = ( ~ ~ [I-Ib] + ~ ~ [I~OJ ) d + S where n is light extinction, Z is wavelength, d is vessel width and S is wavelength independent light scattering (2). Using this technique we evaluated OzSAT in 100-200 micron arterioles and vanules of the retina in I/uman subjects. We intend to use this system to study oxygen uausport in diseased retina. I. Delori F.C. (1988) Appl. Optics 27, 1113-1125. 2. Pit'anan R.N. and Duling B.R. (1975) J. AppL Physiol. 2,315-320.

S-34

Microcirculation

I

157

160

MOLECULAR ARCHITECTURE OF ENDOTHELIAL CAVEOLAE: POSSIBLE STRESS-SENgING ORGANELLEg; J.E. Schnitzer; H a r v a r d Medical School; Beth Israel Hospital; Dept. Pathology; Boston, MA 02215 Caveolae or noncoated plasroalemmal vesicles are very a b u n d a n t in many endothelia of the continuous type. Although a function in transport is likely, they may also play a role in cell surface signaling including even responding to mechanical forces. Mechanical forces can disrupt caveolae. Increasing intravascular p r e n s ~ cause significant attenuation ot the endothelium and progressive loss of caveolae without disruption of the junctions. We have purified caveolae from the luminal endothelial cell plasma membrane which was isolated directly from rat lung tissue (Schnitzer et al., P.N.A.S. 92: 1759, 1995). Further molecular mapping of these membrane fractions reveals that caveolae are specialized microdomains on the cell surface containing many molecules required not only for transport but also transmerobrane signaling. As such they may act as molecular filters capable of signal and transport screening not only through binding events but also stretch effects. So far, w e found that the caveolar structural p r o t e i n caveolin, the inositol trisphosphate receptor and calcium ATPase reside concentrated, ff not exclusively, on the cell surface within caveolan. Members of protein families responsible for different aspects of vesicular trafficking including vesicle budding, docking and fusion exist enriched in the caveolae (VAMP, annexins, NSF and SNAP). Various signaling molecules are found in caveolae including GTP-binding proteins and tyrosine kineses. Specific binding proteins such as insulin receptor and albondin are enriched. Scanning electron microscopy shows a substructure for caveolae but the molecules of this are tmlmown. With the molecules so far identified in caveolae, it is possible that caveolae may act as pressor or stress-sensing organelles that could relay messages into the cell, for instance via calcium signaling or local phosphorylation events.

SHEAR STRESS ALTERS ENDOTHELIAL MONOLAYER TRANSPORT PROPERTIES H W gill, Y S Chang, J A Frengee and J M Tarbell Physiological Transport gtudks Laboratory, Department of Chemical Engineering, The Pennsylvania State University Blood vessel wails are lined by a monolayer of endothelial cells (EC3) which regulate water and solute transport between blood and surrounding tissue. Blood flow imposes mechanical shear stress no endothelial cells and this can affect transport rates. We have used bovine aortic endothelial cells plated onto polycarbonate filters at 9 days pastplating as an in vitro model of the endothelial transport harrier. Parallel plate and rotating disk shearing devices have been developed to impose controlled mechanical stress on EC surfaces while allowing me~nsement of hydraulic conductivity (LIp) and albumin permeability (Pc). Both Lp and Pc respond acutely to step chengr in sheer stress at levels as low as 0.5 dynes/coO. At 10 dync~cm.?., Pe is elevated 10-fold within 30 minutes while Lp is elevated 3-fold after 4 hours of exposure to shear atre=. When shear is removed,'Pe returns m baseline within I hour whereas Lp remains elevated for at least 2 hours. This data suggests that more than one transport pathway Is affected by shear stress. The mechanism of shear stress on EC transport is not wall understood, but a purely physical one, such as the deformation of fluid conducting channels, seems unlikely, since the response of I..p can be completely inhihited by elevation of intrarellular cAMP and highly attenuated through inhibition of G-proteins, Acknowtedeement: T h ~ work was supported by NIH Grant H L 35549.

158

161

ARTERIOI.AR WALL SHEAR STRESS, BRANCH ANGLE AND FLOW DISTRIBUTION ARE MODULATEDBY ENDOTHELIALCELLDILATORYPATHWAYS. Mary D.S. Frame and lngrfd H. Sarelius Department of Biophysics, University of Rochester Medical Center, Rochester NY 14642

IN SITU CAPILLARY HYDRAULIC CONDUCTIVITY: RESPONSE TO ALTERED SHEAR STRESS D A Williams and V H Huxley Department of Physiology, Unive~ity of Missouri, Columbia, MO 65212

Our purpose was to delermine whether endothelial celt dependent dilalory tone regulates flow distribution in Ihe arterioles which control tlow into capillary nelworks. Bifurcation geometry and hemo~namic variables (cell flux F, wall shear stress T=) were delermined Ior 4 sequential branches along a transverse feed arteriole (-25 izm) in the superfused cremaster muscle of anesthetized (Nembutal, 70 mg/kg) hamsters (N=37). CONTROL F was significantly greater into the upstream branches (1561+315 cells/s', mea~s.e.) versus the downslream branches (971• cells/s) along the teed. Supeffusate exposure to 50 IzM N~ plus 50 p.M indomethaein (BLOCK) constricted branches 14+4%" and decreased F, with the eet result that red cells were diverted downstream. The correlation between red cell distribution (% of feed inflow) and diameter during CONTROL (r=0.60) was disrupled during BLOCK (r=0.29). CONTROLTr was significantly higher in upstream branches (17.5.+.2.6 dynes/cm2.) compared 1o downstream (7.2:1:1.1 dynes/cm z) reflecting differences in cell velocity and consequently shear rate, but not v=scosity. During BLOCK, To)decreased significantly upstream (9.7+9.4 dynes/cm2") thus equalizing T,o wifhin lhe group. The angle of bifurcation was significantly greater upsiream (110.+-8")versus downstream (57• ; from CONTROLto BLOCK the angles changed • with a significant increase downstream (to 69• The relative increase in cell flux to downstream was preferentially diverted from those upstream branches that increased in angle from CONTROL to BLOCK (1935• to 455• cells/s'), while F was relatively preserved upstream when angte decreased (1037• to 593• cells/s). A~gle change and diameter change were unrelated (r=0AS). Inhibition of these pathways redisftbuted ted cells within this microvascular region with respect to branch location, b=furcation angle (related to upslream versus downstream location) and the changes in angle. ('p
Endothelial cells, which line the cardiovascular system experience continuous shear stress, the tangential force induced by blood flow. In culture, these cells produce intracellular signals and release products in response to changes in flow-induced shear stress. Working in intact tissue we have applied this information to studies of the acute response of capi0ar7 hydraulic conductivity (Lp). Mesentery of pithed frogs was exteriorized and suffused with frog Ringer's solution (14-16~ Each capillary was caanulated at 30 cm H 2 0 using glass micropipottes filled with 10 mg.ml "1 bovine serum albumin in Ringer's and marker red cells. The modified Landis technique was used to assess Lp in arteriolar (diverging flow at both ends), true (diverging and converging flow at either end), and venular (COnverging flow at both ends) capillaries. In vanular capillaries Lp correlated (Kendalrs Correlation Coefficient, Tau = 0.5; P = 0.04) with changes in shear stress calculated from red blood cell velocities measured before and after cannulation (time delayed integration). This relationship was shifted right (less sensitive to shear) when nitric oxide synthase was blocked by adding an L-arginlae analogue, NG-monomethyl L-arginine (10-6 M), tO the soffusate. When indomethaein (10 -5 IV0 was used to block r the relationship shifted left indicating greater sensitivity of veuaisr capillary Lp to changes in shear stress in the absence of prostaeyclin. These data indicate that, in addition to vasedilatation, another fundamental function of the cardiovascular system filtration (permeability to water) can be altered acutely by flow-induced force. Further, flowo induced filtration has a spatial component, occurring in venular capillaries, and appears to be exquisitely sensitive to two release products associated with shear stress. Supported by Missouri Affiliate - American Heart Association Fellowship to DAW and HL34872.

159

162

M E A S U R E M E N T O F L O C O M O T I V E F O R C E S G E N E R A T E D BY S I N G L E L E U K O C Y T E S IN VIVO AND IN VITRO.

ENDOTIIELIAL CELL ISOMETRIC TENSION DEVELOPMENT Z, M~ Gocckelcr and R. B, Wysolmerski Dept. of Palbelo~, St. Louis Urdversity School of Medicine. St. Louis Mo. 63104

ROBERT W. GORE. Ph.D. Dept. of Physiology. College of Medicine. Universily of AHzona. Tucson. AZ, 85724. We have developed a remote-sensing isometric force transducer for measuring forces tn sfngle cells with femtonewton-to-micronewtoa resolution (Am. J, Physiol. 263:C700-707, 1992.). We have used this transducer to make time-resolved measurements of forces exerted by single leukncytes as they migrate in vlvo end in vitro. In studies o f spontaneous force generation, neutrophtls were studied in vfvo in th0 intenfifium o f the h~.mstor cheek pouch. Alveolar macrophages were obtained by lavage from mice. rats and hamsters and cultured on glass coversllps. C c ~ were allowed to engulf ferromagnetic micrnsphcres (- 6pro die.), and forces were measured with the isometric force transducer. Maximum locomotive forces ranged from 1.9 to I0.7 nN (tenths of miomdynes). FIVILPstimulated macrephages produced less force than spontaneously active control cells. Also, fMLP caused a change in the pattern and frequency o f force generation from control, but did not increase the magnitete of force production. Spontaneous force genemtion was periodic and correlated with the length o f the leading lamellipod but not with generalized cell rnn'ling. Although extension o f the leading lamella is critical to locomotive force generstlon, our direct measurements suggest that lamellar extension may not arise from the same contractile prneesses driving forward motion o f the cell mass. Indeed, cell ruffling, lamellar extension, and locomotive force generation may be differentially controlled and may have different origins. (Supported by NIH Grant HL- 13437)

~[ial cells (EC) lining most ~ i s form a continuous layer that nonvaBy constrains proteins as well &s formed blood elcmcn~ to the vascular lumen. The loss of continnity of the cnd,JthaliaI sheet leeds to increased pemleabd~, md the (kvdc~aent OFedema. We have previoesly sbewn that phasph~ 'lati~ of EC myoem ligla chains (MLC) by myosin lighl chain kinnse (MLCK) is occessary end sufiScient to iniuate EC retraenos. To direcOy test fee a myosin-based contractile system, we monitored the developmeat of EC isomeCic tension in tissue culture. Human umbilical vein EC develop a stable basal tcnsic~-tof 30-35 dynes w i ~ 5-6 day~ in c~,lt~-'c. T'hrombin atimulafi~ results in a rapid sustaJagd isom~ic contraction which increases 2 to 2.5 fold within 5 mi~ and remains elevated for at least 60 rain_ Resting isometric tension is associated with a basal pbosph(xylation of 0.54 mole PO,hnole MLC. After thrombin stimulation. MLC phosphaWlstinn inoreases te 1.61 moles POd~mole IvlLC within 60 soc. and ~ elevated ov~ the e~kmmg60 Tt'A36cp l x = p h o p ~ maps from both control and da'ombin stimulated culturesresolve both r ~ l a t e d Set-19 and t h p b e s p h o ~ Ser-19,Thr-I 8 indicative of MLCK a~ivatio~ Changes in the polymerization of ectin and myosin II correlate ~ y with the development of isometric tmsion and h~C pbosphegyisfir Activatios resolts in a 57% lacrosse in F-sofia cotaant within 90 s~, while 90% of the soluble myosin It associates with the reergAni71%oF=sct~ F-actin initallyforms a free re'work OFfilaments md thee reorBainzes into prominect stress fihers. Myosin II rapidly forms discrete aggrcgratns associated with the actin network and aasamaes a periodic dislffot~kn along the stress fibe~ Two disth~ forms of MLCK have bclm immanologieally defined in EC extracts, A 130 kD MLCI<. akin to the smooth mu,sele enzyme ~ d a anique 210 kD MLCK (MLCI~. MLCY~,ophespl~Tls~ gizzard and EC MLC in a Ca2+/calmodalmdependent maua~. Phospbepaptide maps show that MI,CK2t0 pbospho~latns MLC at Set-19 end Sor-19, Thr-18 (:~asistent with the known pbospbocylation sitos of smooth muscle MLCK.

Microcirculation II 163

166

A NEW COMBINED FLOW AND STRETCH CHAMBER FOR DIRECT VISUALIZATION OF ENDOTHELIAL CELLS DEFORMATION IN THE MICROSCOPE Y Tardy, AJ ljspecn and JJ Meistor Biomedical Engineering Laboratory, EPFL, Switzerland

Macroscopicand MicroscopicCharacterizationof LymphaticTransport Using

The effects of mechanical forces on vascular endothelial cell structure and function are generally studied using specially adapted flow and/or stretch chambers. The objective of this study was to design and test a new flow/stretch chamber that would allow to: first, combine, in an independent and controllable fashion, beth shear forces due to a flowing medium and stretch forces imposed by a deformation of the substrate, and second, do so on the stage e r a high magnification microscope providing direct visualization of the cultured cells as they respond to their mechanical environment. A rectangular compliant channel (50 mm long, 5 mm wide and 0.5 mm high) is created by flattening a cnstom-mnde waasparent silicone tube (120 mm long, 8 mm in diameter, 180 [zm thick)~ with two rigid sU-uts introduced longitudinally in the tube and pulled apa~ The struts are held externally by claws attached to piezoelectric translators which coo~rol the lateral deformation of. the channel. The tub~is connected to a flow system controlling the fluid shear stress in the channel. The bottom of the channel is plated with endothelial cells and is placed on the stage of a con-. fecal microscope, Real-time phase-contrast and fluorescence imaging of the cells can be peffonnod for various shear stress and substrate (channel) deformation. Up to 20% uniaxiel deformation perpendicular to the flow direction was obtained with a very good homogeneity (N=3fl points spread over the surfuce, mean defonnation=l 8%. SD=0,6%) in a: third of the channel surface. Preliminary results on endothelial cells under no flow condition showed that confluent ceils could follow the substrate deformation up to 18% whereas isolated ceils stopped deforming much earlier. This seems to indicate that intrecellniar junctions are important for the mechanics of the endothelium.

S-35

R T D T h e o r y and F R A P M.A. Swastz 1,2, D.A. Bork2. and R.K. lain 2. IMassnchnsetts Institute of Technology. Cambridge, MA 02139 and 2Massachusetts General Hospital and Harvard Medical School, Boston, MA 02114 The goal of this study was to quantify uptake by and flow through the initial lymphatics in the mouse tail. Interstitially injected FITC-destrans of varying size were used to visualize and quantify these processes. On the macroscopic scale, residence time distribution (RTD) theeW was utilized to measure net flow velocity through the lymphatic network as well as provide a relative measure of lymphatic uptake. The effects of concentration and injection pressure were examined to establish the usefulness of this characterization for studies of lymphatic function. On the micrnscale, fluorescence recovery after photobleacblng (FRAP) was used to m e , urn velocities in individual cupillaties, and microspheres were injected to visualize flow patterns. For an injection pressure of 40 cm water, the RTD approach resulted in a net steady-stete velocity (3.8:1:t.0 mlcrons/s) that agreed well with the average FRAP velocity mee~ured in I l l vessels and corrected for tonuosity (4.0 + 2.5 microns/s). Injection pressure greatly influenced the uptake rate but not the net velocity within the network, suggesting the pe~ance of a systemic driving force for lymphatic flow which is independent from the Iocni driving force for uptake. Finally, micrnspheres revealed oscillations in flow that conelated with respiration. These two complementary approaches provide a look at the in rive dynamics of lymphatic transport and establish the groundwork for studying lymphatic transport issues such as drug delivery pathways and cellular wansport. Furthermore, it shows evidence that pressure gradients generated by respiration are ttansmined to distant lymphatic capillaries and therefore should be considered as a possible baseline driving force for lymphatic flow.

164

167

SHEAR-CONTROLLED DILATION IN SINGLE OR BRANCHED ARTERIOLES LEADS TO UNSTABLE BEHAVIOR WHICH CAN BE STABILIZED BY SERIES FLUID RESISTANCES Timothy P. Hurrigan Orthopedic Binmechanics Lab, University of Texas Health Science Center at Houston

A NEW T E C H N I Q U E TO ESTIMATE THE HYDRAULIC CONDUCTIVITY IN SOLID TUMORS. Botcher, Y,. Brekkan. C., Netti, P.. Baxter. L.. and Jaiu R.K. Department of Radiation Oncology. Massachusetts General Hospital, Harvard Medical School. Boston.

Shear stress related vasodilstion is one of several mechanisms which regulate the diameter of atteriolas throughout the body. However, simulations which implement this mechanism have led to numerical instabilities. The purpose of this study is to assess the character of the instabilities seen in network models, and to see whether series ~-sistanees which are constant in time can stabilize these models. Two vascular regulation models considered in this study. In model 1 a shear-reactive arteriole is in series with a passive resistance, with the model subjected to a constant flow or a constant pressure. In model 2, two parallel sheur-eensitive arterioles are coupled to a passive resistance network, which i$ modelled by two series resistances and a coupling I~istance. The sum of the flow within the two arterlolns is assumed constant, or the total pressure drop is assumed constanL Shear stress regulation was implemented by a fir~-ordor rate equation. Nonlinear coupled differential equations derived for the urterinlas radii were linearlzed for small pen'urbations about equilibrium states to assess stability.The lineasized stability analysis was used to fred the series resistances needed fur stability. With an imposed pressure drop across model 1, the series rasistence I~ for stability was 1~ > (8 p L)/(3 11 r,*) with r, the equilibrium radius of the arteriole, L the length of the segment, and p the blood viscosity. With an imposed flow or pressure in model 2, divergence is prevented if the series resistances satisfy R,/(1 + 2R,/P~} > (8 tt L)/(3 n r,4) with R~ the ~ansveme resistance. Future work on the stability of con~rul schemes for vascular diameter should account for the fact that although the Murray optimum is unstable in simple shearsensitive ne~wurks, the predictions of the Murray model are berne out by many anatomical studies of the vascular system.

The elevated interstitial fluid pressure (IFP) in solid tumors is a potential barrier to the delivery of therapeutic macromolecules finch it opposes the convective flow of fluid and mncromolecalas out of tumor vessels. Mathematical modeliug has shown that the steep pressure profiles in tumors are dependent in part on the ratio of vascular to ioteratitial (K) hydraulic conductivity. The pressure gradient between the vascular and the interstitial space could be increased significantly by increasing IC In order to estimate K in solid tumors, we performed intratomoral infusions using a specially constrec~I needle designed to produce a radially symmetric fluid source. AI constant flow, radial steady-state pressure profiles were measured with the micropuncture technology. TO evaluate the influence of fluid reabsorption by the tumor vessels, K was also estimated in tumors after ci~alato W arrest (CA). K estimation was done by differential analysis and curve fitting of the pressure profiles based on an established fluid transport model (Baxter and Jain. 1989. Micrevasc. Res. 37: 77). K by differential analysis was in vivo 2.6.10"7[0.8-17.5] and post CA 2.9-10"711.9-4.6], and K by curve-flttiog was in vivo 2.1'10"711.3-9.0] and post CA 2.7.10"711.6-4.4] cm2/s.mmHg. Values ate given as median with range io pazonthesis. No significant differences were found between K In vivo and K after CA. The absence of reabsorptinu could be due to the fact that the central region of some turaors is known to be poorly perfnsed. In conclusion, we have developed a new method for the estimation of the in rive hydianlic conductivity in lumor tissue.

165

168

SIMULTANEOUS MEASUREMENT OF PROPAGATED DIAMETER AND M E M B R A N E pOTENTIAL CHANGES IN ISOLATED R A T CEREBRAL ARTERIOLES INDUCED WITH MICRO-STIMULATION. H.H. Dietrich and R,G. Dace)', Jr. Department of Neu~surge~y, Washington University, St. Louis, M O 63110.

TRANSMURAL COUPLING OF M I C R O C I R C U L A T O R Y N E T W O R K A N D INTERSTITIAL FLUID FLOWS IN TUMORS. J.W. Buish. P.A. Netti, M.S. Fan, and R.K. Jnin. Department Of Radiation Oncology, Massachusetts General Hospital, Medical School, Boston, MA 02114.

Propagation of vasomotor responses may be mediated by membrane potential changes conducted along mefioles, which is said U3 be an Important mechanism to regulate miceoclrculatoty blood flow. To test this hypothesis we Isolated and cannulated peneUafing atteriolea f~3m tat ce~brui cortex. ~ pressurization to 60 mmHg, heating to 37*(2, and viability tasting, adenosine trl phosphate (ATP, 10"2M) was applied locally using micro pressure ejection at the proximal end of the vessel. Membeane potentials o f smooth muscle (using fine mlcroeleetrodes) and vessel enameler (using digital video imaging) were measured simultaneously at the opposite vessel end. 50 to 450 ms pulses o f A l l " caused an average vessel dilation of 4.6%'(at 50 ms pulse) to 9.8% (at 450 ms pulse)of control respectively, 1129.3 4- 165.3 lun distal from the site of stimulation anti was maximal between 1.3 to 1.7 s afte~ stimulation. The remote vessel dilation was preceded by a membrane hype~olaxization of 7.3 to 16.6 mV which was maximal after 1.0 and 1.3 s after stimulation. The results indicate that a membrane potential change ptx)pagated along an ane~ole may be the mediator for the ptupagation of vasomotor responses with an estimated minimal electrical conduction velocity of 870 to 1130 lttm/s and the vasomotor response following at 660 to 870 Ilm/s. Supported NIH ROI NS30555-2 and Hoffmann-La Roche.

The growth of tumors and their response m treatment are determined by delivery of diffusible substances to their cells and hence by their blood supply. Relative to that in most normal tissues, tumor blood flow is highly heterogeneous. In this study we confider a potential source of the nonuniform blood flow: the coupling of the interstitial and vascular flows due to the transmurul flow through the leaky walls of the tumor blood vessels. A aimple medel is prasenaxl to dew.~ihe the beale fcettaes OFflow Uwoogh a cetwork of permeable and compliant vessels embedded in an ianuopid porous medium. The network is cor to be a regular mesh of identical vessels. The flow through each vessel of the netwo& is described by Pnisanifle's law; the transmural flow between the vessels and the external porous medium, is governed by Starling's law.- the fluid movement through the punster medium is described by Dercy's law; and the vessel wall is assumed to he linearly elastic. Our reanlss show that the hehaviur of micmci~ulation may he strongly modified as a result of vasonlur compliance and enhanced vascular leakiness of the tumor vessels. We find that not only does the vascular pressure generate the wuil-lmown, high central interstitial fluid Wesmre, bet the elevated interstitial fluid pressure in tam alters the vascular pressure distribution. These changes in vascular pressure disttibutim result in a modification of the bleed flow ~ As the leakiness and compliance of the vessels increases, the blood is divexted away from the center of the tumor to a mine peripheral flow path. The elialeal s i g u i f ~ of these results is that drug defivory and oxygenation needed for radlo0grapy will hampered in the cenWal regico of the tumor. This work was supported by CA-56591 and CTS-9057412.

S-36

Blood Flow in Arteries I

169

172

FLOW-INDUCED DIAMETER OSCII.J.,ATIONSIN RAT ARTERIES PERFUSED IN VITRO C.-A. Pon'et, H. Achakri, N. Stergiopulos,J.-J.Meister Biomedical Engiueoring Labuea~y, Swiss Federal lustitota of Technology, Lausanne, Switzerland

INVESTIGATIONOF THE CORRELATIONBETWEENPRESSURE AND VELOCITY IN RIGID MODELS OF ABDOMINAL AORTIC ANEURYSMS R. A. Pea~e and T. 1. Riehle Department of BiomedicalEngineering,Tniane University, New Orleans, LA 70118

It has been observed that significant vusomotion (Iow-fieqonuey diameter oecillatiom)exists in arteries of the human arm. It has also been observed that often, mean 0~uuetes oecillationslag mean flow oacillationsby a time shift of 10 to 20 s. Thus, it was hypothesised that flow may drive arterial vasomotioo.The effect of flow variations on arterial diameter wese investigatedin vino on isolated rat modes. Astedni segments of feBxlflti lind mezelUeric wet"e exnised, mestnted OI1 m i ~ n n n l ~ ~ p e ~ with Tyrode's solmion. Perthsion pressure was kept at 80 mmHg. ~ equilibration at a constam flow (tO0 #/rain), aneciee were precoaseicted by nempinephrine (I t~)- The amp'inswere subjected to different flowconditions:step change (in~.._~ and decrease)and Iow-fleqnancy sinusoidal flow variatious (frequeacies ranging from 0.005 Hz to 0.1 Hz). The extetml diameter was measured comimamalyusin8vkleo-micmzec~. The results show IJmt the flow rate increase (decrease) induced vessel contxaction (dilation). WhenUaeflow varindalnuanidelly,the arteay diameter showedoecihati~ witha deiny (in the oeder of 10 s). in the oase of mesenterin arteries, the flow-indncedvadatious were anpedmpused by norn#uephdae-thdueed vasomotion.An inclausein the frequency of the flow lead to a deornase of both the amplitudeof die flow-indue.eddiameter oseillatious and the phase-shiftbetween flowand diameter.

It has been estimatedthat abdomin,~ anrfic anetmyslasocour in as mnoh as 2-3% of the genernipupulafion,and that their mptore produces a mortalityrate between 78-94%. As an investigationinto the mechanical factors that lead to aneurysm rupture, flowfield and pressure measurements are presented for a range of anemysm sizes and Reynolds numbers. Seven rigid models wi~ elliptical anomysm geomeW/wereconstructed with unfform lengths of 4d and of diametors that rangedfiom l.5 to 3.0d, wbere d is the innor diameter of the undiintedentranoe tube. The flow fields wore ~ qualitativelyby color.-Dopplerflowimaging and were quantifiedby laser-Dopplor vel~. Pressere was mnamred by a set of tnmdnoen connected to a leries of taps along the model walls. Vismlizau'onimages, velocityprofilesand pressure measoremems ore plasonte~ It wus foond that over a lange of Reynolds numbe~ correspondingtodn v/voflow ra~, the flowvaried from I~tmimtr through J ~ t l ~ t i y to fully tm'bulent. When the flowis tusbniont,the torbnionca intensityiuereases with the bulge diametor. Imensityleveisapproaching 40% are achieved in tha largestmodels, m ~ that the Imlgeacts as a steongtmbulence amplifior. Tudo~lenoeproporfies are discossed, and the correlationbetween the flowand the wall s~rmsfield is investigated. The reanlts implythat the wnil ~ prodnoedby torbulent flowplace large anemysms at greator risk of mlamr than smallanoorysmsin vlvo. Suppuned by a WhitakerFonndadon Research Grant.

170

173

ADAPTATION AND PATHOLOGY OF ARTERIES FROM HEMODYNAMIC WALL SHEAR STRESS David N. Ku, Christns P. Markou*, Stephen Hanson*, James E. Moore+, and Seymour Glagov#. Georgia Institute o f Technology, Atlanta, G A 30332-0405, *Emory University, +Florida International University, and #The University o f Chicago.

DYE VISUALIZATIONOF FLOWIN STENTED ARTERIALMODELS James E. Moore Jr., Joel L. Ben'y*, VirginiaS. Newman*, WilliamD. Rondi* Mechanical EngineeringDepartment, Flonde InternationalUniversity,Miami,Florida *Bowman Gray MedicalSchool, Wake Forest University.Winston-Salem,North Carolina

The human arteries are part o f a physiologic system designed to transport blood and nutrients between different organs. These arteries must be able to adapt to differing flow conditions as well as react to abnormal situations such as bleeding. The focus o f our recent studies has been to characterize the physiologic response o f arteries to pulsstile hemodynamics under specific flow conditions. Experimental and computational studies are used to quantify the time-vmying wall shear stress conditions. Biological experiments are used to elicit the short-term mechanism ofplatelet adherence, the medium-term adaptive response o f intimal thickening and arterial diameter, and the long-term localization o f atherosclerotic plaque to differing shear stress levels. Arteries appear to adapt to these widely differing hemodynamic conditions by normalizing wall shear stress by several biological and physical mechanisms consistent with the functional needs o f the cardiovascolar system in the short and long-term.

171 MEASUREMENT OF ABDOMINAL HEMODYNAMICS IN SWINE USING LAPAROSCOPICALLY-POSITIONEDFLOWMETERS L. A. Capitidi,J. F. ]~[ahefty, B. D. Kuhan, D. L. Fry, M. H. Friedman Biomedical E~gineedngConter, Department of Medicine, and UniversityLaboratory Animal Resources, The Ohio Slate Univel~ity,Colu~has, OH, and US SurgicalCorp., Norwalk, CT Hemodynamie factorsmay conm'boteto the localizationof athorosclercais throughtheir influence on vascularpermeabilityto maoromolecules. We hypothesizethat changes in the/oc~l nnar-wall hemodyusmie e a ~ e ~ t t , comeque~R to more global changes in flow rate, heart late and flow I~trfition at branches, elicit an adaptive response in the vascular endothelitan, during which p e ~ i l i t y is Irauslentlyelevated. This hypothesis is being tested by placing a reve~ible artetiovonoue shunt between one femoral artery and vein of a pig and meesm'ing the uptake of Evans Blue dye-laggednative albtanin in the terminalaorta and both iliac astorles. The uptake dala see suhaeq~.mly related to the fluid dynamic changes induced by the shunt asing lasor Dopple~ velocimetryin flow-throughreplicas of the gntm~l'~abdominalvessels fal~icatedfrom post-mortem vascular casts. Measurements of flow in the relevant IXrtions of the swine vazeuismlaprior to ~anting and throughoutthe protocol are made usingtranalt-timeuhiusomld flow wanedueea~, modifiedto facilitatelapamecopicpositioning. Thisapproach providesmore ~ c flow data alld minlntal sorgicaJ invasive11~l~;. Initial eape#~oats have sought to charnctedze the hamodynamica of the region of integest,both at haselineand in the presenoe of a shunL Fmdinns: at baseline, 13~ of the flOWtlgough the e.,xtm'nalihac artery (El) entela the ciromdlex branch (CXI) (N-3); o ~ g the shunt increases the ipailateralEl flow by a factor of 2.5 (N-2) and reduce~ conlralatend El flow by 12% (N-2), but has no significanteffect on il~ilatorniCXI fow (N-4). (Supportedby NIH Grants HL50442 and HI.29095)

Intravescular stents were developed in older to improve putency rates following balloon angioplasty, but thrombosis and neointimal hyperpl~ia can develop within the stem and compromise patency. Even though the original motivation for stenting is largely sofid mechanical, the patency of the stna~d artery may be more strongly determinedby bemodynamlc flow patterns. Dye injection flow visualizationwas perfonued in a scaled up (300%) in vitro model of a stented artery. A transparent compliant artery model was coastmcted from Dow Coming Sylgard 184, and a modelof the Johnson and Johnson Pnimez stem was fabricatedusing cle~ acryfic. Physiologicflow conditionsrepresentativeof the femoral artery (Reynolds nmnber = 170,Womersleyparameter = 4.4) were providedby a gear pompcontrolledby a computer. dye was injectedusinga small tube positionedsix diamete~ proximal to the stem. In the first set of experiments, the Sylgardtube was constructed to have approximately the same diameter as the stent model, representingexpansion of a stem to match the artery diameter. Stagnationof dye was observedbetween the struts of the stem in systole. In late systole, a large vortex f~-med approximately 1.5 diameters proximalto the stem. Washoutof dye between the stuns occtmed after 4-5 cardiac cycles. In n second set of experiments,a Sylgard tube with a diameter 10% less than the stent modelwas used to represent a 10% stem overexpansion. Dye stagnation again occurred during systole. In late systole, a large vortex formed 0.5 diametees proximal to the stenL Completedye washout occurred after 12-15cardiac cycles. These results demonstrate that stunt placement creates flow patterns which may lead to the formation of thrombosis md neoindmal hype~lasi~L Factors which appear to most affect the hemodynamlc environmentale the compfiancedifferencebetween the proximaland stented segments, the degreeof overexpansinn and the amount of"pillowing"of arterial wall materialbetween the struts.

Blood Flow in Arteries II 174

177

EFFECT OF GRAFT CALIBER UPON HEMODYNAMICS IN BYPASS GRAFTS Keynton, R.S., Evancbe, M.M., Sims, R.L., Rodway, N. and Rittgars, S.E. Department of Biomedical Engineering, The University of Akron, Akron, Ohio.

VISCOUS FLOW IN STE/~OTIC ELASTIC T U B E S W I T H F R E E M O V I N G BOUNDAR.[ES, PART I: S T E A D Y FLOW D. T a n g

Wall shear has been implicated to be inversely correlated with vascular disorders such as intimsl hypespinsia (IH) [Rittgers SE, et al., Circ. Res., 1978; Einns RL, at al., J Vase. Surg.) 1989]. Few studies have directly shown a relationship between wall shear and the development of vascular diseases. It is the purpose of this paper to evaluate the effects of graft caliber upon hemodynamlcs and IH at the distal anastomosis of vascular bypass grafts. Tapered (4-7 m m ID) e-PTFE grafts were placed with 30 ~ end-to-side distal anastomoses in the common carotid arteries of 26 adult mongrel dogs. The distal anastomoses consisted of graft-~ diameter ratios (DR) of either 1.0 or 1.5. A.qefini velocities were measured with a n e w in vivo triple crystal ultrasonic (PUDV) transducer [Keyntou RS, et ni., IEEE Trans. BME, 1995] at 9 far wall and 6 near wall axial loeatinnsat 0.5 L/D uicrements. A cubic epllne regression was performed on the PUDV velocity data vs. axial position and wall shear rates were computed using a polynomial curve fitting procedure. Using the g~ft toe as the initial reference point, it was noticed that along the far wall D R = 1.5 had a larger reclrculation region and the stagnation zone was located further upstream than DRffi 1.0. Along the near wall, DRffi 1.5 was found to have higher normalized shear rates compared to D R = L 0 . in studying the region of greatest IH along the far wall, the maximum shear rates were found to be 527 < mean= 1333 <2175 s4 and 898 < meuaffi2189 <3302 s"t for D R = I . 0 and 1.5, respectively, while the maximum peak shear rates were 755 < mean=1920 < 3060 sq and 1357
Department of Mnthematie.ad Sciences WeseAmter Polytochair Institute Worceater, MA USA 01600 Arterles with high grade stenmea caused by ntheroedasotin plaque growth m a y coIlnpae under physiological conditions. The compreuion meultiag famn thin collapse m a y lead to accelerated f~tlgue and rupture of the fibrous r which contains she plaque. The plaque cap rupture ce.n lead directly to heart a t t a c h and strokes. The fluid dynamics auocinted with this is not fully undetitood. A nonlinear mathematical model with free moving bounda."y is introduced to study vincous flow in a=inymmetric e]estlc tubes with itenosca. T h e tube law introduced by Shapiro et ILl is adapted to reflect the ehatic property of the tube. An iteratlve nu~eelcni method using boundary iterations, psendocompremibility and ADI technique in developed to zolvr the model. Three k i a d l of gtenoeee with different stiffoeaa vas~tiene are considered. Effects of severity and atiffne~ of the atenoeea, pressure drop conditions, stiffnea of the vessel wall and v h e m l t y o~ the fluid on the flow are studied. Flux curves vs. different parameters, velocity profiles a n d shear stress fields are obtained and studied to understand the fluid dynamics. The results are consistent with the l--d modele and provide more details about the local flow properties. The mode] and numerical method have the potential to be used to ~ u d y blood flow in collapalhle arteries with stenosea.

175

178

ANASTOMOTIC DISTURBED FLOWS AND ENDOTHELIAL CELL FUNCTION IN HYBRID VASCULAR GRAFFS Michael P, Burns and Natac,ha DePzoin. BiomedicalEugiueering Departolent, Rensscinar PolyW.chnicInstitute,Troy, NY 12180.

THE EFFECT OF WALL COMPLIANCE ON SEPARATED ARTERIAL FLOW F.S. Henry Department of Mechanical Engthr162 & Aeronautics. City University. London. UK

Auastomotic intimal hypet~lusia is the ultimate cause of failure of all vascular grafts. Unavoidable compliance mismatch between anificini grafts and host vessels can significantly niter anastomosis geometry resulting in local flow disturbances at the anastomotic sites. We investigated the fluid dynamics in modeLs of end-m-end graft anastomoses. To identify the role of hemedynamice in graft failure we evaluated endothelial cell function in the anastomotic sites of endothelial seeded vascular grafts peffased in vitro. The computational code Nekten was used to solve the Navior-Stokes equations in ~ e n s i o a n l graft geometriesthat considered vasseFgrafl compliance ratios (Cv/Cg) of 3.75, 1.7 and 0,5 at a Reynolds number of 250. The graft and vessel walls were considered rigid. However, the geometry used was based on estimates of the dynamically-adapted dimensions at intravasCniar pressures for normal and h ~ s i v e humans. The results demonstrated areas of flow separation and renircidation on the graft surface downstleam to the proximal anastomosis for Cv/Cgffi0.5 and on the host vessel surface downstream of the distal anastomosis for Cv/Cg=3.75 and 1.7. Endothelial seeded PTFE and Sflustic grafts (4 to 7 mm diameter) of various anastomotic geometries' were perfnsed for 24 hours in a flow loop. Preliminary evidence indicates increased cell proliferation in areas of flow disturhancas with decreased cell retention at flow reaUachment (predicted high shear stress gradients). These results suggest that the complex fluid forces, characteristic of compliance mismatch related flow dismrhaacss, may lxomoto cellular and biochemical events that lead to graft failm'e.

Local regions of separated and recircuinting flow occur naturally in the cardiovascular system; e.g., at arterial bifurcations. Due to the puinatile nature of the flow, the extent of the separated flow region changes cyclically with time, and hence, the area of arterial wall over which the separated flow reanachex is subjected to low and fluctuating shear stress. As these regions appear to correlate to sites of plaque deposition, it has been suggested that spatial and temporal gradients of shear play a role in stherogenesis. This paper addresses the question as to the extent to which the compliance of the arterial wall affects the shear stress dynamics in the rcsnaehmeut area. Specifically, the wall shear stlz~ in the reanachmeut region of a model separated flow is studied with and without wall compliance. The pnisetile flow of a Nevnonian fluid in a compliant tube and the motion of the tube wall were anlved in a coupled manner using the Finite Volume Method. Embedding the wall rolution in the flow solver resulted in an efficient solution scheme for the Coupled equations. The pnsition o f the wall/fluid interface was found using a predictor/corrector scheme. As expected, the reattachment point was found to change with time, and the sepol~ed zone was largest towards the end of the deceleration phase of the flow cosve. However, while differences in the extent of the separated zone and in the values of maximum shear were predicted between the compliant and rigid models, the general features of the two flows were found to be similar. Hence, these initial calculations would suggest that wall compliance has only a seenodaly effect on the dynamics of the reattuchinent region of separated anerini flows.

176

179

AN O RG A N CULTURE MODEL T O INVESTIGATE IIEMODYNAMIC INFLUENCE O N VEIN G R A F T T H R O M B O S I S DA Vosp, D A Severyn, H Kc, EK Dick, MW Webster, SC Muluk Depar~teat of Surgery, University of Pittsburgh, Pittsburgh, PA 15213

STEADY, THREE-DIMENSIONAL FLOW SIMULATION IN A RIGHT ANGLE BIFURCATION: COMPARISON TO EXPERIMENT N. Chesler, H. Loreet, R, Yamagechi t, R. Lect and R. Kamm Department of Mechanical Engineering, Massachusetts Institute of Technology tDepartment of Medicine, Brigham & Women's Hospital. Boston, MA *Department of Mechanical Engineering, Shihaura Institute of Technology. Tokyo, Japan

Early failure of saphenoas vein grafts is attn'bated primarily to thrombnsis. The beating heart induces cyclic distension, longitudinal stretching, and twisting strain on coronary grafts. The~;e hemedyuamlc stimuli may produce n prothrombogeaic response by human saphcoous veins (HSV) that could lead to graft failure. The purpose of this work was to demonstrate 9 new apparatus to invastigate this hypothesis. Using this device, fleshly excised, intact HSV segments were peffcsed ex-vivo while exposed to cyclic twist (+20 ~ and longitudinal stretch (0-7%) or palsatile flow consistent with thnse found In coronary arteries. Three 6-hour illustrative experiments wase performed to measure HSV production of tissue factor ( T ~ a key pmthmmbogeaic factor in the extrinsic euagnistion cascade. The paafusina steatugiea sad 'IF activities measased (expreased i s Factor Xa units generated per 0.113 ema of luminsl surface area) are shown in the table fur beth the specimen end paired control for each experlurent P is picture (mmHg), Q is flow (ad/min). This study indicates that our apparatus allows novel expaKmen~ to be ~ out which may yield new information regarding the effects of binmechadiexl st~atnii (e.g., cyclic wall atrsin) on the thrombogenlcity of vasenlar segments. gneelmen test I f0nU'ol I test 2 control 2 test 3 control 3

Perfuslo~ Conditions P~100, Q f l ~ P-15. Offi15 Pffil00, Q=I00 P=100. Offil00 P=120/80, Q=80 P=15~Q=15

Protoeol steady flow, twist/atretoh @ l H z ateadv flow steady flow, twlsVstrctch @ l H z steady flow :pulsatile @lHz, uo twist/atratch isteady flow '

T] r Activity funlls'J 4.2 xlO ") |,~ x | 0 "s 4.4 xl0" j 1.8 x l 0 "j ~ xl0" ) 2.2 xl0 "3

Numerical simulation of flow in a right angle bifurcation modeled after an aorto-renal bifurcation has been performed to evaluate wall shear stresses and secondary flow characteristics during steady flow. The spectral element solver NEKTON was use~ to perform the simulations. The geometry of the fluid domain was based on the bifurcation built and used in experiments by Yamaguchi, et al, (ASME BED 1993 vol 24:pp383-386) to study the effects of secondary flow on wall shear stress in steady laminar flow. Novel structured grid generation techniques were used to create a three-dimensional scale wireframe mesh of the domain. Simulations were performed at several Reynolds numbers up to Re = U D ~ ' ffi 255 where D ffi 2.4 cm, p = 1,0 g/cm 3, U = mean velocity = 2,0 cm/s and p. = 0.0180 cmVs; boundary conditions are parabolic velocity profile at the inlet and main outlet such that Qm/Qin = 0.78, and outflow condition with P=0 at the side branch outlet, Despite a two-fold difference in Reynolds number, the calculated results agree well with experimental results taken with the electrochemical technique at a Reynolds number of 510 when scaled for the difference in velocity. In beth cases, peak shear stresses occur at the proximal and distal aides of the right angle branch, falling rapidly in beth (side and main) downstream directions. Farther, shear stresses along the proximal wall of the right angle branch exhibit damped oscillations before approaching a steady value consistent with fully developed flow. Support for this project has come from the MIT-Japan Foundation, National Science Foundation and the National Institutes of Health.

S-37

S-38

Modeling and Measurements of Cardiovascular Function II

180

182

NONINVASIVE CIRCULATION P A R A M E T E R ESTIMATION J.L. Palladino I, J.P. Mulie~, F.Wu s, M. More~, T. Kenne~, R.M. Baevsky4, end A. Noordergranf s ITrinlty College, Hartford CT USA., =University of Lenven, Belgium, aUniversit&t Gras, Austria, 41nstitute of Biomedical Problems, Moscow, Russia SUniversity of Pennsylvania, Philadelphia PA USA

Evaluation of the state of the cardtovescolar system via variables versus parameters is contrasted, with the conclusion that the latter are far more attractive than the former. An isovohimicpressure-based description of the leR ventricle as a pump [I] is used in a simple, closed-loop systemic elrenintion model to show that the system variable stroke volume can be normal for a wide range of system conditions, ineludin8 a normal vantrlele and normal circulation, or a weakened ventricle and altered cirrulation. Altecnstively, isovoinmic pressure p,,(t) directly reveals the heart's contrsctile state, and may be embodied as the parameter ventricoinrcompliance, C.(g), computed via OV./Opv. Noninva~ve estimation of circulatory parameters is proposed by combhiing the experimental sceeleration ballistocasdiogram sBCG with a distributed nmthematical model of the arterial system, originally in analog form [2], taki,~ ventrictdas ejection flow as the independent variable. The simu]taneons matching of all available information will narrow the range of possible solutions, yielding characteristic circulation parameters. 1. Mulier, J.P. (1994) Ventrlculas pressure as a function of vohime and flow, Phl) dissertation, Univ. Leuven, Belgium. 2. Westerhof, N., Boama,; F., DeVries, C.J. and Noordergranf, A. (1969) Analog studies of the b,,m-- system/c arterial tree. J. Biomerh. 2:121-143.

181

PRESSURE-AREA CURVE OF THE BRACHIAL ARTERY WITH ALTERATIONS IN SMOOTH MUSCLE TONE

C~ty Orzewisdd, James Rlla, Shawn Reid, and John K-J. U Rutgem Unlversny, Biomedical Eng. Dept., Cardiovascular Res. Labs, Piscetaway, NJ The lransmural pressure - luman area corve (P-A) d the human 10r artery is examined for resting tone and ma,~rnally dilated conditions using ecdustve cuff plethysmography. "lTds data Is usedtoconetn~an'mttwnm'dcalmodel P-Acuwethat accents for smcoth muscle effect A calibrated oeclusive ann cuff is employed to measure the artedal volume pulsations and to vary tr~mmural pressure. Arted~ compliance Is obtained from the retlo of volume puise to preesure puise ampitude e~l Integrat~ to 0btaln the P-A cunm. The measurement ts mpeeted followtn9 a 2 rain.

of blood flow to Induce reactive h~pemn~ end ma.~mal vamdiletlon. Prevlou~y a magmmat/cel model of the P-A oJo/e ban 10eendeveloped m fo41ov~, P = a(exp (h (A- AI~/Ab)- 1)- E ((AgA)n- 1) + PI> Tm,ml,w,dPmeaeo where a, b, At=.E, n and Pb ~ Conetsnts d e t e r ~ ~n=n t~=,~ ~w/ from expedrnantel measurements. P is the ~ J found that at low and negative pre6~ree the P-A"~~ J" oJwes for different tone needy coincide (fig). ~" | Smooth musc~ is apparenlly only effective while wan =o~ I" stretch lumen Is drcolar. ~ | mathematical model is found to represent both the ~=' ro~/n0 and dilated conditions. The primary [ ~terations are in the value of b,with b increasing with e,o.,~ -eo o tone. x =. , ~ m. , = = ~~ . = .,.

m

,oo

(mm

183 E M E R G I N G C A R D I O V A S C U L A R M O D E L I N G STRATEGIES J.L. Palindino z, G.M. Drsewlecki 2 and A. Noordergrasf s XTrinity College, Hartford CT, 2Rutgers University, Piecataway NJ, sUniversity of Pennsylvania, Philadelphia PA

Modeling is a tool used to identify and quantify the mechanhm(s) responsible for observed phenomena in physiology. The strict rules to which the modeling process h subject generally lead to the formniation of new, specific experiments, thereby broadening the pool of experimental data and the modeling database. On occ~fion, this process suggests new areas of experimentatic~t and concepts outs/de the scope of the original problem. Conaequently, successful modeling can dcepan a researcher's i n ~ , h t and breadth of view. A major chaileoge in the bioselences is integration of s large body of molecular level information, with the ultimate goal of onderstandiog organ function. Three examples from the cardiovascu/sr field illustrate novel modeling approaches toward t h h goal. The first, a distributed arterial system model, combines features of two separate, flawed classical methods to interpret a wide range of previously unexplained experimental observations. The second shows how s modal based on muscle structure can predict a molecular mechemsm responsible for muscle's mechanical properties (function). The third prelumts an approach for predicting ventdco]ar hypertrophy (structure) based on cardiac metabolic function. PalJadino J.L., Dr=ewiecki G.M. and Noordergranf A., Mode]rag Strategies in Physiology, Ch. 156 in: The Handbook of B/creed/ca/Enf/neer/ng, J. Bronzino, ed. CRC Press, Boca Raton, (1995) in press.

SENSITIVITY OF MEASURED AND MODEL-DERIVED PARAMETERS FOR ASSESSING MYOCARDIAL ISCHEMIA AND HYPERTENSION. John K-J. 1.~ Ying Zhu, Jia-Juog Wang and Gary Drzcwiecki. Biomedical Engineering, Ratgers University, Piscatsway, NJ 08855-0909. Assessment of cardiovascular function in normal and diseased conditions reties on the alterations of measurable variables and/or derived parameters. Their sensitivity is important in differentiating the severity o f the undedying disease. We examined this aspect in myocardial ischemia (MI) and hypertension (HBP) conditions. Directly measured variables, such as blood pressure, flow, vessel diameter and myocardial segment length and lumped cardiovascular model-dmived compliances, elastances, and resistances were obtained. Results show that, dependent on the specific condition, certain variable and parameter are more sensitive than others. For HBP, pressuredependent arter~l compliance is the most sensitive for describing severity and treatment efficacy;, while for MI, cardiac muscle segment shorten/ng. Although the maximum values of both left ventricle and arterial system elastsnces are important parameters for assessing global function, pressurediameter and pressure-segment length loops are more sensitive to local myocardial and arterial alterations. I'4o single variable can sufficiently describe a disease process. In general, cardiovascular function can be more precisely assessed by a carefully selected combination of variables and parameters.

Cardiac Mechanics

S-39

184

187

BIAXIAL STRAINS IN CULTURED ADULT CARDIAC FIBROBLASTS A.A. Lee, T. Delhaas~ J. Lecng, F.J. Viliarreal, and A.D. McCulloch Departments of Bioengineering end Medicine, University of California-San Diego *Wilhelmina Children's Hospital, Utrecht, The Netherlands

EFFECTS OF PRESSURE OVERLOAD HYPERTROPHY ON PASSIVE STIFFNESS AND VlSCOUS DAMPING IN THE ISOLATED CARDIOCYTE. Michael R. Zile. lohn M. Bucklev. Ken E. Richardson. George Cooper. IV~ Medical University of SC, Veterans Adirdnistration Medical Center, Cazes Cardiac Research Institute, Charleston, SC 29425.

Stress or strain has been postulated to be a major determinant in the remodeling process of the cardiac extraselluiar maldx (ECM), which is pdmadly regulated by cardiac fibmblasts. Because wall deformation in the heart is multiaxial, it is important to determine whether mechanical regulation of cardiac cell function may depend on different magnitudes and pettems of strain. To examine which mechanical parameters (e.g., tensile or compressive strains, area changes or shears) may be the primary determinants of cellular responses to stretch, we have combined the use of uniaxinl and equiblaxial cell stretchers for systematic mechanical testing in cardiac sell cultures. In both systems, fluorescent microspheres were attached to the silicone substrata end the cells for strain analysis. In our unialdal studies of adult rat cardiac fibrobiasts, representative longitudinal and transverse strains in the substrate and cells were El1=0.11 and E22=-0.04 with little average in-plane shear, E~2. We observed that different magnitudes of strain dudng uniaxial stretch may induce different responses in ECM gene expression and the release of autocrine growth factors. We have developed an equibiaxial sell stretch apparatus to further examine the dependence of such responses on specific strain parameters. Strain analysis of this stretch system confirmed that the principal strains (El=E2) were equally distributed throughout the surface of the silicone membrane and the ceils at different magnitudes of membrane stretch. The strain magnitudes were calibrated to the number of turns of the circular equibiaxial stretch apparatus to allow for reproducible application of strain to thu cell cultures. Quantitative characterization of the fundamental effects of mechanical strain and its mechanism of action in cardiac fibroblasts may contribute to the understanding of the process of ECM remodeling in the head.

One potential mechanism causing diastolic dysfunction during myocardial hypertrophy is an increase in the stilfnese uf the cardiocyte itself. We examined 2 components of cardiocyte stiffness: passive spring (Kps) constant and viscous damping (Cvd) constant. Cardiocytes from normal cats (control) and cats with right ventricular hypertrophy (RVH), Indoced by pulmonary artery banding, were embedded in a 2% agarose gel..Cardiocytes were then subjected to a stepwise change in force (stress,o) and the resultant changes in cell length (strain, e) were measured. Changes in Kps were determined by .exa.mining the slope and position of the o-e relation during an increase in o apphed at a constant rate. Changes in Cvd were determined by measuring the area between the a-c relation obtained during increase and decrease In force. This loop area reflects the amount of mechanical energy converted to heat onergy by damping. The slope and intercept of the o vs e relation obtained during an increase in o were similar in hypertrophied and control cardiocytes (y-----441+ l.le + ~ x , r=0.99 for control and y=-678 + 1.2e + 05x r=0.99 for hypertrephted cardiocytes). However, the loop area between the o-e relation obtained during an increase and decrease in force was significantly greater in hypertrophied (102.+.3) compared to control (83:/:4) cardiocytes (p<0.0S). Thus, hypertrophy did not alter the spring constant but did increase the damping constant. We hypothesized that this increase in Cvd was caused by an increase in microtubnie density. When microtubule density was decreased by treating hypertrophied cardiocytes with colchicine, the loop area (Cvd) decreased to normal. Thus, the increased damping in hypertrophied cardiocytes was caused, at least in part, by an increase in cardiocyte microtubnies.

185

188

INTERACTIONS BETWEEN M E C H A N I C A L FORCES A N D T H E EXTRACELLULAR MATRIX I N THE REGULATION O F C A R D I A C P H E N O T Y P E IN VITRO. DG Simpson, L Terracio & T K Borg. University o f South Carolina School of Medicine, Col. SC. 29208

VISCOSITY OF PRESSURE HYPERTROPHIED CARDIAC MUSCLE CELLS Hirofumi Tagawa, Ning Wang, Donald E. ingber, and George Cooper, IV Gazes Institute & VAMC, Mad Univ of SC, Charleston, SC, & Harvard School of Public Health, Children's Hospital & Harvard Medical School, Boston, MA We have shown that increased microtubule (MT) density causes cardiocyte contractile dysfunction in pressure overload cardiac hypertrophy. To determine how, in physical terms, this increased MT density overloads the contractile apparatus at the cardiocyte level, we measured cytoskeletal stiffness and viscosity in isolated cardiocytes via magnetic twisting cytometry, a technique by which magnetically induced force is applied directly to the cytoskeleton through integrin-coupied ferromagnetic beads coated with RGD peptlde. Measurements were made in two groups of cardiocytes from pulmonary artery banded cats: 1) those from the pressure overloaded right ventricle (RV), and 2) those from the normally loaded same-animal control left ventricle (LV). Stlfiness (dyne/cm 2) increased from 5.15 + 0.33 in the normally loaded LV cardiocytes to 8.90 + 0.39 in the hypertrophied RV cardiocytes (T 73%). Viscosity (poise) increased from 19.1 + 4.4 in the normally loaded LV cardiocytes to 42.0 + 8.3 in the hypertrophied RV cardlocytes (T 120%). In addition to these baseline data showing differing stiffnesses and especially viscosities in the two groups of cardiocytes, MT depolymerization by colchicine was found to return both the stiffness and viscosity of the pressure overload hypertrophied RV cells to normal. Conversely, MT hyperpolymerizatiun by taxol increased the stiffness and viscosity values of normally loaded LV cardiocytes to the abnormal values given above for pressure hypertrophied RV cardiocytes. Thus, increased MT density constitutes primarily a viscous load on the cardiocyte contractile apparatus in pressure overload cardiac hypertrophy.

The metabolic response o f a neonatal heart cell (NHC) to an episode o f static stretch in vitro appears to be highly dependant upon the orientation of its myofibrils (MFs) with respect to that strain. A thin film o f type I collagen prepared with a def'med orientation will promote NHCs to spread, as a population, along a common axis that is in parallel with the gel lattice. The individual NHCs of these cultures display a rod-like cell shape and polarized arrays of MFs. Altering the tertiary structure of the gel or blocking the ftmction of the alpha I beta 1 integrin inhibits the evolution o f ilm tissue-like pattern of organization. Applying a sustained static strain in parallel with the long axis of these NHCs (increased eccentric load) appears to have little, or no effect on M F structure or overall rate of total contractile protein degradation. In contrast, stretching myo~ytes that are aligned perpendicular to the axis of strain

(increased concentric load) altersmyofibriliarorganization and has a marked effect on suppressing total contractile protein degradation. Randomly arrayed cells display a response that is intermediate to these effects. These data suggest that mechanical forces and the composition of the surrounding extracellular matrix can interact to regulate cardiac protein metabolism, myofibrillar structure and cell shape. Supported by N I H H L 42249, H L 37669 & South Carolina Heart Fellowship to D G Simpson.

186 INTEGRIN-MEDIATED ADHESION STRENGTH O F TRANSFECTED FIBROBLAST MUTANTS Daniel A. Hammer, Kelly A. Ward, and Jun-Liu Guan* School of Chemical Engineering and *Department o f Veterinary Pathology, CoreeU University, Ithaca, N Y 14853 Inleknlns constitute a major class of cell sm'face molecules involved in receptorhcterodlmc~ (a[3) glycoproteinscolocalize with ECM proteins and cy;oskaletel molecules in focal contacts - discrete mceptor clusters - in the cell-sobsu'ate contact reglon. We have hypothesized that this om'tceauadon of receptam significantly increases adhexion s~regth, end depends on the ability of the ~ to bind both F.Clvland cytoshakton OVmd and Hammer, 1993, Bienhvs. 1. f~936-959). To test this hypulhexls, we uss 3'r3 fibmblasts that have beea mmsfected with different fomts of the avian 131integrin receptor. We have meastned the edhedon streagth of thexe coil line~ ta ssbsUates roared with eithcr llbroseclin or on arsthody against the exuacellular domain ofthaavian [~l iulegrin mcepter (CSAT).Adhaskm sU'engthofnonnslmonss 3T3catison fibronectin is in agree.meatwith IHevicusmpom. For transfectod cells, adhesion meagth hr.resse8 with CSAT sob~lrato coacestrat/on, but is always less than seea far wild type colls on fibrmecfin. F ~ e t ' e , iacre~ng the size of the mmcatinn of the cytoplasmic domain in geaerel deereases adhesion suength, althongh A2. a mmcation that leaves only a dRmsicRR residue, exhibits greater adhe~iea tium A4, whlch has both the dihasic RR rexidue as well as one Y resides implicated as a slte of phesplmfylatlon. Flow cytomelry cc*'d'mnsthat adhesion eaeagth differences are not due to diffemnc~ in receptor expression. I m m m a o ~ t confocal microscopy shows adhesion s~atgth is most closely associated with phcsphorylailon in the.se mutants, not receptor clttstering, refuting our hYlx)thssir We have developed a mathematical model to expisin the msoha of adhasion strength m ~ with rsspect to heterogeae/ty of ~ polmledon, binding inopenies of the receptor, and tim exteat of spreading stimulated by the different receptor a-assfectants.

mediatedadhssim to extracollular matrix (ECM). ~

S-40

Poster

Presentations

189

192

CONTINUUM MODEL FOR CONVECTION AND DIFFUSION OF A TRACER 1N A THREE DIMENSIONALTISSUE STRUCTURE Daniel A. Beard and James B. Bassingthwaighte Center for Bioengineering. University of Washington. Seattle, WA 98105

AORTIC IMPEDANCE ALTERATION ON THE RESPONSE OF SINGLE SEGMENT INTRA-AORTIC BALLOON PUMPING: A MODEL-BASED ANALYSIS Jiwang Dai and John K-J. Li. Cardiovascular Resea/ch Lab. Biomedical Engincering. Rutgers University. Pif~naway, NJ 08855-0909.

A new approach to mathematically modeling myocardial tracer kinetics is developed and used to study the interaction of neighboring micrevessels and the effects of diffusional shunting on the washout of highly-diffusible substances. A tissue volume of I mm2t is modelled ~s a three dimensional arrangement of 16x 16• 16 elements. Each element consists of one conveeting (vascular) region and one or more nonconvecting regions. Convection is governed by a three-dlmensional velocity distribution generated by solving a continuity equation for arterial soorces and venous sinks randomly distributed in a randomly inhomogeneous and anisotropic conducting media representing the 4096 vascular regions. Dispersion of the tracer proceeds by simultaneous convection in the vascular regions, first order transport between regions within elements, and diffusion across elements. The model is similar to a compartmental model consisting of 4096 vascular compartments and 4096 tissue compartments, distributed in three dimensions. This is a simplified approach to modeling more tortuous microvascular arragements than classical Krogh cylinder-type models which are generally distributed in only one dimension. Complex flow geometries are verified based on knowledge of the microvascular anatomy and the fractal correlation of spatial flows. Simulation of the washout of a diffusive tracer from the microvasculature demonstrates that diffusional shunting may be important for tracers with effective tissue diffusion coefficients greater than approximately 10":~cm2/s. (This research was funded by NIH grants HL50238 and RRI243.)

The bcncficial effects of intra-aortic balloon pumping (IABP) have been reported to be variably influenced by aortic impedance To im,estigato this aspocl, we developed a distributed parameter medct representing the effects of cardiac assistance by IABP. Asocnding aortic blood p~asme (Pa) and flow (Qa) ~ere inltially calculated from the rondel. when balloon's inflation was set at tbe end ofsy~ole and deflation at 40 ms before the end of diastole (left figure). With inflation and deflation time unchanged, the peak value of diastolic pressure (Pdmax) was changed by varying the aortic compliance (Ca) or characteristic resistance (Zo) (right figure). Decrease of Ca or Zo tmused the increase of Pdmax and increase of Ca or Zo caused the decrease of Pdmax. Since Ca is the ratio of the blood volume change in the asecoding aorta (dV) to change in Pa, the increase ofdV induced by IABP causes a greater increase of Pa in the aorta with a smaller Ca. Smaller 7,o causee lower resistance to the back flow induced by IABP, thus also increases Pdmax. In addition, ~ , ,~. . ~ L ~ ~) ~ the change of Pdmax was

I

0 .50 ~ 0

I morese ,ti to

Qz 0

TimeO'S(t)

9

1

-

,oo

80 0

"'"-':~-- :~:l "I 9

Ca(~05

induced diastolic pressure augmentation is greater in a stiffer aorta and Ir in an aorta with higher resistance.

193

190 MONITORING HEMOGLOBINCONCENTRATION CHANGESAND OXYGENCONSUMPTIONIN SKELETALMUSCLE USING NEAR-INFRAREDSPECTROSCOPY

Paul B. Benni MS, "BoC ~ MS, 'DavidAreoffMD,PhD,and John K-J.U PhD. Dept.of BiomedcalEngineedng,Ridges Urt~ty, and 'DepLof/~eslhesla,RdoeaWoodJohnson Medicalg c ~ , NewBn.mwick,NewJenmy08903,USA. Near4nfraredspectrou:op,/(NIRS)pa~ides s new technologyto contJnucosl,/andnoMnvasbely monitor changes in tissue ox'/genaSon. NIRS is based on Iba absoq~on d~nc~isfics o~ the clwmophuesox,/hem~n (Ht~) andd e o ~ l ~ n (Hb)In the ne~'..~ra~m~ (To0.1ooonm). Usings NIRSsyslemof our desk, we monitoredleg skeletalmuscle~ l i o n in heallhyvoluntee~ befog, dudi~, end afterm'ledalocclusim st 220 mmHgby n ~ t placed on ~ ~gh. RMalive ~obln and deoxyhemogloblnconcenLm6on(tti-.~3~ Al"lb)cJ~mgeswee celcelatedby u s ~ e mod~d Beer-l.end3ndlaw. V~ was calculatedvla the line~ Icetemenlof Hb in mmde ~er ceduslon: d(~l)~t ~ 0.O2O[ V02= '1.63idA/n'i~1OOg,l Tba I~xe ~ o t'~.~ petlem O~ AHb, AHb02. at~l AIt~.AI-~02 Release ~ "" medal ocdu,~on. Vc2 was Shoed so utueied

I'

'"

to be 1.e3 + 0.34 -~ o.oto

~r~ m e~e de.~ty of musde. "~ resulissbaw ihatourNIflSs?slemiss iwa'r~r~ monitor o~ ~eleta rr,t ~ e ortTmatlon dud~ ~ t gOieoew.t

~ ~ 0.000 ~,, g -0.fll0

121 AH~z~-,,,..~

-

~rne (seconds)

191 DIRECT OBSERVATION OF NITRIC OXIDE AND BLOOD FLOW O S C I L L A T I O N S ( V A S O M O T I O N ) I N C A T O P T I C N E R V E HEAD. D.G. Buerk, C.E. Rive*, and S.D. Cranstoun*. Depts. Physiology, Bioengineering and *Ophthalmology, Univ. of Pennsylvania School of Medicine, Philadelphia,PA. Nitric oxide (NO) was measured simultaneously with infra-red laser Doppler flowmetry in the optic nerve head ofisoflurane anesthetized cat eyes (n ffi 12). Spatially detailed etectrnchendcal measurements were made with Nation polymer coated recessed gold microalecU-odas (tips - 5 Fro) held at the oxidation potential for NO (+800 mV). Both N O and blood flow increased with greater neuronal aetivity(flicke~fing light stimuli, 15 to 2 0 H z for 1 to 2 min). N O nsually inereased to its maximum before the blood flow did. Both responses were attenuated aRer intravenous administration o f NO symlmso inhibitor (L-t~tro arginine methyl ester, 60 to 100 ms, n = 5). In many preparations, oscillations in N O and blood flow were observed With similar frequencies (5 to 7 per rain) during resting (darkadapted) conditions, as well as during flickering fight stimuli.At some locations, the blood flow lagged behind the increase in NO. Although N O oscillations were attenuated after NO synthnse inhibition, blood flow continued to vary, usually at slightly slower frequencies. These data confirm that N O is important for vnsodi]ation with increased neuronal activity,and that NO probably has a minor role in modulating vasomofion. (Supported by EYO9269 from NIH).

REGIONAL PERFUSION OF L U N G AND MUSCLE TISSUE IN RESPONSE T O CHRONIC HEAT F L U X C. Davies, H. Harasaki, K. Fukamachi, F. Fukumura, G. Saidel" Department o f Biomedicel Engineering Cleveland Clinic Foundation and "Department of Biomedical Engineering Case Western Reserve University A totally implantable artificial heart (TAIl) that produces heat will affect the surrounding lung and muscle tissues. Evaluation o f these tissues after chronic exposure to supranormal temperatures is required to provide safety specifications before a T A l l can be implanted clinically. Characterizing the adaptation o f tissues to chronic heating includes the evaluation ofperfusion as a function o f heat flux. To simulate heating by a TAIL experiments were conducted with heated disks implanted adjacent to the left lung and latissimus dorsi mnsele o f nine calves. Identical, unheated disks were implanted contralaterally as controls. For seven weeks, a constant flux o f 0.04, 0.06, or 0.08W/cm 2 emanated from the heated disks. To evaluate perfusion, colored microspheres were injected into the systemic and pulmonary circulation o f calves prior to utcrifice. After sacrifice, tissue samples were harvested for the lung and muscle tissue adjacent to the heated and unheated disks. The tissue capsule and underlying tissue were separated and digested to allow counting ofspberes. The number of microspheres in a tissue region is proportional to the regional perfuslon. In tissue capsules and underlying lung and muscle, preliminary data show an increase in perfusion relative to controls with a maximum at 0.06W/era 2.

194 A DIFFUSION WAKE MODEL FOR TRACER ULTRASTRUCTURF~ PERMEABILITY STUDIES IN M I C R O ~ B. M. 1~ t, F. E. C u r r f and S. Weinlmum I tDept, of Mechanical Engineering, The City College of the City University of New York, New York, NY 10031 2Dept. of Human Physiology, School of Medicine, University of California at Davis, Davis, CA 95616 We developed s time dependent diffusion model for analy~fingthe eoneeutration pmftlec of low molecular weight tracers in the interendothellal tiers of the capillary wall, which takes intowxoont the three-dimonsiomd, time dependent filling of the sncroonding tissue space. The model provides a critical connecting link between two methods to investigate traasvasctflar exchange: electron microscopic experiments to study the time d e ~ t wake formed by low" molecular weight tracers (such as lanthanum nitrate) on the tissue side of tbe jtmctlon atlmad discominuitiesia the inter-endothelial cleft of frog meseatery capiIlari~ (Adamum end Michel, Journal of Physiology (London) 466:303-327) and confoeal mict-mcopoexperismats to memure the spre~l of low molncular weight f l ~ t tracers ia the tissue tpa,:e ~fmund~g these miemwaeais (Adereson'at el., Miemcireulation: Vol.l, 4:251-265). We show that the inteqpratafioo of the preaemce of trar,er as en idl-or-~aone indication of a pathway aem~ the jtmetlonal strand is likely to be i~eorrect for *mall taluses. Large pore pathways in which the local tracer flux densities are high resell a threshold con~mtratien for detee~on sad are likely to be detoctndarea"relativelyshortperfoslon times whereas distributed mall pore pathways may not be detected tmtil the tissue concentrations tmrvmmdingtile entire vessel tgpmaeli threshold conceatratioes. The analysis using this approach SUplmrtsthe hypothesis ~vmeed in Fu et ILl. (Journal of Biomedieal Engineering: 116:502-513) that the principal pathways for water end solutes < 1.0 nm diameter across the intertmdotbelial deft may be differ~t and suSgeats new experiments to te*t this hypotheeia.

Poster Presentations 195

S-41

198

ANALYTICAL MODEL OF AORTOMYOPLASTY: EFFECTS OF SKELETAL MUSCLE CONTRACTILE DYNAMICS David T. George, Jayson W. Bauman, Alexander S. Geha, Brian L. Cmolik Division of Cardiothorecic Surgery, School of Medicine, Case Western Reserve University, Cleveland, Ohio. Aortomyoplasty is a new surgical treaUnent for chronic heart faiinre in which skeletal muscle is wrapped around the aorta and electrically s6mulated to provide external diastolic counteqtulsation. The principle of aortomyopinsty is similar to the intra-aortic balloon pump (IABP); however, aortomyoplasty does not require an external power source and could be used long-term. Muscle contractile dynamics differ from IABP dynamics and the muscle changes from a predomlauatly fast phenotype (type I]) to a slower type 1, as it adapts to its increased contraction rate. To investigate the effects of mascle dynamics on the timing and dumtinn of muscle contraction in anrtamyoplasty, we used o lumped-parameter model of the cardiovascular system. The heart was modeled as a time-varying elastance while the passive elements of the cardiovascular system were modeled as resistances and capacitances. We modeled a muscle contracting around the aorta by manipulating aortic resistance and compliance within the cardiac cycle. Muscle ceat~ction and relaxation were modeled as filst order and we varied the time constant to manipulate muscle speed. Fast and slow muscles produced similar increases in mean aortic pre'~ure and coronary blood flow and decreases in peak lefl ventricular pressure and stroke work - suggesting the magnitude of diastolic augmentation and aflerlond reduction does not depend strongly upon muscle speed. Stimulus paradigms producing optimal effects differed, however. The slower muscle required activation earlier in the cardiac cycle and its stimulus duration needed to be shorter. Also, the extremum were narrower for the slow muscle, suggesting that greater care must be take when choosing stimulus parameters - relatively small increases in heart rate could result in increased aflerload. This model suggests that muscle contractile dynamics are important in aortomyoplasty.

A NONLINEAR MODEL FOR MYOGENIC REGULATION OF BOOD FLOW TO BONE: EQUILIBRIUM STATES AND STABILITY CHARACTERISTICS Timothy P. Han'igan Department of Orthopedic Surgery, University of Texas Medical School at Houston A simple compartmental model for myogenic regulation of interstitial pressure in bone is developed, and the interaction between changes in interstitial pressure and changes in arterial and venous resisUmces is studied. The arterial resistance is modelled by a myogenic model which depends on lransmurel pressure, and the venous resistance is modelled using a vascular waterfall. Two series capacitances model blood storage in the vascular system and interstitial fluid storage in the extravascolar space. The static behavior of this model mimics the observation that vasodilators do not work well for bone, and it shows that interstitial pressure increases substantially if capillary permeability is increased.The dynamic model shows unstable behavior at small values of bony capacitance, and at high enough myogenic gain. At low myogenic gain, only a single equilibrium state is present, but at high enough myogenic gain, two new equilibrium states appear. At further incereases in gain, one of the two new states merges with and then separates from the original state, and the original state becomes a saddle point. The appearance of the new stales and the transition of the original stato to a saddle point does not depend on the bony capacitance, and these results are relevant to general fluid compartments. Numerical integration of the rate equations con finns the stability calculations and also shows limit cycling behavior in several situations. The relevance of this model to circulation in bone and to other comparanents is discussed.

199

196 SELECTIVE STIMULATION OF SKELETAL MUSCLE FOR CARDIAC ASSIST Dirk R. Thompson, John J. Michale, Erlk A. Cheerer', David T. George Division of Cardiothoracic Surgery, School of Medicine, Case Western Reserve University, Cleveland, OH, *Department of Engineering, Swarthmore College, Swarthmore, PA In skeletal muscle cardiac assist, the contractile power nf skeletal muscle is used to augment function of a failed heart. In present clinical practice, a pair of intramuscular electrodes is woven through the latisalmus dorsi muscle (LDM) in the region of its main oerve. This electrode configuration effects contraction of the entire LDM for each electrical stimulation. However, because there are three separate anatomical segments to the LDM, each with distinct innervatinn, we can selectively activate regions of the muscle. Selective stimulation opens up a wide range of possible spatial contraction patterns, including peristaltic-like motion, not achievable with the electrode system now used. in seven studies, we measured the regional EMG activity of an in-sire canine LDM while activating the individual muscle segments. We stimulated each of the three nerve branches individually, while recording EMG signals

A FIBER MATRIX MODEL FOR T H E FILTRATION THROUGH FENESTRAL

P O R E S IN A C O M P R E S S m L E ARTERIAL IN'HMA Yaqi F i n a l , David Rumsshit.,ki:, She China3, and Sheldon Weinbeum t IDepanment of Mechanical Engim~ring, 2Depattmem of Chemical En "gilleering, The City College of The City University of New Yosk, New Ycek, NY 10031; 3Instltuta for Biomedical Engineering, and Departments of AMES-Bineoginenring and Medicine, Unive~'sityof Collfcrnia,San Diego, La JolinC A 92093

0. ....."-iiIJl[[k/lll 1

This paper adveaces a new hypothesla to explain the changea in hydraulic condoctivity of an intact arteay wall with i n c ~ u ~ g mmsmural pressme that Tedgui and Lever (1984) and Baldwin and Wilson (1993) have observed experimemally. This h . ~ d s ~ that the compactian due to ptussure loading of the proteoglycan mattix m the artegial intima near fonestsal pores of the inUnmalalastic lamina leads to a nmmwing of the i~re enmmce area mul a large decreaso in the local intrinalc Darcy psrmeabilRy of tire matrix. To quantitalive/y asseus the feasibility nf this mechanism, a local two-dimeamoonl model is proposed to t~d.y the filtration flow in the vicinity of the feaesWal pines in a compresmble intima. Using a heterog..coons fiber amt~. theory, which includes pmtonglycan and collagea componeats, we first predict the cinmge te Darcy permeability with intimal thickness I~. The local velocity and Im~ssure cEstributions in the intima and medin are then caleaintod from the model ts show that there is a marked ann-llnear steepening ~ the infimal Wusstwe Wofiles near the feaustral pores wben the imima thins at higher lumen i~essme~ The predicted relative change in the ~ of tha intima (with the IEL), P~, and the merlin, R=. as a function of intimal thielmuss shows tbst these is a steep inesease in Rt/R= wben Lt is less than 20 pe~eat of its anstre~od value. The numerical results ~ l y support the qeantitafive feus~billty of onr hypothes~ and.also provide a rational explanation as to why the d e ~ arle~ does not exbibtt this ~ dep~deat t-dirmion babavi~'. They also predict how diffesent transmural pressurm alter the growth of an intimal HRP spot that derives from a localized (single cell boundary) endothelini leakage. Such a prediction is amenable to experimeotal verificatioo_

197

2O0

oblique, and transverse muscle segments, in | each study, we successfully achieved selective activation of the canine LDM. In fu~her studies, our laboratory is now applying 1[ selective stimulation of the LDM in skeletal muscle cardiac assist. This work was supported by a grant from the Whitaker Foundation (DTG).

'

"

"

"

"L~"

"

"

"

"

ENDOTHELIAL DERIVED NITRIC OXIDE INHIBITS CVTOMEGALOVIRUS REPLICATION

Computer Simulation of the Penn State Total Artificial Heart N.C. Kelley. R,P. Gaumond, J.F. Gardner, A.J. Snyder. D.B. Geselowitz Bioengineering Program, the Pennsylvania State University, University Park, PA 16802

K~ J. Gooch I and J. A. Frangas~ 1Department of Chemical Engineering. The Pennsylvania State University, University Park, PA 16802 and 2Department of Bioengineering, University of California - San Diego, La Jolla, CA 92093. Ather~clerests, manifesting itself as coronary heart disease and stroke, is the most common cause of death in industrialized countries. One controversial theory, the virns-indoced model of atheruscleresis, attributes the origin and development of this disease to infection of arteries by cytamegalovirns (CMV) or a related virus. CMV infection, like atheruscler~is preferentially affects portions of arteries exposed to relatively low fluid shear stress. The reasons for these locniizetious are unknown. Relatively large concantratioes of nitric oxide (NO) produced by macrophages inhibit viral repficatian, perhaps by inhibiting DNA synthesis in the host cell. We have shown that smaller concentrations of NO derived from unstimniated endothelial cells (EC) also inhibit viral replication in a dose dependent manner. The nitric oxide synthase inhibitor L-n-amino-arginine significantly reduced this inhibition. Since fluid flow stimulates EC to increase production of NO approximately ten fold, we fiypothetize that increased shear stress prevents local formation of atheresclerotic lesions, in part, by elevating the production of the antiviral compound nitric oxide. PreliminaW data have shown that fluid flow inhibits viral infection in one cell type.

A comprehensive mathematical model of the Penn State Total Artificial Heart CI'AH) coupled to a mock circulatory loop has been developed. This model characterizes various pressures and flows throughout the systemic and pulmonary circulations, pressures and volumes within the prosthetic ventricles, and the velocity and position of the pusberplatus which produce blood flow by compression of the ventricles. Nonlinear elements, including pressure-flow behavior of the inlet and outlet valves, as well as ventrleular compliance, have also been included to belier characterize the topography of the system. A simulation which provides velocity control of the pushetplate through an adaptive mechanism, as well as overall control of the cardiac output of the TAH, has been implemented in conjunction with the model. The controller relies on a simple mechanical linkage model of the TAH motor, which provides information on the degree of ventricular fillit~g, as well as an estimate of the mean arterial pressure. The relative left and right pump filling times are periodically adjusted so that the left pump fills completely (preload control) and so that the mean arterial pressure falls within a prescribed range (afterlond control). This simulation has proved useful in evaluating the control system's response to step changes in physiological parameters. The simulation has also been used in an attempt to quantify the effects of parameter mismatch between the motor and the controller. This work was supported by National Heart, Lung, and Blood Institute Contract No. N01-HV-38130.

S-42

Poster Presentations 204

201 ASSESSMENT

O F RETINAL B L O O D F L O W C H A N G E S

USING RES[DENCE T I M E

DISTRIBUTION FUNCTIONS ET Lee, PG Rehko~, JW WamickJ, KN NitowskJ, TR Friberg, ON Finegold, EG Cape Children's Hospital and Eye and Ear Institute, University of Pittsburgh, PA purpose: Progressive retinal disease is often reflected by changes in retinal blood flow. We have developed and verilied a new method to assess changes in retinal blood flow using residence time distribution functions (RTD). Composite RTDs (CRTDs) are constructed by combining information from multiple imaged vessels to descdbe the global flow of blood in the retina. These CRTDs are constructed by utilizing microdensitometric techniques in conjunction with scanning laser ophthalmoscopic: (SLO) data and Image processing techniques. From the CRTDs a value for the mean', circulation lime (MC~I) is cak~lated, which is the average time it takes the blood to pass through the retinal microvasculature. This study addressed the hypothesis that precise! combinations of RTDs provide a quantitative marker of MCT ehangas. Changes In. blood flow induced by O2 breathing were used to test this approach in humans. Methods: Sixteen normal volunteers (8 Female, 8 Male) were studied before and after 0 2 breathing. After injection of dye via the antecubital vein, SLO/IP obtained the relative fluorescence In all vessel branches (64 frames at 2/see). The digitized data were then normalrzed using the RTD method. Vessel diameters were measured from fundus photos. CRTDs were oonstrocted for both air and 02 breathing by weighting each RTD according to vessel size. Results: In response to 02 breathing, increased MCT representing decreased flow was found in all 14 subjects (2 excluded dua to eccentrio anatomy) and averaged from 6.1% to 63.1%. ~oncluslons: CRTDs take full advantage of the data available from SLO/IP to provide an evaluation of retinal blood flow. Evaluation of global retinal blood flow can only be obtained from multiple vessel studies. This MCT is more representative of overall retinal blood flow than those extrapolated from single vessels and may be used as a means of detecting retinal blood flow changes at varying stages of disease.

EFFECTS OF THORACOTOMY AND PER/CARDIOTOMY ON MYOCARDIAL BLOOD FLOW HETEROGENEITY IN THE ANESTHETIZED DOG Sandra Maszian, Christian W, Rascbe*, Andsens Dcessen Znutrum for Physiologic mad Zentmm filr*Anaosthesiologie, Heimich-ndine-UniversitSt D~sddorf, FRG. Regional myocardial blood flow (RMBF) is heterogeneous. The present exl~diments were undertaken to assess the effects of various experimental conditions on local myocardial blood flow ia 6 anesthetized dogs. Differentlyisbr tracer mi~heres (15+3 (SD) diameter, 256 samples per heart, average mass 0,34 + 0.02 g) were injected under copAitious of closed chest (control),open chest and open pericardium with time intenmis ofonu hour. Under control onnditious, mean h e a r rate was l l S _+ 34 bpm, mean aortic pressure 129 _+ 18 mmHg, left ventr/enlar systolic p ~ 148 +_ 11 mmH& pressure in the pulmoan~ artery wus 14+ I mmH,g, mean RMBF of the left and right ventricle were 1.83 +_ 0.50 and 1.34 + 0.42 ral min-~g"s. Relative diapersiou (RD) of RMBF was 0.27 and 0.32 for the left and fight ventricle, reapacdvcly. Simultaneously injected trecer micraspberes gave a correlation coefficient for RMBF of ru = 0,87 + 0,03 (n ffi 6). Thoraontomy had no significant effect on global hemndyunimc parameters and the RD. Correlation b e t w ~ RMBF before and after thoracotomywas 0.85 _+0.03 (tLS,versusr~) and 0.79 + 0.11 (rtS. versus rM) for lefr and right venlricie. Pericardiotomy slightly changed the RD of left and right ventrionlar RMBF (0,31 and 0.25, reapnetively). Correlation of RMBF before and after poricarthotomy was 0.64 _+ 0.26 (p = 0.048 versus rM) for left and 0,29 + 0,18 (p ffi ll.00002 versus rx~) fur fight vanlricin. ~ resolts demous~te: 1) The degree of myocardial blood flow hoteregnneity is similar under closed cl~-'t az,,dopenchestconditi0us, 2) While thoraentomyand pericardin~omyhavemoderato effects on left ventrlcular RMBF, they induce a marked flow redistri~tioa in the right vantrionlur wall.

202

2O5

STEP-WISE HYDROTHERMALISOMETRICTENSION (HIT) TESTING FOR THERMOMECHANICALANALYSISOF MATURING COLLAGENOUS TISSUES J.M. L~, W.A. Naimark. S.D. Waldman,R.J.Andenon, 0, S ~ulti, CA- Ik-reira Centre for Binmxterials.Univer=iWof Toronto. Canada

CHARACTERIZATION OF WAVE REFLECTING SITES OF THE ARTERIAL SYSTEM E Pythnud, N. Stergiopulos, J.-J. Moisten" Biomedical Engineering Laboratory, Swiss Federal institote of Technology, Lausanne, Switzerland

Crmdinklng of collagen moleculel and fibl~l during maturation incJr strength and r to transmission of tcmile forces. While cor~latlon= have been e=tablhhed between mechanicalbel~viour and change= in crotdinking (asasteraedby chemistry), thi* approach doe= not euablith db~,tnt lin~. HIT besu tin: isometricallyconstrained *pr162 to study the rehtionthip betwen load and temperature. ~qben =u~cient tbermal enerl~, is available,the collagen triple helix denaml~ into a morn random, globular form (a rubber). At higher temperature=, thermally labile c ~ n k s are bmkn. Finali~ under u.utained temperatur~ (imthenm) of 80 m 100 "C. m~.t rel=ctfiondrOUthdue to hydmlyfis of pepri& bends within the denaturedcollag~n----nowrei~fo~'d only by themtally stable cm~linhing. The adazafion me, typicallycxp~aed a the hell-time of a Maxwdl-raodel decay, is a direct me=muteof the stable, mechani~lly important erm=fiahing in the m[lagenom maxxix. In a mmputet-mntmlled, mulri-=traple~r *Tttern, ~e have inedited d~ HIT ~dmique m employ #tep-wi~:changez ia =mperaturr With multiple isetbe~m. Anhenim-typr activationenet'gicz can be calculated from tingle uwaple=(pmviomly HIT reacton muldple utmpka were required), and finer dltcriminarion of wamition temperamt~ b obtained. We have taed this tedmique to extra. ine cronlinking of pcricazdlafrom 119 and I~0 day fetal (wbe~ tx~mis 145 dayiFSUmon)and21 day neonatal lsmbe. Steps of 5 ~ from 60-90 *C were applied to Jampkn which had been load-equib ibeated OVer16-1g hrl Of ve]a~'~n with data mile=ion &hyed to hog=reeffect=of heating h~tor F Denaturation temperature* were tlgnilica~riy lower in step-wi=e HIT un• than in mndard HIT tern for both age group= (e.g. fetal: 68.6 e 0.3 ~ersus 66.8 9 0.1 *C). Thi= likely a~qem a lag betwem structtaal transitions and solution temperatu~ during mote rapid HIT heating. Activationenergy cab cuhted from step-wise HIT test= (23.4 t 1.6 koal/rnol) wa= dmilar to literatme valuesoi"27 kcal/mol measured aom~ speci~ and age groups [Ix Lous M. et al. (1985) C~nn Tm Re~13:145].Step-wise HIT tests offer a more icnslti~ probe of changing srzuctme/func~on w.]afons during maturation, and reducethe mtmherof tezt =Fr162 reduc.~g the effecetof in~er-anlmaiwariahillty.

Pressure and flow waves eraitted by the pumping heart am rnflacteA at different locations of the a.neriul wee and mmm to the ascending aorta, influencing the total heart load. Different methods have been proposed for characterizing the location and intensity of arterial refinetion sites, i.e., minimum of input impedance spoclrum, impulse response function, etc. In this study, we introduce the concept of wave reflection function,which yields a more precise pictose of the location of redectlon sites and thok relative intensity. The wave reflection function i$ a time domain funcdon which is obtained by the deconvolution of the backward running pressure component out of the forward atoning pressure componenL This concept is based on the sr of the aortio preasu~ and flOWwave into their forwasd and harkward running wave cumpor~nts. It assumes that, despite the complexity of wave reflncfions and re-reflnetions within the arterial system,the backward pressure component is the direct rasdt of a reflected fonvasd preSSure wave component. The applicability of the method has been tested using in vivo experiments on anc~h~ised dogs. Different locations ~om the flrorecic aorta to the lilac bifurcalion have beea occluded by inflation of an i~ma-arterial balloon catheter. Occlusion sites have been well predicted by the wave reflection function. No significant differenceshave been observed in the reflection function between the control situation (no occlusion) and occlusion at the ilinc bifim:aboa which suggest that the main reflection sites are upsl~.am the lilac bifm'cation.

203

206

MULTI.WInDOW POWER SPECTRUM ESTIMATION FOR HEART RATE VARIABILITY Eric G. LOveRand Joel B, Myldebast Department of Biomedical Enginuedag, Marquette University, Milwaukee, WI 53233 Laboratory of Sensory Moto~"Performance, VA Medical Center, Milwaukee, Wl 53295 Pow~ sp~-tral a~alysis of heart rate varinbility (HRV) provides a frequency-based decomposition Of h~trt rate varintiens, Spectral decomposition permits quantification of ceolrihoticos to I'~V variance by different physiologic processes. Power spectra of HRV andex ~ t r o l i e d condRions generally o0esist of three ccml~eents disjoint in frequency:, low frequencies (LF) are less than 50 mHz, medium frequcecies (MF) range ~tween 60 and 150 mHz, and high frequencies 0W) lie between 150 and 400 mHz. In general, IF, MF, and HF are rasl~tively associated with varindcos in vasodiladon, blood pressure, and respiratinu. Commonly used methods of estimaling HRV spectra include estimated antcregr--~sive (AID models, Welch pedcdograms (effected by intea-polating I~,V onto uniform sampling grids), and periodograms designed for nenaniform sampling. These spectral estimation techniques o f ~ exhibit complications when applied to ~W.V: (1) Estimates may have bias or variability abcot trends, (2) Estimators make mechanistic zssumplices about HRV (e.g,. AR esdalarors asstmaeoHRV is Ganasian white noise filtered by a ratioeal transfer fnuelion), and (3) Consistent and unbiased auto and ~ spectrum estimates (such ~ c~lerence) are difficult to lmplealenL We develop a mttlli.windew nonparametric spectral estimat~ for HRV which has approximate minimum bias, minimal variability, and consistent auto and czcas specuum estimates. Using a single data segment, the multi-window astimat~ produces smooth spectra. tmequivocally separates LF. lVlF,and HF. makes only standard assumptions about HRV (ergodicity and stadoanrity), allows maltlchannel spectrum estimation, and is easy to implement. Facilities for carrying out this w~k were provided by VA Rehabilitation R&D. Mr. Lover is supported by a Whitaker Foundation Predoctorsl Fellowship.

UNIAXIAL CYCLIC STRETCH OF RAT AORTIC S M O O T H MUSCLE CELLS K. Schnetzer t, P. Delafontaine*, and R. Nerem t eGaorgia Institute of Technology, *Emury University Department of Medicine Major arteries experience circumferential expansion with each heart beat due to internal pressure changes. To mimic this r in vitro, a noveI system was developed which applies a uniaxial, 1 Hz sinascidal atrain regime of 0% - 10% elongation to an elastic silicone subatmto upon which rat aortic smooth muscle cells (RASMs) are grown. Stretched confluent RASMs reorient from a random direaion to an angle 60* - 90* from the direction of stretch over the course of 24 hours, This orientation is maintained threugho~ cerainund application of stretch. C-one regelatin n at the mRNA level was studied with a modified RNase protection assay utilizing biotin labeled RNA probes, In cyeficelly stretched RASM cells, as compared to controls, the amount of mRNA for GAPDH. a metaholie enzyme, is decreased at I hour. returns to control levels by 4 hours, and is elevated after 24 hours. The effects of Insulin-like growth factor I (IGF I), an autOcfine / pamerine growth factor essential for anmml developmental growth, are mediated by the activation of a cell membrane tyrusine-kimlse receptor. IGF IR. The expression of IGF IR mRNA is decreased in cyclically stretched RASMs compared to controls. Likewise, mRNA levels for c-mye, a pinto-on.gone strongly tied to cell proliferation, are decreased in stmtebed v m control cultures, Fluorescent staining for proliferating cell nuclear antigen (PCNA), n eyclin e~ential for DNA synthesis, indicates that the percentage ofFCNA sinini~ cells, (and by itffereace, cell proliferation), is decreased as early as I hour after the onset of substrata strain, in comparison to controls. Mouecyte cbemotactio protein-I 0dCP-I), when sec~ed by v ~ ' u l a r cells, attracts monoeytes flora the blood stream to invade the arterial wall. MCP-I mRNA expression is eievated in stretched RASMs in comparison to centrois throughout 24 hours of stretelt These findings indicate an active cegular response to cyclic stretch and a potential link of mcehanical stimuli to nell regulatory mechanisms implicated hi cardiovesonlsr disenses.

Poster Presentations

S-43

207

210

MODELING OF THE CARDIOVASCULAR RESPONSE TO MICROGRAVITY MK Sharp, GM Pantahis, T Beamed*. S Wcodmff, S O1.,eary.JB Charlest Univea~ty of Utah, Bellasmioc College* and NASA Jchusou Space C.eatsr~"

PRESSURE INDUCED ARTERIAL WALL STRESS AND ATHEROSCLEROSIS Mano J. Thubrikar, Francis Robicssk. Heineman Medical Research Laborator/ Carolinas Medical Center, Charlotte, North Carolina.

The ciJ~ulato~ system of aslmnauts responds to miedogsavity 04-G) by shifting blood ~om the legs tuward the head, reducing circdating fluid volume, inc~a.(mg heart rateand changing Cardiac slrcke volume. Adaptation tu wG, which occors on seversl time scak~ is of wlafively minor health ecocem for short flights, but reedaptutinn to I-G ispotentiallymission and life.-threateain$. Orthostatic i n t o ~ can be manifested in dizziness, blurred vision and Ioas of c o n s e ~ To study the effects of gravity, cardiac funetitua tests wea'e condaated in I-G, wG m d appmnimately 2-G aboard the NASA KC- 135 aircraR with an hydrsulic ~ a u l a t ~ of a~e curdinvasoular system in the elaight (UP), supine (SUP) and leR iatea'al decubim (LLD) postures. Veutriealm" function w u determined by varying cl~euinfing fluid volume (CFV) while syatemle realstan~ was adjuated to maintain a mean afiarlcod pressure of 95 mmHg f ~ d~ UP posture in I.G. All results we.re compared to I-G UP, fcc which tim intrsveatricolar (apex to I~aso) pressure difference 0VAP) was 5 mmHg, tim mean inflow pw.ssure (MIP) wes 16 mmHg and the atroke volume (SV) w u 62 mE At a ncnml cardiac ootput of 5.2 I/rain, Iransitina to the 1-G SUP postme eliminated the IVAP, resalting in a 30% reductlco in SV mula I mmHg drop in MIP, to I-G LLD, 1VAP and SV retomed to the I-G UP value while MIP was 8 mmHg below 143 UP. IVO'was zero for allIt-G tes~ All postures ha tt-G resulted in similar hemodymmir values. MIP wonid increaso or deureaso with entry intu wG depending co Ihe level of CFV and SV was 25% lower than I-G UP. IV&P w ~ niso zero for 2-G SUP, but M]P was reduced by 6 mmHg and SV was 19% inwet than I-G UP. IVAP was incve.~aod to I0 mmHg in 243 UP, while MIP response varied with CFV and SV was 14% higher than 143 UP. in 2-43 LLD, IVAP increased a similar amount, while MIP dropped by 13 mmHg while SV was 15% higher than I-G UP. This accdaration-depcodant cardiac response is likely responsible, in part, for both short and long term cardiovascularadaptationsto the space flight environment and must niso be comidered in tbe soqecnce of t~racfionsleading Io oRImstatic intulerance.

The authors present the hypothesis that high wall stress and accompanying stretch particularly that caused hy ededal pressure, are the pdmary factors responsibia for the topegraphy of atherosclerotic lesione. In their view the pettem in the localization of atheroscierotic iasione indicates that the artery behaves as .beth a pressure vessel and a conduit of blood flow. The phenomenon of "stress cuncentration" in the artery wall is censidered and the aree of pressure induced high stress is related to the sites of atheroscierctin plaques. Data are presented indicating that reduction of pressure induced stress may lead to absence of atherosclerotic changes. The proposed mechanism explains the prevalence of athercsdemtic lesions at the ostia of major arterial branches, at the aortic bifurcation, at the carotid bifurcation, in the descending thoracic aortar and also explains the absence of athercsclerosis in the intramyocardial coronary arteries, and in the intraosseal portions of the vertebral vessels and why a reduction of heart rate, blood preseure, or wall etress by external support reduces the occurrence of atherosclerosis. The effect of wall stress and stretch on atherosderosis could be mediated by the endothelial cells, the smooth muscle cells, and the penetration of low density lipoprcteins. The comprehensive presentation made could lead to a better understanding of athemselerosis, its treatment and prevention.

208

211

MEAN COMPLIANCE AND PERIPHERAL RESISTANCE DEFINE SYSTOLIC AND DIASTOLIC PRESSURE N Sterginpulos, F Pythoud, JJ Meister, and N Westerhof* Biomedical Engineering Laboratory, EPFL, Switzerland *Laboratury for Physiology, Free University, Amsterdam, The Netherlands In an earlier study (Annals of BME, 22:392-397, 1994) we have shown that a 2*element windkessel model of the entire systemic circulation gives very accurate estimates of the aortic systolic and diastolic pressure, despite an overall poor quality of the predicted pressure pulse. This means that from the aRerial side, the only two important parameters needed to define systolic and diastolic pressure are the mean volume compliance and total arterial resistance. This also means that the actual location of reflection sides is not an important determinant of systolic and diastolic pressure. To test the above hypothesis we compared the systolic and diastolic pressure predicted by the 2-element wiedkussol model with the actual measured pressure in the aorta of anesthelized dogs (n=7). The dog data included pressure and flow measured in the ascending aorta, Measurements were performed for control as well as for total occlusion of the aorta (by intraannle balloon inflation) at the abdominal and thoracic (diaphragm) aorta level. The total occlusion at the two aortic sites is chosen to evaluate the influence of strong reflection rites, located at different sites in the aortic tmnL on systolic and diastulic pressure. The results showed that the predicted diastolic and systolic pressure were typically within 3% of the measured values. This was true for control as well as for total aortic occlusion. This quite paradoxical and significant result means that. despite the overall complexity in the geometry and properties of the systemic tree, systolic and diastolic pressure are determined essentially only by total arterial resistance and compliance.

A CLOSED-LOOP NUMERICAL MODEL OF CARDIOVASCULAR CIRCULATION Jianhua Zhou*, Pieter M. Vandervoort, James D, Thomas ** * Ohio State University, Columbus, Ohio ** Cleveland Clinic Foundation, Cleveland, Ohio Doppler transmitral and pulmonary venous flows have been widely used as noninvasive indices of diastolic function. A closed-locp computer model of cardiovascular system was developed to investigate how isolated changes of various diastolic parameters such as left atrial/ventricniar compliance, left ventricalar relaxation time, mitral valvular area transmitral and pulmonary venous flow profiles. We modeled the complete circulation system as eight individual chambers: four cardiac chambors (left/right atrinm, lefl/right ventricle) and four peripheral chambers (systemic mery/vnin, pulmonm 7 artery/vein). The input to this model is a set of parameters including atrial/ven~cniar time-varying compliance, valvular area and initial chamber pressures. 'ntr outputs am instantaneous pressure, volume and flow wavefurms for each chamber. The mathematical description of this model consists of 24 differential equations : 8 for pressure changes based on pressure-volume (P-V) relationships of each chamber, 8 for flow based on valvular fluid dynamics; and 8 for volume changes. These 24 equations ale numerically integrated with an interval of 5 ms using a fco~h-order Runga-Kutsa method. The input parameters can be selectively changed and the resulting effects monitored using a customized program developed within the LabVIEW (National Instruments, Austin, TX) environment- Sensitivity analysis has been done on major parameters ( ~ t e d by Jacobian matrix) to demonstrate how the output flow profiles es'e affected by isolated changes of the input parameters. With different input parameters, the model outputs of wansmitral and pulmonary venous flow profiles have been compared with Doppler images from patients with different pathophysinhigical conditions: normal, delayed and resWictive diastolic filling patterns, cannon A waves and mitral regurgitation. "rialsnumerical model will improve oar tmdet~tanding of the pathophysiology of the cardiovascular system.

209

212

MECHANICAL STRESS IN THE DEVELOPMENT OF A TRANSVERSE TEAR IN THE AORTIC DISSECTION. Mano d. Thubdkar, Ph.D., Prason Agali, M.S., Francis Robicsek, MD., Heioeman Medical Research Laberatory, Carolinas Medical Center, Charlotte, N. C.

Excessive W a l l Stretch Leads to Loss of Function in Stunned M y o c a r d i u m Jia-Jung Wang, Gary Drzewiecki,, and John K-J. Li Rutgars University, Biomedical Eng. Dept., Piscataway, NJ 08855

In the aortic dissection usually a transverse tear develops in the inner layer of the aorta involving more than half of the cimumferanca. To investigate the mechanism of development of aortic dissection we analyzed the pressure induced wall stress in the aortic aneurysm as a function of growth of the aneurysm. To determine the stress, we used a method of finite element analysis in which the aorta was modelled to have isotmpio, nonlinear, end hypereiastic material properties. To simulate growth of the aneurysm, a segment of the aorta was dilated in steps of 10% of the initial diameter. At each step the aorta was loaded with a pressure of 120 mmHg end parameters such as circumferential and axial stresses and strains, radial displacement and wall thickness were determined at the bulb and in the undiiated segment. At the crown of the aneurysm, circumferential stresses changed very little as the bulb became larger, whereas the axial stresses increased dramaticelly, almost by 100%, as the bulb diameter became twice its initial size. In the undiiated segment the circumferential e~'esses increased only a little and =odal stresses showed no change. Tnickneaa at the crown decreased more than did the thickness in the undllated aorta. The dramatic increase in the axial Stress at the crown was pdmedly due to change in shape of the aorta from cylindrical to ellipsoidal to sphericaL We, therefore, propose that as the aneurysm grows pressure induced axial stress increases very rapidly and causes the transverse tear to develop. Furthermore, the tear occurs on the inner wall because this stress is highest on the inner wall.

Reversible motion abnormalities of the myocardium during stunning have been observed, although the actual mechanism of myocardial stunning is still unclear. Previous studies have shown that collagen fiber matrix in the stunned myofibers, associated with the muscle passive characteristics, appeared damaged. In this work, the hypothesis that left venalcular CLV) functional abnormalities during stunning result from an increuse in the passive myofibril resting length (Ix) was examined using a computer model. The model consisted of three parts: the preload, including a filling pressure source (venous return) and filling resistance; the truncated-spbere LV, constructed of a series and parallel alignment of sarcomerus; and the aftorioed, represented by a three-element modified Wiodkassol. Similar to those in canine experiments, the results (Table) obtained with the compute* model showed that an i n ~ of only 5% in the resting sarcomme length could result in a decmuse in percent systolic wall thickening (SW) from 33.7 % (control) to 23.6 %, and percent systolic shortening (SS), from 14.0 % (control) to 10.3 %. Thus, the current mathematical model supports the theory that the alterations of LV wall motion during stunning are a result of an elevation in the resting sarcomare length due to passive matrix damage, which imiuce~ a diminished muscle contractility, in agreement with experiments. Condition Control Stunned I Stunned 2 Stunned 3

Lr (urn) 1.60 1.65 1.68 1.74

EDW (mm) 6.50 6.15 5,93 5.53

ESW ~mm) 8.69 7.83 7.33 6.51

EDL (urn) 2.10 2.16 2.20 2.28

ESL ~m) 1.80 1.91 1.97 2.10

SOt/ (~) -33.7 -27.3 -23.6 -17.7

SS (9[,) 14.0 11.6 I0.3 8.0

S-44

Poster Presentations

213 PROTEIN KINASE C i~mBrrioN INCREASES ENDOTHELIAL PERMEABIIATY UNDER FLOW CONDITIONS

C..M. Waters and C. Craig Departmente of Anealhesla and Biomedical Engineering, N o r t h w e s t e r n University Medical School. We have previously shown that shear stress stimulates an increase in endothatlal permeability, and the incruase may involve modification of cell-to-ecll junctions. Protein 16nase C (PKC) is a key regulator in the recreation o f bonds betwnen the cellc3,toskeleton and junctional proteins. We investigated the effect o f PKC inhibition on endothelial permeability under static onaditims and when exposed to flow using the PKC inhibitor calphostin C (calC). Measured u n d e r static conditions in a t w o - c o m p a r t m e n t system, sodium fluorescein (NaFIsc) permeability was not affected by calC (control 2.5 + 0.3; calC 2.3 + 0.6, mean + SEM x 10"~cm/s, n=10), while cyanooobalamin (1312) permeability increased slightly from 1.7 + 0.4 to 2.3 + 0.5. When measusemonts were taken in a cell onhimn with cells cxponad to 2 dynes/cm~ shear stress, talC stimulated a steady increase in permeability which was s i ~ t l y different (p<.05) from baseline 50 to70minafletexposuretocalC(NaFlsc: 14.0 + 3.9 to 30.3 + 8.4; B12: l l . 0 + 3 . 0 t o 19.3 + 4.3; mean + SEM x 10"~cm/s, n=5). These results sugg~t that PKC is important for the maintcnanon of endothelial harrier function under flew condition, but has a minimal effect on endothalial pecmeability under static Ooadilions. In~bition o f PKC while cells are expoca~ to shear stre~ may affect the ability of the nell eytoskeleton to rcmodel and adjust to maintain barrier. This work was supported in part by NIH grant RR-05370, the Whitaker Foundation, and the Dept. of Anesthesia.

214 THE MECHANICAL PROPERT[ES OF THE THORACIC AORTA IN THE PERINATAL PERIOD AND ADULTHOOD S . M Wells. B.L. Langilie, S.L. Adamson Samuel Lunenfeld Research lnsli|ute, Mount Sinai Hospital We have determined the mechanical properties of the sheep thoracic aorta as it remodels during the perinatsl period (+3 weeks of birth) and in adults (4 to 10 at each age). Two methods were used The wave propagation coelficient (describes the velocity and attenuation Of a travelling pressure wave) was measured in vivo in acutely-eathetefizod shnep. The static elastic modulus (ratio of static stress to strain), and the dyonmic viscoelastic modulus (rcintes llme-va~ng stress to strain) ~ere measured in vitro, using a mechanical testing apparatus, The: stiffnees of the thoracic aorta increases ~s~thdevelopment from the fetus to the lamb and adult: the static elasticmodulus (F,.~)in vitro of thc fetalaortawas lower than of both the lamb and adult anna, whose static elastic modull ,,','ere similar. These results were supported by calculatious based on changes in pressure unve velcci~" (c) in vtvo. The thoracic aorta also becomes less viscous with age. Both pressure u~ve attenuation (~) in viva and viscoelastic phase angle (~P)in vitro were rcducad with age frqm the fetus so the adult. Both methods indicate that the thoracic anna becomes stiffer and less viscous uith age, and most of these changes occur in the pennatul period. We h)pothesize that this is largely a result of the pefinatsl remodelling of this vessel, v.'hich invoh'es a rapid E~ r @ a accumulation of elastin and ':106dvWcm:; (m/s) (de[reck) (cm") collagen (elastic elements) fetus 0.53 + 0.07 o 3.6 • 0.2 a 16.6 + 2.7 a 0.067 + 0.017 a ",s~th unchanged amounts of inmll 1.03+0.10 b 5.8+0.4/' 14.2• a'~ ).035• a/' smooth muscle (viscous adull 1.07+0.06 b 4.8• c 10.6+0.8 b 0.019• b elements). Supported by MP.C studentship and HSFO. Mean • SEM. For each variable, different subscripts indicate significant differences, (P<0.05)

215 PULSATILE FLOW OF A GENERALIZED OLDROYD-B MODEL FOR BLOOD K.K. Yeleswaranu. J,F. Antaki, M.V. Kamcoeva, and K.R. Rajagapul University of Pittsburgh Schools of Medicine and Enginec~tg. Pittsburgh, Pennsylvania, USA A generalized Ol&oyd-B fluid model was proposed [1] to describe the obsczved shear thinning and visoneinsdc behavior of blood. The generalized viscosity fanctlon was shown to capture the shear dependence of the apparent viscosity. Further. the steady Poisonilie flow velocity lXOfilepredicted theoretically was shown to be in aSrecmem with the expedmso~ ~ e a t . Present work is a pantmetric stody of the pdsatiio flow Of a gonecalized Oldsoyd-B fluid in a stralght rigid robe ondec a sinusoidul lnessure gradleat. Duo to the complexity of the geanralized viscosity foncfion, the equafico of motlan becomes a highly annliuaac perdal ~fiul equation n~quh'inganmodcad solution. A collocation method was employed for this purpose. "[he noadlmeuainnal velocity pcofiles for various combb~ious of material pmameters uad Womcndey anmber (a) were examinod. It was obsesvod that the ve~iation in the viscoelastic perameten (A: and ~ alters the amplitode of the centedine velocities. The deviation flora, the steady shape wets observed to ecctEr at re]ativeAyhlgh~ Worar numbe~ due to shear thinning. The q~m!imtlvenature of the anndimeusinonl velocity l~Ofiles oncomages further consideration of the generalized Olthuyd-B model as a ~ c~tstitotiv~ equation for blood. [1) ILK. Yeleswampu, J.F. Aatal~M.V. Kameuava, and K.R. Rajssopal; "A Ganeralized Oldmyd-B Model as a Constitutive Equation F ~ Blood," Presented at the 1994 Annual Fall Meeting of the Biomedical Engineering Society, ASU, Tempe, Anzoua. Oct 14-t6, 1994.

216 ESTIMATION OF CARDIOVASCULAR BED PARAMETERS Yih-Chonng Yu I , J. R. Boston 1"2,James Antaki23, Marwan Simuan I , DmTylS Breitcostein I Depamnent of Electrical Engineering, 2Bioonginnexing Program, 3 D ~ e n t of Surgely University of Pittsburgh, Pittsburgh, Pennsylvania 15261 Hemodymonic patametera ~-flest the metabolic demand of the body and the pumping cal~ility of the heart. Using the simple electricmodel shown below to represent the cardiovascular system, a ba~h least-squareaalgorithm that can be applied for the multi-inpat multi-output sys~ms in continuous time domain was used to identify parameter values. The algoritiun allows independent parameter esfiraates to be obtained during the ejection phase and during the rifling phase of thc cardiac cycle. Preliminmy performance evaluations using data faom a simnhdon of the cardiovasculac model are presumed. The antic presst~c (P.O. atm~ flow fix), left atrial p~essta-e (!~), and vcac~ return flow (iV) in one cardiac cycle were taken as me~uremonts. A filtered differen~on technique was applied to obtain the time derivatives of the measurements. The actual values used in the simnla~on model and the results obtained by the esfiraator are listed in the Table. This algorithm performed quickly and could be used to determine the constancy of model peramete~ over the cardiac cycle, allowing a compmison of values during systole and diastole. Summar~ of the Estimation Results

L.

p ~ t l .~ no(tJ ~ '(fi ~*l~-~ -" s*0)"~, ~. TI~ eler

~(t] ~.,

I ec. I

.039s

.0s9~

I.~ C:S 1~ CR El

.001025 2.896 .8738 4 -

.00102 2.975 .8M2 4.029 .0338

2.859 .8711 4.079 .0129

A~tleg of the Caldtovascular Mod~

217 T H E R M A L EQUILIBRATION IN RAT CREMASTER MUSCLE L. Zbo,* D. E. Lemons,** and S. Weinbsum* of Ml~.huaicaiEnsionoring* and I ~ n t of Biology** "me City College of The City Unlverfity of New Yeck

MICROVASCULAR

I~at

A new e x p o ' ~ m l appronch was developed to obtain the f'ust direct measm'emeats of the axla] counte~'m*reut thormal equilibration in a mlcrovascular tissue Wcperadon using high-resolution ins thr Detaikd stufacetempr162 mcasu~n~nts were obtalnod for an emr rat cremas~" musclein which l:k~macological vaso~tive agents were employed tOchange the local blood flow Pcclct onmbegf~omI to 14 in the reading e~ezy. Under normal condition&,only the IA re'reties(> 70 pm di&)shona:dthmual nonequllilxafioa with the surrounding tissue. The theoretical model developed in ~ m sed Weinbanm (1995) for a two-dlmeesionel tisme t~,j~etlon arbitrarilyembedded ononterctmentvesscls is mcclified to incinde axinl oneductlon and the pt-esonon of the mppor~ug glass slide. "l~is modified model is employed to i n t ~ the expe~imontalre~ul~ 8~1 m wJate thc m r f ~ ~ x r ~ c prardca to ~ bolk tampa~m~ Wofilea in ~ ~ a ~ arteTy and vein and the local averase flume tempm'atare in the ero~-socr plane. Surfane tsmporaw~ pe0~ke ~ to the vessd ar~s are shown to depend ~fiSnificaatlyon tbe ~ r iulet temporally. "lhe #~-,~uoction fro"the axiM thermal cqulitbcafiondepends p~ma~y on ~ blood flowPeeler~ b e r asd theeavhunnz'mal coove~ve eooff~ont. Tbo thoon~cul reeultspredi~ that whoa p,,*Pe klem d~on1 mm (the range in onr experlmeats), a.~ul oneduction is the demi~mt made ofwdal tbermal~ For 1 < p~'Pe< 3 ram. o n u a ~ t blood flow beonmm c~mt.,~,~bin to axial c o n d u c ~ v,hereas, v,ben p~*Pe > 3 ram, onuate~cm'reutbhied flow is the domionat mode of axial thermaleqm'itTmlfian.Thas, for p,," Pc > 3 mm the axinl equilibration Iongth is proportional te tbo blond flow Pesint anmber as previonsly predlctod in Zhu ~.edWeinbanm (1995) whom aalal condectiou was neglected. It is also shown that the axial decay of the tissue tamperatmc at Iow pelTm~e rat~ can be described by a simpin oon-dimeasioual Wcinhaum-liji equation with a newly derived conduction shape factor.

Cellular Engineering in Bone

S-45

221

218 BONE FORMATION USING POROUS POLYCLACTIC-C'O-GLYCOLIC ACID) SEEDED WITH STROMAL OS'I~OBLAST CELLS Genevieve M. Crane, I Susan L. Ishaug, t Michael J. Miller,2 Alan W. Yasko,2 Thomss B. Aufdemoste? Michael J. Ye.~emski,4 and Antonius G. Miknat qnstitote of Biosciceces and Bioengineering, Rice University, Houston, TX; 2University of Texas M.D. Anderson Cancer Center, Houston, 'IX; suulversity of Texas Health Science Center. San Antonio, 'IX; sWilfot'd Hall Medical Centar, LacHand A l ~ , 'IX

TEMPORAL EFFECTS OF SUB-LETHAL TI-6AI-4V ION SOLUTIONS ON OSTEOGENIC CELLS DERIVED FROM BONE MARROW STROMAL CELLS GJ. Thompsee and D.A. Puleo Center for Biomedical Engineering, University of Kentucky

Autulogous bone tissue flaps are tcotinely used is skeletal reconstructive surgay to fill defects following tumor removal: however, many limitations exist with this technique. We are studying the feasibility of segineering natural benc flaps by the induction of bone tissue into cull-seeded pomes hiedogradable polymer foams implanted is e sultsble In ~ site. ~ prefabS.cared vsscal~rized bene flal~ will ulinw fer the exl~,,s/ce of donor tis.~ae, will be lass ievasive, ned may ho formed into a variety of irrogular shapes. The ~ of this study was to investigate beec formation in vitro by culturing sfromal osteoblssts in 3-dimessinsel degradable polymer foams to study the effects of foam morphology and culture conditions on osteoblsst prelifesafion and function. The IXXOUS biodegradable 75:25 poly(DL-lactic-co-glycolic acid) foams were made of cco~olled pcessity, pore size, and thlcrmess by a solvent cantiog-particulate leaching technique using sodium chloride as the Icuchablc componseL The stmmal c~teoblasts were obtained from cultured sheep bece nuemw aspiratu and ware seeded onto the center of the pomes foams sed cultored tn ~tro fc~ up to 60 day~ Light m ~ and video imegisg software were used to determine the effects of pore size, polymer thickness, and seeding deasity on esteoblsst prdiife~ation and fu,',ctiee. Histological sections were stained with hematoxylin and eesin to visualize tissue formation, ulkelioe phosphatase stain to assess the antivity level of the eusyme, and yon Kessa stain to deteunine the extent of mineralization in the const:ucts. These results will enable us to determine the appropriate foam morphology and cultme conditions to be used in future In vh,o studies.

INTRODUCTION: Metal ions and particles released from the implant surface are suspected of playing some role in ostcelysis surrounding hip and knee prostheses [C/In. Orthop. 276:75, 1992]. Previous work demonstrated that sub-lethal doses of the ionic constituents of Ti-6AI-4V suppressed expression of the es~ohisstio pheno~3e and deposition o f a rniucreilzed matrix [Trans. SFB 18:417, 1995]. The pasposs of this work was to further explore this suppression as a function of the setmal time.course of phenotype expression. MATERIALS AND METHODS: BOne marrow stromal cells (BMCe) were harvested from juvenile male Spmgue-D9 rats and seeded at 9 density of 6500 eeUs/cm2. BMCs were then exposed to 9 5 ppm solution of ions repre~ariog Ti-6AI-4V alloy (ASTM F136) beginning from 1 day, I week, 2 weeks, or 3 weeks 9 seeding. Cells were fed weekly for four weeks. At the cud of each week medium, lysato, and inorganic matrix were collected. Lysate was assayed for total cell protein and for llkaline phospluttase, medium for osteecalein, and matrix for calcium. RESULTS AND DISCUSSION: Ti-6AI-4V solutions were found to produce little diffeseace from control solutions for total protein or alkaline phosphatase levels. Whcu ions wese added at or before a critical phase of osteoblastic differeatiation Csetweea 2 and 3 weeks after seeding), both ostoucalcin and calcium levels wore reduced, but if added 9 this time only osteocalcin was affected. These results indieata that ions associated with Ti-6Al-4V inhibited the normal differeatiatico of BMCs to mature ostanblasts, and that if miceralir~tion was successfully initiated, calcium deposition proceeded unhindered by the presence of ions.

219

222

ENGINEERING BONE CELL ORGANIZATION IN T W O DIMENSIONS K.E. Heely, u C.H. Thomas, u A. Resania, u P.E. Hockberger, 3 and P.J. McKcown 4 IDiv. of Biological Materials, 2Dept. of Biomedical Engineering, ~Inst. for Ncuroscience, Northwestern Univ. Dental, Medical, and Engineering Schools, 4physical Electronics Inc.

AN IN VITRO CELLULAR MODEL OF BONE ADAPTATION TO LOADS: FLUID SHEAR STRESS ELICITS SYNTHESIS AND RELEASE OF OSTEOPONTIN AND BONE SIALOPROTEIN BY OSTEOB~ H. Y. Shin, F. A. Blumenstock', R. Bizine Dept. of l~omedieal Engineering, Renuziser Polytechnic Institute, Troy, NY 12180 *Dept. of Phymology and Dept. of Cell Biology, Albany Medical College, Albany, NY, 12208

The perfmmaeee of device~ used in dentistry and medicine could ha improved by engineering theorganizationofdifferent tissue in discrete locations on the materlal. In this study, we measured the kinetics of spatial organization of hone.derived cells on chemically patterned subsl~ates, and whether our observations of initial attachment and sp~..edlng affected the rate of matri9 mineralization (i.e., hone formation) in extended cultures. Experiments were conducted with patterns of N-(2-amlnoethyl)-3-amianpropyluimethexysilaoe (EDS) and dimethyldichloresilaee (DMS), EDS/DMS. Tune-lapse video microscopy was used to observe the kinetics of cell organization on the EDS/DMS patterned surfeces. Extended culture.q of cells, 15r and 25d, on the patterned surfases wen: assessed for the ability of the substrate surface chemistry to alter the rate and location of mineraliz~ion. The bone cells reo~red the presence of se~m or adsorption of sennn proteins to the pattsnaxi EDS/DMS surface to organize ascutdlng m the lithographically defined swface chemistry. Trine-lapse video microscopy indicated that cells were aided in their movement by natural convection attributed to thermal variations throughout the medium during the start of the experiment. After 30 rain, the cells aligned and spreed on the EDS regions, and became confluent on EDS regions after 24h. In extended cultures cells eventually spread on DMS regions and funned n confluent layer-; however, mineralization of tissue preferentially occurred on the EDS regions of EDS/DMS patterns. results demonstrate the ability of surface chemist]ry modifieatioes to organize cells and form mineralized tissue/n vitro. The methods employed should have general value to the engineering of tissues in r

220

Cultured mteoblssts (the bon 9 rells) were exposed to steady 9 stress to investigate t h e effects of mechanical forces on synthesis and release of proteins. Neonatal r a t ca]varial cotecblants ( p n ~ w e 2 to 3) were exposed to steady, fluid, s h e a r stresses (2 to 12.5 dynes/era z) under serum-free Dulbecco's modified Eagle medium in a 37~ C, humidified 5% CO2/~5% a i r environment for 6 hours. Experiments were also conducted using mteoblasts pretreated w i t h either 14 p.M cycloheximide (an inhibitor of protein synthesis) or 30 i~g/ml monensin (an inhibitor of protein secretion) for 3 hours prior to 9 stress exposure. Controls w e r e estsoblast preparations maintained u n d e r static (no-flow) conditions but, otherwise, similar experimental conditions for 6 hours. The proteins present in supernatants of osteeblants from the shear stress experiments a n d from static controls were isolated a n d identified based on reactivity w i t h polyclonal antibodies to setespontin and bone sialqpretein. Compared to 9 from static (noflow) controls, the supernatsnts from esteoblasts exposed to s h e a r stresses contained osteopontin a n d bone sialoprotsin; most important, these proteins were absent in the s u p e r a a t a n t s from oeteoblssts pretreated with either cyclohe]dmide o r moneeein. The results of the present study provide evidence t h a t expceure of coteeblants to fluid s h e a r stresses elicits de novo synthesis a n d release of eeteopontin a n d bone siainprotsin, two protsinn which are revolved in bone formation a n d remodeling.

223

OSTEOBIJ~I FUNCTIONS ON PEPTIDE-MODIFIED SUBSI2ATES IN THE PRESENCE OF G 2 O W I H FACFOs K.C. Dee. T.T. A n d e r s e n ' . R. Bizios O e ~ r n e n t Of Biome~cad Engineedng, RensseloerPolytechnic It~litute. Troy. NY 12180 * Oep(~Irnenl of B ~ f n i s h y ond Molecutof Btek~y. ~ Medical College, Al~ny, NY 12200

Cellular- and/or molecuiar-level methods of manipulating cell/tissue behavior are being used in order to induce clinically desirable biological events at the tissueimplant interface. For example, rapid deposition of mineralised matrix at the site of newly.implanted dental/orthopedic prostheses supplies crucial mechanical stability to these implants. It is, therefore, desirable to define send;bonn at the bone-implant interface which elicit swift deposition of mineral by ostcoblasto (the bone-forming cells). To this and, the present andy investigated cetsoblast functions (adhesion, ehemotaxis, proliferation, and mineralization) /n vitro on substrates modified with the immobilized bioactive peptide Arginine-GlycineAspartic Acid-Serine (RGDS) in the presence of various concentration 9 (0.1 - 100 ng) of either Transforming Growth Faetor-I~, basic Fibroblant Growth Factor, or O9 Pretein-I (OP-I). Substrates modified with the nonadhesive peptide Arginine.Aspartic Aeid-Glycine-Serine were used as centrels. Mineralization of t h r e e . w s e k oeteoblant cultures on nubstrates modified with RGDS, in the presence of 100 ng OP-1, was enhanced compared to all other culture conditions examined in the present study. These results suggest that combinations of select bioactive agents may synergistically enhance clinically desirable cellular functions; such approaches could be utilized in the development of biomatsrials which can elicit specific, timely, and desirable responses from surrounding ceils and tissues.

RAPID FLOW-INDUCED PRODUCTION OF NITRIC OXIDE IN OSTEOBLASTS John A. Fmngos and Dameron L. Johnson Department of Bioengineering, University of California, San Diego, La Julia, CA 92093-0412 Interstitial fluid flow may mediate skeletal remodeling in response to mechanical loading. Previously it has shown that fluid flow stimulates synthesis of prostaglesdin E2, which may act to increase hone formation. Evidence suggests that hone resorption is also affected by mechanical loading. Since nitric oxide (NO) has been shown to mediate resorptice in hone, we investigate and characterized the role of fluid shear on the release of NO in osteohiests. Rat celvminl cells in stationary culture produce undetectabln levels of NO, as determined by Gseiss reaction. Fluid shear stress of 6 dyn/cm2 increased NO release to 80 nmols/br/mg protein. This release rate was sustained during the course of 12 hrs of exposure to flow. Cytokines (100 ng/ml TNF-a, 10 ug/ml lipopolysaeeharule, and 100 U/ml interferon g) also induced NO synthesis, but only after 9 12 hr lag phase where no NO was produced. After 48 Ins of eytokise treatment, 35 nmuls NO/rag protein were produced. The cytokine-induced release of NO could be inhibited with dexamethasone, while that stimulated by flow could not. The stimul9 production of NO-in both cases was inhibited by N-amino-L-asginine, an NO synthese inhibitor. It thus appears that fluid shear stress can stimulate a constitutive isoform of NO synthase in estcoblasts to produce NO at rates much greater (10-fold) than is produced by the cytokine-inducible NO synthase. These results suggest that skeletal interstitial fluid flow may regulate osteeclastic resotptioe as well as esteoblastic formation activity.

S-46

Gene

Regulation

by Mechanical

224 AtTEEATION iN ENDOTHEUAt CeLt ADHESION MOLECULESIN ItESeONSETO SUSTAINED HYDEOSTATIC FEE$SUEE E.A. Schwartz, R. B~os. and M.E. Gerdtsen* Dept. of Biomecl|cal Engr.. Rensselaer Potytechn|c Instilute, Troy, NY 12180 =Institute for Adl'vitis and Autoimmunily, Bayer Corp.. West Haven. CT 06516

Forces

in Mammalian

Cells

227 DIFFERENTIAL REGULATION OF ENDOTHELIAL VCAM-1 GENE EXPRESSION AND NF-'~BACTIVATION BY LAMINAR SHEAR STRESS Signa E. Vamer. Robert M. Narem, R. Wayne Alexander. and Russell M. Medford Georgia Inslitute of Technology and Emery University, Atlanta. Georgia

Sustained hydrostatic pressure alters the morphology and function of vascular endothelial cells/n v/Era. In particular, bovine pulmonary artery endothelial cells elongate, proliferate in multiple layers, and eeerete basic fibrobhlst growth factor from internal Stores in response to sustained hydrostatic pren~urss of up to 15 cm H 2 0 for 7 days. The present study investigated molecukax.leval effects of sustained hydrostatic pressure on human umbilical vein endothelial cells. Immunofluorencent micrenenpy was used to visualize specific integrin subunite known from the literature to be present on endothelial cells; integrins are a large family of dimeric glycoprotein adhesion molecules, composed of specific combinations of an o~and a 13subunit. Compared to controls (ceils incubated under 0.2 cm H20 sustained hydrostatic pressure under identical culture conditions and for identical time periods), endothelial enlis exposed to 4 - 8 cm H20 sustained hydrostatic pressure for up to 4 days exhibited redistribution of integrin subunite u2, ct3, r ~ and 1~3 on the abluminal surface of the cells. There was no change in the distribution of integrin subunit 131at any pressure or time point examined in the present study. Exposure of endothelial cells to pressure of 2 cm H20 did not affect the distribution of any integrin subunit examined at any time point in the study. This study provided evidence that exposure of endothelial cells to sustained hydrostatic pressures greater than 4 em H 2 0 affects at the molecular level the physiological mechanisms of endothelial cell adhesion.

Atherosslamsls is an inflammatory disease of the vasculatura chemcteriznd by oxklation-reduction (radex) sensitive endothelial expression of the cell adhesion molecule VCAM-1 and by 9 predulectionfor focal sites on the vessel wall. These edtea era characterized hemndynamloally by low and oscillaloly shear stress. Previous studies in our lalxxatoly with human umbilical vein endcthelinl cells (HUVECs) have demonstrated parallels between ahear streets and antioxident e~orsssion of cytoldna induciole VCAM-I protein and gene ~preoslon. VCAM-1 transcription is depandant on NF-KB binding to the tandem consensus sequences -,d.-lcR (ceords. -72 and -57) on the promoter. The antioxidaut PDTC suppresses VCAM-1 promoter activation by pmvanting the cytoldce activation of this important nuclear factor After axpooura to a shear stress of 5 dynes/cm2 for 24 hours nnd stimulation with IL-11~,gel shift analysis was performed to determine NF-r binding activity in HUVEC monolayem. In preliminary experiments, steady laminar shear stress had no effect on cytoldno inducible NF-~B binding activity compared to static controls, while shear stress alone (without cytoidne stimulation) slightly augmented binding to the KL..~R eaquance. A human microvsscular endothelial cell line (HMEC-1) was transfeotnd with a region of the VCAM-1 promoter (o0oeds.-288 to +22) ceutaining the r element fused to the reporter .g.anechlora..mphanicoleoetyltran~erase (CAT). ARer shear Ixeconditioning and cytoldns stlmuintton, CAT enzyme actNity Was almost completely inhibited, while a constitutwely expressed pSV2CAT expression vector transfected in parallel was unaffected by shear. This raises the intriguing possibility that in buth endothelial cell types, shear stress may regulate VCAM-1 gene expression through distinct transcriplional mechanismsinteracting either directly or indirectly with NF-'
225

228

ROLE OF SHEAR STRESS RESPONSE ELEMENT (SSRE) IN REGULATION OF PLATELETDERIVED GROWTH FACTOR-B (PDGF) IN ENDOTHELIALCELLS (EC) EXPOSED TO STRAIN, Wei Du, M.D., Gary Gal~agher,M.D. MichaelGimbmne.Jr., M.D.*, Nitzan Remick, Ph.D.*. Bauer E. Sample, M,D,, Ph,D,, DopE.of Surgery, Yale Univ, Sch. of Mad., New Haven, CT 06510 and *Dept, of Pathoingy, Harvard Medical School, Boston, MA

BQ-123 antagonism of endothelin-l/ETA receptor pathway p r e v e n t s vasoconstriction d u r i n g c h r o n i c low flow in r a b b i t . W.J. C_aDo, G, Hajduezok 1, and S.L. Diamond Departments of ChE and Physiology 1, SUNY at Buffalo, Buffalo, NY 14260

Exposure of EC to repetitive stretch or shear sla~,sshas been demonstrated to result in upregulation of PDGF. Previous studies have indicated that an SSRE site at -125 in the promoter region of the PDGF gene is important in the activation of PDGF by shear stress. In order to determine whether the SSRE site is involved in PDGF activation with cyclic strain, bovine aortic EC were subjected to 60 cycles/~tin, 10% strain for up to 24 h. Nuclear extracts were obtained and gel shift studiesconfirmedbindingto SSRE. EC were then bansfeated with deletion constructs of a 450 bp 5' flanking fragment of the promoter coupled to a ehloramphenicol acetyltraesferese (CAT) repor~r, subjected to cyclic strain and then harvested and CAT activity determined. The results indicate that in EC exposed to 10% slrain (n=5) there was a 2.40% increase in CAT activity in the 450 hp promoter construcxat 4 and 8 hrs (*p<0.01 compared to control by ANOVA). In ennwaat, there was no sigafficaut inerense in CAT activity in EC exposed to 6% sWain(n=3). There was a diminution in CAT activity between the -313 and -153 region of the promoter with complete abolition with the -101 promotm"construct However,erausfectionwith a 450 bp PDGF promoter with mutation of the SSRE site, failed to abrogate the sWain-induced increase in promoter activity. Likewise, transfaetion of EC with a hcterelogous SSRE.SV40-CAT construct, failed to induce an increase in CAT activity. Thus, cyclic strain activatesnuclear transcription factors which can hind to elements in the PDGF promoter, including SSRE, leading to PDOF expression. However, binding to the SSRE site is neither necessary nor sufficient for this response.

226 I n d u c t i o n o f endothelial n i t r i c oxide s y n t h a s e gene expression

fluid shear stress is PKC and NO-independent. V. Ranjan. Z. Xiao. S. L. Diamond Bioengineering Laboratory, SUNY at Buffalo, Buffalo, NY 14260 Shear stress exposures on endothelium causes induction of constitutive nitric oxide synthase (eNOS) protein and mRNA levels. In beth endothelial cell types, the intracellular level of eNOS was elevated within 3 hr of flow exposure at 25 dynes/era 2 and remained elevated at 6 and 12 hours of flow exposure, compared to stationary controls, as indicated by digital immunofluorescence mlcroscepy. Shear stress exposure for 6 hr caused a 2.2 + 0.3 and 2.8:1: 0.3-fold elevation of eNOS proteins levels in BAEC (n = 3, p < 0.01) and HLrVEC (n = 3, p < 0.01), respectively, in the presence or absence o f 1 p M dexamethasone. The flow induction of the eNOS protein levels was not blocked by preineubation o f BAEC with 100 to 400 g M of NG-nitro-L-arginine methyl ester, indieatlng that a positive feedback loop for flow induced N O (or eGMP) was not involved in the elevation of eNOS levels. Also the protein kinase C inhibitor, H7 (10 I.d~_) ~2 . had no effect on induction of NOS protein in BAEC exposed to 25 dynes/am The eNOS m R N A levels were found to be elevated by 2 to 3-fold in BAEC after 6 or 12 hours of flow exposure at either 4 or 25 dynes/era 2, and this induction of NOS m R N A occurred in the presence of dexamethasone. PMA or LPS had no significant effect on eNOS m R N A induction. The contrasting effects o f arterial shear stress on the regulation of endothelial NOS and endothelin-1 gene expression suggests a mechanism for vessel diameter matching to prevailing blood flow.

We have sought to identify molecular events that cause vasoconstriction during chronic low flows in arteries. To create low flows, we stenosed (>50%) the left femoral artery of N Z white rabbits, while the fight femoral artery served as a sham operated control. A marked vasoconstriction of 20 to 35 % reduction (n>8) in diameter and waviness of the internal elasdc lamina were observed within 3 days as well as at 7 and 14 days in stenosed vessels relative to sham operated controls. The constriction was observed both upstream and downstream o f the stenosis indicating that the vessel was responding to low flow and not pressure dynamics which would be vastly different in these two regions of the stenosed vessel. Using a 25 % platonic gel containing 1.0 p M BQ-123 placed on the adventitia of the vessel at the time o f stenosis, we tested whether the vasoconstriction induced by low flow w a s due to pathways susceptible to endothelin antagonism. We found that the vessels stenosed for 3 days with BQ-123 had essentially identical diameters as the sham operated vessels treated with BQ- 123. Also, we have found that BQ-123 caused dilation of vessels after 3 days relative to pluronic gel treated vessels under normal flow conditions. This suggested that the endothelin pathway, in addition to control of basal vasotone, may play an important role during vasoconstriction under chronic low flow conditions.

Cell Adhesion under Dynamic Conditions 229

231

FLOW MODULATION OF MOLECULAR MECHANISMS OF ADHESION OF T CELLS TO HqDOI'tlELIUM KoasUmenm Kcoat~tolx~ce, Shared Kela'ee, C. Wayne Smith*, and Lmsy V. Mchie~re The Cox Laboratory for Biomedical Engineering, Rice University; *Bayl(x College of Medicine. Hou.qon, TX.

NEUTROPHILS USE BOTH SHARED AND DISTINCT MECHANISMS TO ADHERE TO SELECTINS UNDER STATIC AND FLOW CONDITIONS MU. Nollert5, K.D. Patal 1 , K.L. Moore1,4, and R.P. McEvar 1,2,3,4 Departments of 1Medicine end 2Biochemistry and Molecular Biology, the 3W.K. Warren Medical Research Institute, University of Oklahoma Health Sciences Center, the 4Cardiovascular Biology Research Program, Oklahoma Medical Research Foundation, Oklahoma City OK 73104; and the 5School of Chemical Engineering and Materials Science, University ct Oklahoma, Norman OK 73019.

The objective of this work was to investigatethe specific me]easier me('h~n;~mao f T r162 adhesion to vasculareedothellum under defined condifioas of flow. A parallel-pinta flow chamber was usedto mode] physiological coedillons that mimic blood circulation in postcapillary vonulas. T cell saspunsions, isolated from human ~)esipbe~ blood, were passed through the flow chsml~r at a wall shanr sl~ess of 2 dynes/cm~, and tbe tatarantlon cg T cells with the eedotbenal moeoinyer was obse~red under phase.contrast microscopy foe Ill minutes, nmman mobitical vein endothelial cells (EC) were lxetranted with histamine (10-4 M - l0 rain) or int~feson~f (2000 U/ml - 24 hoers) or intedeukta 4 (50 ng/ml - 24 been). results indicate that extensive T cell adhesion was detected in hiatamino-atimalated EC ('792.~77 T coil#ram2), and intelcokin 4-stimalated ~ (582:1:119 T celLVmm2), bet not in hitesfct~0e-';,trealed EC (57+41 T cr AEFoxlmately 97% of the T cells raged on histamine-trsated EC at 24.5 I,tm/soc. Moanr antibody (MoAb) 01, specific for Peclectic, completely blocked association of T cells with Idatamine-stimdlated EC, whereas an anti-L-aclec~a MoAb inhibited T cell adberceee by aboat 00%. The average T cell rolling velodty on tatarlenlda 4-sttmuinted EC was 10 Ixm/sec, and a significant portion of the tatefacttag cells were fu'mly adherent (98:1:27 T cellshnm2). Adhesion was completely abolished by "~Jmoat of T cells with an anfi-CD49d Moat), and re~inced by 45% and 62% by Moabs specific far the vesoalar cell adhesion molecole-I (VCAM-I) and anti-CD29, respectively. Preliminary data indicate that an anll-L-sdleetin Moat) also inhibited T cett ~ e s e n c e to inteJcukia 4-atimelated EC by 35%. This study contributes to otw understanding of the molecular mechanisms involved in the interactions between lymphocytes and EC d u n g eX~T.~VS~qafionat ~ of infl~mm~Joe.

S-47

Both P-cetactin glycolxotain ligand-I (PBGL-I) and L-selantin ere localized on the microvilli of neutrophils and have been implicated in the attachment of neulrophils to P-selectin or E-solectln. We directly compared the attachment and rolling of neutrophils on P-selantin and E-solectin under flow, with emphasis on the functions of PSGL-1 and L-ealantin. FLowing neutrophils attached more avidly and rolled at lower velocities on P-salectin than on E-solectin at matched densities. E-salectin bound with lower affinity than P-salectin to this site and bound to an additional airs(s) on PSGL-1. PLI abolished adhesion of neutrophils to P-selectin u~dar shear or static conditions, whereas DREGo56, a mAb to Lselantln, had no effect on adhesion to P-selectin. PL1 inhibited attachment of neutrophlts to E-selectln under flow but not static conditions. DREG-56 also inhibited attachment of flowing neutrophils to E-solectin, and a combination of OREG-56 and PL1 nearly elelminatnd attachment to E-solectin under flow. These data suggest that PSGL-1 functions cooperatively with L-eelactin to mediate optimal attachment of flowing neutrophils to E-selectin but not to Po selectin. Neutrophils attach more efficiently and with greater strength to Psalectin, perhaps because at the higher affinity of P-selectin for the PLl-deffnnd site on PSGL-1.

230

232

SELECTIN-MEDIATED LEUKOCYTE ROLLING: LESSONS FROM GENE-TARGETED MICE Icy. K., ,lung, U., Kunkel, E., Norman, K.E. Biomedical Engineering. U. of Virginia, Health $ci.Ctr. Box 377, Charlonesville, VA 22908

THE ADHESION OF HUMAN COLON CARCINOMA CELLS TO INDUCIBLE ENDOTHELIAL CELL ADHESION MOLECULES. D. Goetz, H. Ding, D. Hammed', M.A. Gimbrone,Jr. and F.W. Luscinskas. Department of Pathology, Brigham & Women's Hospital, Boston, MA 02115 and * School of Chemical Engineering, Corneal University, Ithaca, NY 14853.

Leukocyte roiling results from a delicate balance between the adhesive forces of rapidly feinting and separating leukocyte-endothelial adhesive bends and wall shear stress. The selectins have been shown to be necessasy and sufficient for leukocyte rolling in many in viva and in vitro models. Mice deficient for each of the seleetins (E-, L- and P-solectta) have been produced by gene targeting and homologous recombination in several laboratories. P-seleetin deficient mice show no leukocyte rolling immediately alter tissue trauma, but rolling c ~ be induced by UrJement with the cytokine TNF.,~. Rolling in TNF-a treated P-solectin deficient mice is completely E-soleetin and paxttally L-~lacdn dependent. Nentrophil migration into acute peritonitis is severely reduced el 1-4 hours, demoastrattag that P-seleetin function is naelimiting under these condition. L-selcetin deficient mice show a similar inflammatory defeet wldch is sustained for up to 48 houri. Leukocyte rolling in L-solcetin deficient mice is normal L~ddallybut severely.reduced at > 1 hour al~er trauma. E-solec~ deficient mice ~tve no oven defect in acute inflimm2don or l~kocyte mlItag, but are exquisitely sensitive m antibedy blockade of P-selenfin function, which severely redur~ beth rolling and nentmphil recndtmant in ix~rimaltis or delayed-type hyperss~itivlly. Mice with a combined deficiency in P-solectin and intazCellular Adhesion Moleeule-I (ICAM-I) show a more pronounced deficiency of leukocyte roiling and inflammation than P-solectin deficient mice, suggesting an unexpected synergism between the two adhesion systems. Current data suggest that L- and P-seleetin are rssponsiblc for capturing lenkocytes from the streaming blood, followed by rolling mediated by P- and E-sslectin, E-sslectin may also facilitate det'mitive adhesion, which requires leukocyte integri~s and ICAM-I. Supported by NIH HL 54136 and Miz~tuni Foundation.

A crucial step in hematogenous metastasis is the interaction of the tumor call with the endothelium of a secondan/organ. Consequently, it has been postulated that inducible endothelial adhesion molecules that mediate leukocyte trafficking also are involved in tumor metastasis. Using an/n vitro flow model we have found that a human colon carcinoma cell line, KM12-L4, adhered avidly to 11--113activated human umbilical vein endothelial cell (EC) monolayers, but not to unacfivated EC under condition& Acheslon was signlflcantty reduced by ardl-E-selectln mAb H18/7 and by pro-treatment of KM12-1..4cells with neuramlnldace. Further, KM12-L4 cells express alelyl Le~ end Le', known Iigands for E- and P-selectln. Based on these findings, we postulated that P-selectln supports KM12-L4 initial attachment underflow. In vitro flow assays revealed that Chinese hamster ovary (CHO} cells stably expressing human P-aelectln supported KM12-L4 cell rolling, whereas parental CliO cells did not. However, flow cytometrlc analysis (FACS) using a mAb (reAl0 PSL275, Dr.Dale Cummlng, Genetics Institute) to PSGL-1, a known high affinity Ilgand for P-salecttn, demonstrated that KMt2-L4 cells had undelectable levels of PSGLol expression. FACS also demonstrated that KMf2-L4 cells do not express L-selectin. In summary, the results show that Inducible endothelial-leukocyte adhesion molecules E- and Pselactin can support tumor call adhesive interactions under flow conditions and suggest that KMf2-L4 cells Interact with P-selectin via 9 PSGL-l-lndependont mechanism(s). Suppoded by NIH grants T23-GM08384, HL-18208 & POl-HL36028.

S-48

Cell Specific Bi0materials

233

236 POLYURETHANES CONTAINING COVALENTLY GRAFTED RGD-PEPTIDES

FABRICATION OF A TETHERED GROWTH FACTOR SURFACE P. R. Kuhl and L. G. Cima M1T Del~u'Uncnt of Chemical Engineering, Cambridge, MA, 02139

Horng-Ban Lin* and Stuart L. Cooper

Growth factorswhich exert a broad mitogenic ~sprmse. such as epidermal growth factor (EGF), platelet-derived growth factor (PDGF), and tranHonaing growth factors (TGFalpha, TGF-bets). have been used in disappointingly few clinical pnxlucts, considering the range of effects they produce in via'o. Tran.,qation of the mitogonic effects observed for the ~rget call /n vitro to tissue growth /n v/vo ~ hampered by several issues. Competition with other cells that also ~ n d to the growth factor, and that will overgrow the target cells, is a major problem. Although researchers have attempted to solve this problem by targeting delivery of factoVz at a Specific site, this approach isn't always successful because soluble growth factors can readily diffuse into the blood sue.am and away from the target site, exerting their effects elsewhere. A second problem is lhe complex, nonlinear response cells have to the concentration of grev,,th factor in their environment, especially when one considea'a thc cell response in the time frame commensurate with that required for wound healing. Extended exposure to high growth factor coec~.ntradons may cause cells to lose responsiveness to the factor. A third problem limiting the use of growth factors is that they are expensive.

DeparUnent of Chemical Engineering University of Delaware, Newark, DE 19716 Peptides based on cell-adhesive regions of fibronectin and vitronectin, Arg-Gly-AspSer (RGDS) and Arg-Gly-Asp-Val (RGDV), respectively, were covalently bound to a polyurethane backbone via amide bonds. The polymers studied included a PTMObased polyurethane control, a carboxylated version of the control polyurethane, ~ d three different peptide grafted (GRGESY, GRGDSY, and GRGDVY) polyurethanes. On hydrated samples, X-ray photoelectron spectroscopy (XPS) showed a greater increase in nitrogen concentration for the peptide grafted polymers, which suggests that grafting of the hydrophilie peptides to the polyurethane augments the segment enrichment at the surface. Upon dehydration, the nitrogen concentration dec~..ased for all five polymers, suggesting migration of the more hydrophobia PTMO soft segment to the surface. In vitro endothelial cell adhesion showed an increase in cell attachment on prehydrated RGD-contalning peptide grafted polyurethanes, but not on the other polymers. These results suggest an enhancement of peptide density at the aqueous interface, in good agreement with the XPS studies. * Present address: NJ 07436

Meadox Medicals,

Inc.,

112

Bauer Drive,

The I~'oblems listed above may be overcome if, instead of being defiveaed in soluble form, the growth factors are immobilized on a solid substrata, In this work we have immobilized epidermal growth factor oo a long polymer tether that allows flexibility of the molcoule in solution and permits reg,eptor-ligand binding, but prevents internafization and down-regnlatinn of the ~:eptor-ligand complex. Primary rat hapatecyms were seeded on the tethered EGF su,"facesand on control surfaces that had only nnnspcc~cally adsorbed EGF. The biological activity of ceils grown on the ;etherealEGF am'faceswas significendydifferentfrom that of controls.

Oakland,

234

237

THE ROLE O F E X T R A C E L L U L A R M A T R I X PROTEINS IN THE GROWTH, AGGREGATION, AND MOTRJTY OF PC12 r $ IN A THREE-DIMENSIONAL ~GEN GEL S. P. Baldwin, M. L Kinnann, C IC Krowson, and W. M. Saltznum Del~ranont of Cbemical Engineering The Johns Hopkins University, Bahlmore, MD 21218

MODR=ICATIONOF PLLA S U R F A C E S FOR SELECTIVE CELL ADHESION

When cells m-eslmpanded within gels onntaining exlracelldiar nmn-inpo3teins, the composition and atutetm~ of the matt~ is an impmumt dete.,mimmtof cell bebuviat-and famction, lo these experiments, Pal2 cells (a model for the ceils of the nervous system) were susl~nded in collagen gels, and incubated in the presence of orxve growth fantor (lqGb') for up to twelve days. The collagen gels were modified through the eddidrm of inminin and t'ibrom:r two ~ttracellularmaldx ~ which am known to influence cell-sehatrato intexanticos, and thus adhesion and motility. Ducing tim incubation period,c e ~ move within the matrix, forming aggregates. Qnantitative rnsull~ for the ratoof cell al~e~tdrm wereobtainedthrough the rise of m im~e capoa~ and analysis system. Agg~gate ~ for cells suspended in the tmmodified gels (with and without NGID were compmed to size of aggrogetes incubated in the modified

gels. Inorcean in sine ef the cell al~'egntes wns ~ to be the resdil of ceU migratiort, end not l~Ollferafioa, by the d ~ i n a d o n of cell numbor and thymidu~ uplake. In the i~.scoee of HGF, PC12 cells dilI~llata into a netaal phenotype and develup anmites. The rate of growth of thene antaites wm also correlaled widi gel ~zanture and ndhaslvanees. These studies were designed to ~ i n c the effectof ECMs on the cell.sobsuate intorantirmsl~esant in the collagen ~ These results will aid in the design of matorinls foe roi~ir of CNS injuries.

235 POLYMER SUBSTRATES FOR BIOSPECIFIC IMMOBILIZATION OF ADHESION PEPTIDES Jeffrey A. Hubbell 1, Paul D. Dremheller 2.3, and Diane H e m 2 California Institute of Technology, Division of Chemistry and Chemical Engineering 1 and University of Texas, Dept. of Chemical Engineering 2; Currently at Gore Hybrid Technologies 3 Numerous bioactive peptides and saceharides have been immobilized on polymer and inorganic surfaces to promote the adhesion and spreading of ceils. Some of these ligands are selective in their recognition o f cells, ff a substrata endowed with a selective ligand is also to be selective, the substrata itself must not support cell adhesion or spreading. Such materials were developed, which supported high levels of figand incorporation but very low levels of cell adhesion when inactive ligands were immobilized. Polyethylene glycol (PEG) was dissoIved in hot trimethylolpropane triacrylate with the long wavelength ultraviolet photoinitiator benzil dimethyl ketal (BDMK). When photopolymerization was rapid, e.g. when performed with hikh B D M K concentrations ( > 0.5%), the resulting semiinterpenetrating polymer network was well phase-mixed at PEG concentrations up to 40%. Separation into PEG-rich phases eceurred in networks formed more slowly. The PEG in the phase-mixed networks was very resistant to exlraetion, and the materials were very resistant to cell adhesion. Incorporation o f 6% acrylic acid as a comonomer provided sites for immobilization of peptide adhesion ligands, e,g. the R G D tripeptide. Also, PEG-diacrylate erosslinked hydrogels were formed by photopolymerization of aqueous solutions of PEG-diacrylate. Peptides were incorporated into these otherwise very nonadhesive materials by copolymerization of the peptide after acrylation at its N-terminus.

Ann Park and Linda OfiffithCima

Department of Chemical Engineering, Massachusetts Institute of Technology Cambridge, M A 02139 Surface pmpenies play an important role in tissue regeneration applications becanse tissue regeneration processes are often governed by the interactions of cells with the surface of a b"a'ee-dlmensiona]scaffold. An important factor in the fabrication of degradable scaffolds from polymers derived from lactic acid is the ability to control surface proper~es to obtain cell adhesion to selected regions of the device. Cell adhesion to and migration on selected regions of the surface may gready facilitate cell dis~ibution throughout the device and prevent adhesion of scar tissue during wound healing. Previous studieshave shown that odls euzch to various formulationsof PLMPLGA polymers a few hotws after inoculation. We investigated methods of modifying the ~.trfacnsof well-characterized,sraou~ PLLA films to minimize protein adsorption and cell adhesivity. PLLA films were coated or blended with commereinllyavailable PEO-PPO-PEO tribloek copolymer sud'actan~ with systematically varied chain lengths. Partial treatment of stufaces w ~ pedormed to test for selective cell attachment. Modified PLL,Asurfaces were characterized by XPS and wettability, and assayed for the amount of protein adsorption and cell attachmem. Surfactant.treated PLLA surfaces exhibited significantly reduced protein adsorption and cell adhesion. The degree of abhesivity was dependent on the lengths of the PEO and PPO segments of the surfaetanL Conditions under which selective cell adhesion could be ob~tined were determined. X'PSshowed that signillcendy more surfactancwas relainedat the surface when the su.--'fant,'mt was blended with the surface of PLLA films in an organic solution than when it was adsorbed to the surface from an aqueous solution.

Intracellular Signaling 238

241

GROWTHFACTORTRAFFICKINGANDCELLPROLIFERATION Cartikcya Reddyw Alan Wells~1,and DouglasLauffenburgert w of ChemicalEngineering,Universityof lllinoisat Urbana-Champaign ~[Depanmentof Pathology, Universityof Alabama,Birmingham tDepartmcnt of Chemical Engineering and Center for Biomedical Engineering, Massachnssetts Instituteof Technology

ANALYSISOF MODELSFOR TARGETEDDRUGDISTRIBUTIONIN TISSUES Rceda K. Rippleyand Cynthia L. Stokes Department of ChemicalEnginenting,Universityof Houston, Houston, TX 77204-4792

S-49

Cellular traffickingof growth factors is an importantdeterminant of the proliferative response to growth factor stimulation. We have used the epidermal growth factor (EGF) system to investigate the role of growth factor trafficking in fibroblast proliferation,as assayed on HR6 mouse fibroblaststhat have been transfcctod with the human EGF receptor. The mitogenicresponses elicited by EGF, and EGF-rsriants possessing altered surface bindingand intracellalarfates, are compared ie two distinct experimental scenarios, with or without ligandreplenlshmcat. Dur results show that cellular uptake and processing of these ligandsmodulatesthe mitogenicresponse via a balance between receptor binding and concommitent mitogenic signaling from activated complexes, and the subsequent attenuation of signal due to internalization and degradation of the growth factor molecules. The relative potencies of these ligands are therefore determined by both molecularproperties, and the dynamics of ligand accessibilityas determined by cellular properties and assay parameters. Our results suggest that the choice of optimalgrowth factor ligands for applicationssuch as mammalian cell culture, wound healing therapies and tissue angiueering would depend on specificsystem parameters as well as the molecularproperties of the ligand and its receptor.

The factors that influencethe tissuedistributionof targeted therapeutics includedrug transport and cellularpharmacolofey, which encompasses receptcc/dmg bindingand the processingand routing of the drug withmcells. To examine the importance of cellularpinmnacelogy, l~eVlouslytreated only superficially,we have developeda mathematicalmodel for drug wansp~ ill tiSSUeSthat includesdrag and receptor intel~aliw~tion,recycling,and degradation, es well as drag diffualonin the cxtracolluisrspaceand bindingto cell-surfacereceptors. Dimensional analysisrevealsthat the modelconsislsof a "fast" subsystemof equations for the coll-ussociatod variables(e.g, unboundsm'faneand i n ~ n i a r receptors, d~g-complenedsmface and inuacelloiarreccptms) and a "slow"subsystemequonce that describes extrscollulm"&ng trauspo~ Thus, cell-usanciatedvariablesof the fast subsystemme at pseudo-steady state at any giventime duringdrug diffusioninto the tissue. A steady state analysisof the fast subsystem showed that the cellularpharmacologycould be complctolydescribed by only two Wobeblltties which are combinationsof the kinetic peramete~ for a given ~netic scheme." the probabilityof being bound (as opposed to tmbetmd)~ the pmhabtiityof being on the cellsurface (as opposed to intracelluiar). The steady state valuesof cell-associntedspeciesare always givenby the same combinationsof these probabilities,regardless of the initialkinetic scheme of celluI~"muting and processingof drug. The effect of cellal~ pharmacology,us de,tee'minedby the two probabilities,was incorp~ated into the modelof&ug transport in the overall tissue. This allowedus to reduce the effects of cellularph~macoingyfrom multiple-pacameterto parameter space. The additionalgoverningparameters are di~g thffusivity,initialdrug concentration and receptcc density. The specifickineticscheme selected has someiofinceceon the drug U'ausportand distributionas it contributes tmns to the slow subsystemrate equation. work providesa generalized viewof the effectsof cellularpharmacologyon drug distrihutionin tissues, and providesa simplerway to predict a thug distriboticewhen the kinetics of a given system'sendocytic cycle are known. For instance, one might dete~mionthe effect of changinga particularkinetic parameter on drag distributionsimplyby evaluatingthe effect of the resultingchange in probabilities.

239

242

IMPORTANCEOF SPATIALINFORMATIONIN SIGNALTRANSDUCTION: A MONTE CARLOSTUDY LrmnieShce and JenniferLinderman Universityof Michigan,DepLof Chemical Engineering.Ann Arbor, MI 48109

KINETICS AND PHARMACOLOGY OF PANCREATIC iS-CELLVOLTAGE-GATED CALCIUM CURRENTS/N SFFU D. Meare, N.F. Sbeppard, Jr., I. Atwater* and E. Rojas* Johns Hopkins Univ., Baltimore, MD21218, *NIDDK. NIH, Bethanda, MD, 20892

Mathematical models describingsignal transductionoften assume that the moleculesin the plasma membrane are homogeneouslydistributed; however,in some systems,an inhomoganenns distributionof molecules in the plasma membrane may be an essential aspect of the signal transducdon mechanism. We propose using Monte Carlo techniques to simulatethe reaction and diffusienof molecules in the plasma membrane in ct'dcr to incorporate spatial informationabout the molecules. Althoughthese techniques require more computer time, the spatialinformaticeprovidedmay be necessary to fully underamnd the mechanismof signal transductico. We have used Monte Carlo techniques to examine spatial inhomogcecitlesthat may form during the process of signal transduction. The diffusionof reccptor-ligand complexes to activateG-ixoteiusand ti~ clusteringof receptors by a multivalent ligandan: two aspects of diff~t signaltransduetionmechanismsthat may produce spatial inbemogcecities. We have determined the magnitudeof the effect that these spatial inhomogencitieshave on the response by comparing the simulationresults with models that usanme a homogeneousdlstributloe. Significantdifferencesbetween the two are seen, especiallyat low rates of diffusion. Additionally,we will Inusent resultsfrom simuinticesof two examplesof membrane organization:microdomalns and precoupling. We fred thatthese two phenomena can significantlyaffectthe response and may be importantaspects of the signal[ransduction mechanism.

Calcium influx via voltage-gated calcium channels is a crucial etep linking glucose stimulation to insulin secretion in the pancreatic b-cell. We have studied the kinetics and pharmacology of b-cell calcium currents under physiological conditions by voltege.clamping cells within the intact islet of Langerhans. Currents in response to depolarizing voltage pulses conaletsd of a fast inward component followed by a delayed, sustained outward enmpouenL When the outward component was blocked by adding 20 mM tetraethylammuaium to the perifucate, the inward current consisted of a fast component with activaticu-inactivaticu kinetics and a sustained component. Both components were reduced, but not abolished, by nifed/pino, and were enhanced by BAY If. The inward currents were completely suppressed when extracellular calcium was replaced by magnesium. The extent of inactivation of the initial component Was correlated with the amount of calcium entering the cell, and incubatiua of the islet in the calcium chelator BAPTA partially abated the inastivatien. These observations demonstrate that the rapid inactivation of the inward current is calcium dependent, During long (1-10 see) pulses, the sustained cempenent displayed a gradual, voltage-depondent inactivation. Thus the in o/tu voltage damp technique allows study of b-cell membrane currents under conditions that are as close to physiological as can be achieved in v/tro. Our results indicete that the mDjority of the voltage-dependent celcium influx in the b-~ll is through Ltype, dihydropyidine sensitive channels, although other calcium channel types may also be present in the membrane.

24O

243

CHANGES IN ENDOTHELIAL CELL CONTACT AREA AND TYROSINE PHOSPHORYLATIONFOLLOWINGEXPOSURETO FLOW Lami A. Olivles,Yao Xino. Jesse Yen, W~I. Reichert, and Ge~gn A. Truskey Deparlmcet of BR~edical Engineering,Duke Ualvereity.Durham, NC 277(~-0281

RECEPTOR-MEDIATED CYTOMECHANICAL MODEL OF DYNAMIC MORPHOLOGY, RANDOM AND BIASED TURNING OF LEUKOCYTES R.T. Trenquilio and Walfgang Aft* Department of Chemical Engineering and Materials Science, University of Minnesota and *Department of Theorcticel Biology, University of Bonn

Alteratio~ in the fimctlon of ~ cedothcliam by fluid shear stress may influcece the pathogenesisof ad~esd~o~s. The mechanismby wMChflnid chcar str~s is wsnsmiuedto the eodothalinm is onlmown. In this stody, thn dynamicrcslxmse of cedothclialceU focel centacts to flow/n v/tre was examined with total internal flura~cence reflectionmiero~opy (TUII~). Thn cell meminaneof bovi-e a~tic mdothciislcollswan labeisdwith the c~bocyseloe dye I,I 'd i ~ _ ~ 1 - 3 ~3,3~Y-tetrm~thyl-~doenrbocys~e pe~chlorate and cells were allowed to sptend onto a fibmcocfln-coatod surface. Changes in the separation distance we~ quantified by celcalafi~gcmes-oerrelafi~3cocifieientsof the ssparatinndistancesfor cells mdcr stoflcsed flow conditiol~. C O ~ ~ c ' ~ t s p[ovidc I I {M ~tlmtt~of I ~ r to whlck Enw ~ t~ spatialn:istlonskipof comacts, wi~ unity dr no allea. All celis expeticeced modest cell contour flucgmti~ imdcrstatic coadillcas. Changesin cell con~u" movemcetincrca~xl upon tmposllionof flow, withInamsed deccu~intinaat I~gi~crwall sbeac stresses. The ra~ of cha~e of thn c o c ~ o u coeff~m lmmedimelyafla llowwm algulflcaatlydiffenmtflnm t l ~ prior to flow at 10, 20 and 30 dynes/era2, demomtrating a cellalat response at those shear stranan& Remodel~g of indivithmlCOnL~I~w~ studiedby t t ~ l ~ g focal coatect ~ sud number over time. Focal comact respoase varied widely among the cells. At a wall shear stress of 10 dynegcm2 little change was observedin focal contacts followingimpositionof flow. At wall sheer stresses of 20 and 30 dyl]r the most comm~ n~ponse apon impo$ifioaof flow was a deceeuse in the atce and number of c~atacts followedby folmation of more contacts with smaller areas than Ineflowcoodition~ Oangco in focal contacts were coneinted with tyruslne pbosph~ylafion of focal contact proteins. The results seggest ~ the hydrodynamic force is not Uausmlttedalong the cell membranc. A more pinusiblc hypothesis is that the force is ~smittod along the actin cytuskeleton.

The cytomochanical component of our model describes the distribution of actin and forces in an idealized cortical actin network around the cell periphery in which one or more lamellar regions of may exist (an account is made for forces associated with myosin-mediated contraction, network swelling, cortex-membrane curvature, and cytoekeletaI friction opposing network flow). Cell tracks are related to a net torque associated with network flow aromnd the actin cortex mediated by cell adhesion and a resultant force associated with centripetally-dirocted network flow in lamelllpods alan mediated by cell adhesion (beth cell adhesion and cell shape are presently taken to be funetionale of the local cortical actin concentration). The sensory component describes the distribution of cbemotact, ic receptors in the cell membrane surrounding the cortex, where bound receptors give rise to an intraceliular signal-which modulates some property of the cortical network. As in our earlier models, an account is made for statistical fluctuations in receptor binding, entirely determined by receptor binding rate constants and the local CA concentration field, but generalized here to include statistical fluctuations in the spatial distribution of receptors, entirely determined by membrane diffusion coefficients. We characterize via stochastic simulation and cell track analysis the cell speed and directional orientation under the influence of such fluctuations by making the regulated concentration of assembled actin in cortical cytoplasm a function of the local density of bound receptors on the cell membrane. This is done for both uniform and steady spatial gradient concentration fields of chemotactic factor.

S-50

Hybrid Artificial Organs

244

247 AN IN VIVOMODEL FOR TISSUE-ENGINEEREDBONE FLAPS MJ Millert, DP Goldherg',AW YaskoI, JC Lemont, WC SotterfieldI, and AG Miknsa ' The Universityof Texas M.D. AndersonCancer Center and the a Cox instituteof Binscienceeand Bioengineering,RIce University,Houston, Texas

PURPOSE: Weprosentanmvivomodeltoassessthepotentialofbonesub~imteefurcroafing vascularized bone flaps. METHODS: Polymediyimethacrylate(PMMA) chambers enclosed on 5 sides with a volumerangingfrom 4-9 cm~were fabricated of vurinesshapes. A cuffmede from expanded polytetrafluoructhyloneor polyurethane was bonded around the perimeterof the open side, Four db segmentswere removed from alternatinglevels in 7 domestic sheep, leaving the poriosteum intact. Chamberswere implantedwith the open side secured to the periosteum, sealingthe chamber and permittingtissue ingrowthonly from the perinstanm. Someimplants were completely filled with moreelizedcorticocancellonsbone graN (MBG), and the others were left empty. Implantsfrom 4 aalmalswere harvestedat 6, 9, and 13 weeks, and the contents were examined grosslyand by histology. In 3 animals,chambers were removedat a second operation and formedbone was tmnsfcrrodas p~:licled flaps to adjacent rib levels. RESULTS: Chambers filledwith MBO yieldedformed molded blocksof vascelarlzodbone s~ached to the p~in~onm and conformed to die shape of the chamber. Activebone formationand vescolar ingrowthwere histologicallyconfinded. Specimensharvestedearly contained necrotic bone graft incoq)orated by new bone precisely matchingthe shape of the chamber, Later specimensshowed a greater proportion of vitalbone with evidenceof remodelingthat resulted in a less exact shape. Molded hone segments failedto grow in empty chambers. Trnnsfenedbone flaps demonstrated good vascularityat the time of the second operation. CONCLUSIONS:PMMAchambers implanted against sheep rib poriosteum providca useful in v/vomodel for controlledosteogenesis. Boundary conditionsfor the model may be establishedusing corticocsecellonsbone graft as the optimum physiologiccondition, This model may be used to determine thc potentialof varlons hone substitutesfor creating"pre-fabricsted" vascularizedbone flaps.

245 BIODEGRADABLE POLYMER SPONGES FOR (~.t t TRANSPLANTATION D.J. Mooney*, J. Rowley*, J. Nikolovski*, J.P. Vacanti e, and R. Langer+ *Depts. Biologic & Materials Sciences and Chemical Engineering, University of Michigan; *Dept. Chemical Engineering, Massachusetts Institute of Technology; 'Dept. Surgery, Harvard Medical School. Porous sponges fabricated from biodegradable polymers can be utilized to transplant eelis and engineer new tissues. We have fabricated porous sponges from polymers of hctic and giycolic acid u~li~ng both a p a r t c ~ a ~ leaching technique and high pressure CO 2 gas foaming. The porosity and pore structure of sponges formed with a particulate leaching technique can be oontrolled by varying the ratio of polymer/partlculate, and sponges with porosities up to 95% can be fom,cd with this technique. Highly porous sponges can also be fabricated by allowing solid polymer discs to equilibrate with high pressure COs gas and then creating a thermodynamic instability by rapidly decreasing the CO2 gas pressure. The CO 2 dissolved in the polymer discs phase separates, and forms pores within the polymer matrix. This process is advantageous as it allows sponges to be formed without the use o f organic solvents. Sponges with porosities greater than 90% can be fabricated with this technique. Infiltration of the sponges with surfactants (e.g., polyvinyl alcohol or block copolymers o f ethylene oxide and propylene oxide) reduces the devices' hydrophobicity and enables efficient and even cell seeding within the devises' pores. These sponges have been utilized to transplant hepatocytes, and engineer new liver-like tissues in experimental animals.

246 UROLOGIC ORGAN EECONSTITUTION IN TISSUE ENGINEERING

A. A t a l a , M.D. Children's Hospital and Harvard Medical School

The d e v e l o p m e n t o f a s y s t e m f o r t h e r e p l a c e m e n t o f s p e c i f i c f u n c t i o n a l u r o l o g i c o r g a n s has become a p r i m a r y f o c u s i n o u r laboratory. T e c h n i q u e s o f t i s s u e h a r v e s t , c e l l e x p a n s i o n and r e g u l a t i o n , and m e t h o d s o f c e l l d e l i v e r y a r e a l l i m p o r t a n t components r e q u i r e d f o r s u c c e s s f u l t i s s u e r e g e n e r a t i o n . . W e have u s e d b i o d e g r a d a b l e p o l y m e r s a s c a r r i e r s f o r b o t h t i s s u e IxQplants and c e l l i n j e c t i o n s . Our e x p e r l m e n t a l work h a s demonstrated that organ reconstltutlon and function can be successfully accomplished in vlvo using this technology.

THE USE OF ALGINATE A N D AGAROSE IN CEIL ENCAPSULATION: A COMPARISONOF IN VIVOPERFORMANCE AND OF IN VITRO STABILITY Molly S. Shoichet,t MellssaL. White,2 ShelleyR. Winn2 IDepartment of Chemical Engineering aed Applied Chemistry, University of Toronto; 2CytoTherapestics, Inc. 2 Richmond Square, Providence, RI 02906 In cellencapsulation, a mauix materialprovidesceils wits the appropriate envirnmncotto sustaincellviabilityand seerer release of bioachvemolecules. The matrix mstedal can provide a distributionof cells withinthe device and may maintainits stabilityover tho lifetimeof the devioe. In the/n vlvostedies, algianteond agamsc were o~apared as die matrix for tho immobilizationof calf ad~nal chromaffm(CAC)cell wasponsionswithinamylic hollowHl~r meanbrenes. CACcells delivercatecholaminesand enkaphalim which may be an effective Ircetmcnt for chronic ~ whilethe hollowfiber membrane providesimmtmo-lsofatiou. The in ~ro studiescomparedthe stabilities~ agarose and alginatoin the wesencc of cells and relativeto controls. The single~table of both in v/voand in vitro studies was the mat~ material. EncapsulatedCAC cells wct~ tmptatueAbtisturallyin the ventriclcof rats for t and 6 monl~ and assayed pre-imFlastasd Post.~xplastfur seczetuwteinaseof norepinel~rine(HE)~ epinel~u~e (EPD~ than histologicallyseerione~ The stginste-and agamee4lllodcspsul~ showod similarbasaland nicotine-evokedrele,,~ of NE and Eli from we-implantto I monthpostexplmu. The concentrationof ofcottno-evok~NE and EPI rekased into the assay modimnwas greater for algiantewhco compared to agarnee-rdled capsulesat 6 monfllspost-explantoaly. Alginatoshowedincreased levelsof NE and EPI at 6 mouthe rolafiveto agarose; agaxoseshoweda consistentprofileof catochelaminerelcase ovcr die 6 mondiperiod. Histologicalsectionsshowed that alginatoand agarose were equivalentwhen sca~edfor CAC call aggregatedlstu'butico,lgxcetu CAC cell viabilitywithinthe aggregagesand percent of nma-CACcells present withinthe devices. In vitro stabilitywas assessedover 90 days by measuring gel strength. The ~ gel strengthof agarose was affected by the pmscoee of ceilswhere~ that of alginstewas not. Gel slxcogthsof Ix3thmaterialsdecreased overtimewith and withoutthe weseuce of cells.

248 Detachable epithelial monotayera cultured on thermally reversible poly(N-inopropy! scrylemlde) costed subatretes Homt A. yon Rscum~, TeNo Okano~, Sung Wan Klms I Department of Bloangloesdng, Unlvemky of Utah, Salt Lake City, UT 2. Institute of Biomedical Engineering, Tokyo Women's Medical College, Tokyo, Japan a Center for Controlled Chemical Deliver/, Department of Pharmaceutice, University of Utah, Soil Lake CBy,UT Tissue Englneering is providing us with the means of transplanting many cell types with the goal of restoring lost function. Epithelial cells, however, show a distinct orientation in vlvo, which Is difficult to maintain or restore following transpinntaflon of cultured cells. Traditional transplantation theraplse usually involve Injection of a s;,spanslon of cells, but the random orientation of these ceils prevents optimum attachment and vlahliky. These cells also lack the call4o-ceit Interactions which may be V'~alin maintaining proper function. A solution to this problem is to transplant 9 sheet of cells grown on an implantable aubstrota. Cells In such a construct show proper orientation and interactions, but the ipproach Is limited by blomatarisl compiinstions such as encapsulation and foreign body response. We propose a method by which a confluent sheet of cells is grown on a suitable substrata and can be easily detached. This "cell sheet" is then l~'ensptantod without implanted scaffold, while malotalolng cellular orientation and loteractionc. Our study propose~; the use of a thermal senshlve polymer (poly(N-lsoprapyl act/tam/de) or poly-NIPAAm) as a substrata for cell attachment and growth. A mechanistically thermal reversible polymer demonstrates phase change as a function of temperature. Above the Lower Critical Solution Temperature (LCST) this polymer remains insoluble, but when the temperature is brought below the LCST it anlubilizas or hydrates depending on croas-linklog, causing the cells to detach from the subsVate without chemically or muchanlcally cleaving extracellular attuchment proteins. Corneal endothelium, Intestinal epithelium, end retinal pigmented epithelium were cultured on substrates coated with poly-NIPAAm which had an LCST of about 32*(::. Below the LCST the cells detached in large patches which were easily transplanted to other culture dishns where they attached and continued to grow In a cohesive sheet

Wound Healing 249

252

CHARACTRRIZING THE ROLE OF' F'IBROBLAST MOTILITY IN AN IN VITRO ASSAY OF WOUND CONTRACTION Bruce A. Bcemberek, Victor Barocas, Oayle Hennig*, Michael D. Coldwell* and Robert T. Tranquillo Departments of Chemical Engineering and Materials Science, and *Surgery, University of Minnesota

ACCELERATED ARTERIOPATHIES: POLAR FORMS O F W O U N D HEALING E.R. Edelman, Brigham and Women's Hospital~ Harvard University and Harvard-MIT Division of Health Sciences and Technology

An important aspect of the repair response in dermal wounds is the ability of fibroblaste to infiltrate the wound via migration from the aurrounding dermia. In certain wounds, this flbroplasia is correlated with wound contraction. Our novel fibroblast-populated mincesphera (FPM) Wound assay permits the systematic investigation of the interplay between cell infiltration and tissue mechanics, which underlies dermal wound contraction. The FPM consists of a sphere of reconstituted fibrin gel with concentric outer shell and inner core regions. Using a double capillary injection apparatus, dermal fibroblaste (DF) are initially dispersed in the shell while the enre is initially devoid of cells but contains a dispersed stimulus, thereby mimicking the initial state of a dermal wound. Over time, the ceils compact the FPM as they begin to populate the core via migration and division. In the absence ofo core stimulus, the DF exhibit circumferential orientation due to an apparent contact guidance responso - the FPM compaction is inhomogenanus due to the imposed nonuniform initial DF distribution, inducing circumferential orientation of the fibrin fibrils and thereby a DF contact guidance response (thus, the assay is also a model system for studying a basic morphogenetic process). In the presence of a core stimulus, in the form of TGF~-treated resident immune cells, the DF exhibit radial orientation due to an apparent ehemotactie response that dominates the contact guidance response. The extension of the assay to controlled release stimuli is discussed.

5-51

Cardiovascular diseases account for over 1 million deaths, and more than 100 million hospital days each year in the United States. Almost 1.5 million interventional procedures are performed for attempted relief o f the complications o f these illnesses. Yet, virtually any mechanical intervention designed to limit or alleviate atherosclerotie vascular disease is beset by accelerated vascular disease of its own. The biology o f these lesions is strikingly similar to the biology of wound healing. We have identified three critical phases of the vascular response to arterial injury that involve three distinct forms of wound repair. Initial acute injury to the endothelium abrades and denudes this lining removing the biochemical and mechanical filtering it provides, followed soon by a substantial inflammatory response primarily involving monocytf/macrophages. A second, subacute healing phase involves the proliferation of smooth muscle cells, presumably in response to t h e cytokines and growth factor elicited by the early events. Finally a more chronic response is observed, wherein hemedynamic flow alterations lead to deposition of matrix, collagen, elastin etc. Late remodeling encompasses not only continued migration and growth of smooth muscle cells and giant cells but o f fibroblasts as well. Only continued study of the fundamental cellular and hemodynamic events surrounding vascular injury and repair will shed light on the biology of accelerated vascular diseases and provide a rational approach to their cure.

250

253

H Y D R O G E L BARRIERS TO BLOCK THROMBOSIS O N THE DENUDED ARTERIAL SUBBNDOTHELIUM ALTER MEDIAL bFGF RELEASE Jeffrey A. Hubbelli and Jennifer L. West 2 California Institute of Technology, Division of Chemistry and Chemical Engineering I, and University of Texas, Dept. of Chemical Engineering 2

PERIPHERAL MONOCYTEMODULATIONOF CENTRAL NERVOUS SYSTEM WOUND HEALING C.W. Pa~ck Jr., S. Kni~eti, and L.V. Mclntim Cox Laboratory for BiomedicalEngineering, Rice University

Catheter-based angioplasty procedures, such as balloon angioplasty to open an obstructed coronary artery, damage the artery and denude its endothelium Thrombosis ensues, resulting in a platelet-fibrin matrix upon the treated artery. Such injuries induce the thickening of the intima to form a fibrocellular neointima in a process that may be related to restenosis. Total blockade of blood contact by the photochemical formation of a polymeric hydrogel barrier approx. 20 I*m thick on the surface of the injured artery (barrier formation described in Hill-West etal., PNAS 91, 5967, 1994) completely prevented thrombosis and platelet monolayer deposition in rats. The amount of intimal thickening measured at 2 wk was reduced by 91%. Local release of two platelet-dedved species, thrombin and PDGF, did not reconstitute the effect of platelet deposition in inducing intimal thickening when blood contact was prevented with a barrier. Injury liberates bFGF from beparan sulfate-containing proteoglycans in the arterial media, and this liberation has been shown to be maximal at 4 d. The amount o f bFGF remaining in the arterial media was measured by ELISA, and blockade of platelet deposition with a hydrogel barrier reduced the amount of bFGF liberated by 84% relative to injured control arteries without barriers. These results suggest that platelets play a key role in intimal thickening by releasing a factor that liberates bFGF from the arterial media.

251 FLOW-ASSOCIATED MODULATION OF THE ENDOCAM (CD31) GENE IN CULTURED BOVINE AORTIC ENDOTHELIAL CELLS Eugene A. Sprague and Jeanette Mower,/ University of Texas Health Science Center at San Antonio EndoCAM or CD31 is an integral membrane glycoprotein located on the surface of platalets, leukocytes, and endothelial cells. Specifically in endothelial cells, EndoCAM is concentrated at intercellular junctions and is postulated to play a key role in mediating endothelial cell to cell adhesion. Endothelinm in focal arterial areas prone to develop atheroselerotic lesions is exposed to chronic low shear, reversing flow eonditons and is further characterized by increasr permeability to macromoleeules and increased cell turnover. The present studies were designed to examine the effect of prolonged low shear on the expression of the gone regulating EndoCAM as an important constituent of the endothelial cell-cell junction. Confluent monalayers of bovine aortic endothelial cells (BAEC) residing on polyester sheets were placed in parallel plate flow viscomcters and exposed to either no flow (control), low (0.5 dynes/era 2) or high (30 dynes/era 2) shear stress under continuous steady flow eondtions for times ranging from 1 to 72 h. Upon completion of flow regimens, BAEC were removed for RNA isolation and analysis of EndoCAM gene expression by northern analysis. Our results indicate a significant time- dependent decrease in EndoCAM gone expression in BAEC exposed to low shear stress for periods beginning at 6h and extending to 72 h relative to cells exposed to either static or high shear flow conditions. Thus, our results indicate that prolonged exposure of BAEC to low shear stress is associated with decreased expression of the EndoCAM gone. These data may have important implications related to the enhanced endothelial permeability observed in early atherogenesis.

Wound healing of the central nervous system (CNS) is a complex process involving inte~,.ctioas between cells from both the v~eniar and nem-alenvironments, exlracoIlular malrlx proteins, and a cocktail of agonistic and antagonistic bicoctive molecules. Vascular celts, particularly monecytes and macrophages, are believed to play an important role in organiadngand mediating CNS tissue reactions afro" the o c c ~ of penetrating injmico (i.e., injmies that compromise the blood-brain bani~), although the mechadiswa by which they do so are poorly understood. To begin to elueidale these mcohanianl$, we have investigated the effect of pariphelni monocytes an nsaral mgenesation. We have utilized the hNT neural cell line, which possesses many of the phenotypic neJ~al properfie~ of the human CNS, and an/n v/tro CNS model to investigate famdsmcotalaspects of monocyto-nsmon imemetion~ Cell cnitm~ video micmsenpy, maddigital image ~ i n g am coupled to provide a qnsntitative descrilzion of the effent of mom3cytesco neural regeneration. Briefly, this ~ t s of viewing individual neurons using a CL"D coupled to an inveaed m ~ and digital image In'ecessing module. A computer-coatrolled stage mounted on the mim'osenpe allows precise stage movement, field-of-view solectine mui stefage, and the abUifyto renmt to l~eSniected fields. Analysis of the neorous cco.~sts of digitally u'ar.ing the outline of the previously digitized neurons (ceil bodies and neurltes). Paramele~ me.uumd include neuritic output, arbor output, mean arbor length, semite branching, arbor.henries ratio, and effective cell body diameter. The perameters obtained f~m mon0eyteconditioned media were compared to those under control conditions, as well as those obtained from media containing nerve growth factor. Results show that monocytos affect neural regeneration by increasing anuritic output, arbor output, mean arbor length, and neurita tranching, and decreasing the atbocncorim ratio.

S-52

Poster Presentations

254

257

PHYSIOLOGICALLY BASED KINETIC MODEL OF E F F E C T O R C E L L BIODISTRIBUTION: IMPLICATIONS FOR ADOPTIVE I M M U N O T H E R A P Y Laurence T. Baxter, Robert J. Melder, Hui Zhu, and Rakesh K. Ialn. Department of Radiation Ontology, Masseehuseus General Hospital and Harvard Medical School, Boston, MA 02114

IMPLANTATION OF COLLAGEN-GLYCOSAMINOGLYCAN MATRICES IN CHONDRAL I ARTICULAR CARTILAGE DEFECTS IN A CANINE MODEL

A prerequisite for successful cancer therapy is the optimal delivery of the therapeutic agent to tumors while minimizing normal tissue toxicity. While drug distribution has been extensively studied, relatively few data are available for the the kinetics of distribution of cells within the body. Physiologically based kinetic modeling has provided insight and guidance in the delivery of various therapeutic agents for cancer, and also been successful in scaling up biodistributinu from mice to humans. We therefore wanted to extend this approach to particles and have developed the Erst physiologically based kinetic model for cell biodistribtuinn to describe the following data: i) nonactivated T lymphocytes in rats, ii) interleukin-2 (1L-2) activated tumor inflitratiag lymphcoytes (TIl.s) in humans, ill) nonactivated natural killer (NK) cells in rats, and iv) 11.-2 activated A-NK calls in mice. Model simulations revealed the similarities as well as differences in biodistribution for different lymphocyte populations as results of their trafficking properties. Specifically, activated A-NK cells exhibit a preferential adhesion to tumor vascalatore, while activated TiLs show no such preferential adhesion. The low effector concentrations in the systemic circulation due to normal tissues re~ntion greatly limited their delivery m tumor. Reducing attachment rate or adhesion site density in the lung by 50% could increase the tumor concentration up by 40% for TILs and 60% for A-NK cells. Increasing tumor blood flow, though could lead to a higher A-NK accumulation, may not increase TIL concentration in tumor. Strategies suggested to improve the effector cell delivery to tumor include: i) bypassing the initial lung entrapment with local administration in a tumor artery, ii) reducing normal tissue retention using effectors with high deformability or blocking local lymphocyte adhesion, iii) enhancing tumor specific capture and arrest by modifying tumor microenvironment.

Breinan. H.A. 1, Hsu, H.-P.4, Ramappa, A J,3, Yannas, I.V.2, Spector, M.4 IDivision of Health Sciencesand Technology, and 2Department of Mechanical Engineering, Massachuseus Institute of Technology; 3Harvard Medical School; 4Department of Orthopedic Surgery, Brigham andWomen's Hospital Recent clinical and experimental work has demonstrated limited regenerative capabilities of articular cartilage. Few studies have examined healing in well~lefiecd chondral defects that model clinical cases in which the subchoodral bone remains intact. These chondral defects limit the effects of anbchondrsl vasculatore and marrow on healing. In previous work, we developed a canine model of the choedrsl defect and demonstrated limited repair of untreated defects at early time periods. In agreement with previous authors, defects filled gradually with fibrous tissue. Portions of the reparative tissue at the base of the defect demonstrated early changes suggestive of differentiation into a hyaline material. The objective of this stUdy was to evaluate regeneration in choodrsl defects treated with specific porous, resorbahle collagen-glyeosamlnoglycan (CG) copolymers serving as anaings of extracelhilat matrix. We hypothesize that the C-'Gmatrices will facilitate directed cell infiltration of the defect and differentiation of reparative tissue into hyaline cartilage. CG matrices were implanted in choedrsl defects in the trochlea of adult dog knees. The grafts were fixed by suturing an overlying autogenous fascial flap to adjacent cartilage. Control defects received the fascial cover only. Our canine model includes immobilization of the implanted limb through external fixation for 10 days, after which the dogs are allowed unrestricted movement until sacrifice at four or fifteen weeks. Regeneration is evaluated by staining with safzanin O and H & E, as well as immunohistechemical staining for types I and I1 collagen. This work was supported in part by the VA Rehabilitation R & D service and by the Department of Defense.

255

258

FLUORESCENCE PHOTO8LEACHING AS A NONINVASlVE PROBE OF INTERSTITIAL TRANSPORT IN TUMOR TISSUE David A. Be.A, Fan Yaan. Michael Leuaig, and Rakesh K. Jain Depmlment of Radlalion Oncology, MassachusettsGeneral Hospital, Harvard Medical School. Boston. MA

MASS TRANSFER STUDIES OF TISSUE ENGINEERED CARTILAGE

The mobility of macromolecules within the extracellular space oftissue can be a critical issue in drug delivery and tiSsue engineering problems. We have devised an image-based fluorescence recovery after photoblcaching technique that allows direct measurement of transport in the interstitial space of tissue. The combination of spatial and temporal resolution reveals the ioterstitial motion of floorsscentiy labeled macromolecules in sufficient detail to allow measurement of the effective diffusion coefficient, binding kinetics, and convective velocity. We measured interstitial transport parameters for various proteins in a human tumor xenograB. Serum albumin, intact monoclonal antibody (IgGl) and Fah' fragment all exhibited similar levels of hindered diffusion (D/D0 - 0.3) and nonspecific binding (bound fraction - 0.3). A tomor-apecific (anti-CEA) monoclonal antibody displayed a dose-dependent binding curve that suggested a binding site (tensity of approximately 106 per cell, consistent with CEA expression by the same tumor cell line in vitro. A surface binding model for bivalent binding was used to compare the bindiog saturation curves of intact specific antibody and its mOnovalent Fah' fragment. The results indicate that in the absence of binding, diffusion is an adequate mechanism for distributing proteins from sites of extravnsation throughout the tumor. However, nonspecific and specific binding can have a major effect on the pha~ancokinetics and intmtumneal distribation of proteio. Enhanced drug delivery could potentially be achieved with the appropriate choice of binding kinetics and the minimization of nonspecific binding.

Cartiinge tissue constructs for potential application in reconstructive surgery can be grown in vitro using on isolated cartilage cells and biodegradable polyglycolic acid scaffolds. In the present study, the kinetics of nutrient mass transfer and regeneration of tissue components (i.e. calls, glycosaminoglycan, collagen) were studied for constructs grown in mixed petri dishes for up to 6 weeks. Mass transfer rates were measured by exposing a construct to a small volume of a well mixed solution of tracer. Two tracer molecules were used: glucose (MW = 180) and dextran (MW = 4,400). A transient diffusion model was developed and used in conjunction with physical construct properties to assess the kinetics of mass transfer, Overall mass transfer coefficients decreased to 10 % of initial and 1 % of initial for glucose and dextren, respectively, over 6 weeks of construct cultivation. Strong correlations were observed betwe#n the decreases in mass transfer coefficients and the amounts of tissue components in the constructs.

P.M. Bursec, G. Vunjak-Novekovlo, R.J. Biron and LE. F~,~d Massachusetts Institute of Technology, Cambridge MA 02139.

256

259

A NOVEL "TEMPORARY GLUE" FOR E N D O T H E L I A L CELL SEEDING OF SMALL DIAMETER (< 6 mm) V A S C U L A R PROSTHESES. G a r y L. Bowlin, M.S. and Stanley E. Rittgers, Ph.D. T h e University of Akron and The Willlata H. Falor Center for V a s c u l a r Studies, Summa H e a l t h System, Akron, OH.

PRESENCE OF MYOFIBROBLASTS DURING PERIPHERAL NERVE WOUND HEALING Chamberlain, LJ., Yannas, IN., Landstrom, A.E., Hsu, H.-P.', Specter, M.t Fibers and Polymers Labozatory, Massachusetts Institute of Technology * Orthopedic Surgery, Brigham & Women's Hospital; Breckton/W. Roxbury VA Medical Center

Current endothelial cell (EC) seeding t e c h n i q u e s use various "glues" to enhance EC adhesion. A novel technique using a temporary p o s i t i v e surface charge ("temporary glue") on the g r a f t luminal surface to enhance EC adhesion and m a t u r a t i o n was induced by p l a c i n g the graft b e t w e e n an internal and external conductor. An electrical p o t e n t i a l was then applled across the conductors and the graft treated as a d l e l e c t r i c material in a cylindrlcal capacitor. The entire EC s~eding apparatus is rotated at 1/8 R.P.M. and provides an even EC seeding density. The results indicate that the "temporary glue" (1.0 V) leads to a significant (p<0.05) increase in the EC seeding density (77,054 E C / C m 2) versus a gravitationally seeded PTFE g r a f t (49,389 EC/cm') at a seeding time of 16 minutes. The m a t u r a t i o n phase is also e n h a n c e d p r o d u c i n g more flattened ECs w i t h the "temporary glue". These results indicate that the electrostatic adhesion of ECs may lead to improved p a t e n c y o f small caliber vascular grafts.

Myofibroblasts have been identified in certain connective tissues as important participants in the contraction of healiag wounds and in selected pathological contractares. In addition to containing the apparatus for protein synthesis, the myofibroblast displays the actin isoform, a-sulooth muscle aetin, specific for smooth muscle cells and pericytes. In vitro studies reported in the literature have demonstrated that axoual elangatico can be directed by tensile forces applied to individual notwites. That work provides the rationale for the lXCSenceof cells capable of applying forces for the "towed growth" of axons in the regenerating nerve. "['neobjective of this study was to determine ff myofibroblasts play this role in the healing peripheral nerve. The sciatic nerve of the rat wus wansected and a 10 mm gap created. The cut ends were eusbeathed in either a collagen or silicone tube. The tube was leR empty, or filled with a pemus colingua-glycosaminoglycan copolym~. Tissue samples were allocated for immunohistochmnical staining with antibodies to a-smooth muscle aclin or for tr~nsmi~alon electron miorosCOpy, thgt reveals ulwastructm'al4eatores unique to mynfilxoblasts. Myofibroblaats were found adjacent to the tube and within the defect site. Around the robe, on boththeinner andou~ edges,mynfibroblasts apl~at~Ina ck~mmfcrenlialdimcOonand formed a prominent monolayor adjacent to the silicone tubes. In the defect site, the cells were aligned with the long axis of tiae nerve and appeared in grea~st numbers toward the distal stump ahead ofthe advancingnervegrowthcone,withno diffmenccquatitutively betweua the empty and filled tubes. This is the fwst report of myofibroblasts in the healing nerve. The presence of these cells in the healing peripheral nerve supports the hypothesis that myofihioblasts could be participants in the elongation of axons. (Supported by the VA Rehabilitation R & D Sea'vice)

Poster

Presentations

S-53

260

263

THE EVALUATION OF POLYCLACTIC ACID-CO-LYSINE) AS A BIORESORBABLE SCAFFOLD FOR TISSUE ENGINEERING

DESIGN AND VALIDATION OF A LARGE SCALE FLOW CHAMBER FOR EXPOSURE OF CELLS TO CONTROLLED SHEAR STRESS CONDITIONS. Epstein, MAF, Pcl~um, DR, =ridBcoanR,JC, Jr. University of Conno~:ticut School of Medieian, Farmingtoa, CT 06030-61 l0

A, D. Coot, L M, Jokusnn, U. B. Pajvenl, R. Longer Department of Chemical Engineering, Mnssuchusetts Institute of Technology, Room E25-342, 45 Carletoo St., Cambridge, MA 02139 We are developing methods to enginesa" natural tissues and organs by transplanting cells supported on extracelluinr matrix analogs made from biodegradable, rusorbable polymers, such as poly(inctic acid) (PI.,A) and poly(glyr acid) (PGA). While significant pmgruss has been made using this approech with some cell types (e.g. chcedrocytes I~1 PGA meshes to form cartilage), many other cell types have not performed well with approach. Onc of the reasons for this is the lack of specific interactions between the polymers and the ceils. In order to promote these interactions, we have synthesized poly0antlc ecid-co-lysino) (PLAL), which consists of sppmximately 99% lactic acid and I% lysine. Thc side groups of the lysine residues provide functional amine groups whkh can be used to attach biologically active poptidns. This approach provides for the possibility of initiating and controlling specific inleractions between the polymer and cells, thus bringing the extracelhiinr maulx analogs one step closer to fully replicating natural extraceIlular matrin proteins. Our experimental results include visually observed and quantitatively measured increases in adhesion and spreading of bovine aortic endothelial (BAE) cells on PLAL surfaces containing cevalently attached RGD-poptides. In addition, we have investigated various methods of I~OCe,~ing this copolymer and have determined the effects of these processing methodson cell behavior.

As our knowledge of how field shcer muss effects can metabolism inr h is mmetimes advanlageous tu be able to expose a hirge nambes of cells simultsneneuly to the uxae sheer mess coeditines in order to provide mff~ieat cell fr~ticos for sev~d differunt at~yses (.r severed mR-NA's er adbesioa molecnles), To meet this reqlfiremcet, s parallel plato flow ~ b e r was d e s i ~ and ~x~aru~ed to =x~ept a g l w slide (or odier cell anppo~ matrsinl) lariat, in axe= ~ t sta~lfd ~ sllps of microscopeulkics itadto cxposetheoal] m~aulayerto M~tdyIsmin*r s~Car stresses. This desigudiffere from manyinev:cesdealgushi both the sizeof tbe sbkic(85 mm X 135 ram) andthe lec,atlonof theeuUyto the flow e h ~ t andexit fi~m ~e uhaaedby die flaid meditma uh'cola6n8 throeSh the chamber. The flald cotem die flow dumael through a ~ s~ from the reservoir which is an integral pa~ of the top phite of die flow c2uunber and is lecated direotly abeve ~e ~ of die flow ohanouL Shnilarly, the flnid ex/ts tbe flow chaunel thrunsh a vertioal dit in d:e bese of tha flow chamber which ahm hulds the giass dide. Flow visealizatiee stodics were porfoms:d to verify eat the llow wus hmi..~, steedy usd miform thseeghout the flow channel. Fhioreeccet dyesohiticowasinjecledalmultamamdy at fiveloosticesintheenW/dit and the dye traces were photogrsphod with a video camm-a through the tup of the flow chamber. The video images were capturedusing a Macintosh 680 AV compo~ ~ty~em ami amdyazd using tbe publlo domsin NIH Image Prosram. The pa~tines traesd by the dye w~e straighL and die distances between pathlines rewained ooustam thronghcot the k n g ~ of the flow nkanaeL This flow chamber has been used to expuse oenflucet monoisyem of celuned HUVEC's to steedy laminar shear messes between 0.06-6.1 dynes/(sq can) during invusfigstims of the effects of shear stress on expression of adhesion moicceks.

261

264

STRENGTH OF RBC ADHERENCE AND CIRCULATING PLASMA ADHESION FACTORS IN CHILDREN WITH SICKLE CELL DISEASE: A CORRELATION Cook GA', Wang WC=, Dancy RM 3. Slack SM t, Tmino VT' tTha University of Memphis; =St .lode Children's Research Hospital; 3LeBonheur Children's Medical Center, Memphis, "IN

FLUID FLOW ACTIVATES G-PROTEINS IN HUMAN ENDOTHELIAL CELLS Siva R. P. Gudi and John A. Frangos University of California, San Diego, La Jolla" CA. 92093-0412

The present studies were designed to explore the ability of pis.~na from pediatric sidde cell disease patients to promote RBC attachment to vascular endotbelium. The RBC-plasma ~ o ~ of sickhi plasma + sickle RBC flora crtisis subjects (SSC), sidde plasma + sickle RBC from baseline subjects (SSB), sickle ~ plasma + normal RBC (SNC), and sickle baseline plasma + nmmal RBC (SNB) were investigated in these studies. The sttaclmumt of RBC to cultured human u m b ~ vein endo~.llumwes assayed under wall sheer stre~ coaditio~ of I to6 dyne~em 2. Ace:alum dean~--l~ with incressing shesr, the deca~essewas at~proximsted by a fu'st ordes decay cqantiun and die manimum number of adherent RBC at zero stress (C,.,) was determined. The strength of adaesina, ~ , was defined as the wsil sbear stress st which adberenoe equsis C,../2. The 'usefor SNC = 1.8 and SNB 2. I wes sigaificandylower lhimSSC = 2.8 and SSB = 3.0. In ouder te detarmine which plasma factoes might be involved in the adhesion process, the pI~ma cuncentratic~s of VCAM-I. ICAM-I and t h ~ were det,amined by standard ELISA methods. Tbe basetine(n=l 2) and oris/~n= 12) corteentratinns (mean ~: SEo ng/ml) foe VCAM-I were 1492.3 :t: 323.7 and 1755.3 :It:98.9, ICAM-I wes~454,1 i 108.7 ned $52.1 :t: 54.0 and thrombespoadin ware 128.5 :k 37.8 and 137.5 ~ 63.5. The coneentrstions of VCAM-I, ICAM-1 and thmmbospendin in anfmal plesma(n=6); 769.3 =1:82.43, 313,3 • ~: 16.9,respeelively, were signifieantly lower thaninsickhiplL,=ne. A P ~ CorrelationAnalysisbetween plasma conesntradon and the level of adbealun at all wall shear indicated a positive correlationfor I h r o m b o ~ and a negative correlation for VCAM- l and ICAMt dt.ringcrisis. ']'hus~circu~stlngt h ~ m t x ~ l m may contdbt~e to this adhesinn; e~ulating VeAM-t and ICAM-I may be involved in in.'biting e~tesion by binding to existing receptors on the cell ~rface.

262 ANALYSIS OF MOLECULAR ENGINEERING INTRACELLULAR ADHESION RECEPTOR/CYTOSKELETAL INTERACTIONS ON CELL MOTILITY Paul A. DiMilla* and Stephanle S, Stine Department of Chemical Engineering and *Center for Light Microscope Imaging and Biotechunlogy, Carnegie Mellon University, Pittsburgh, Pennsylvania 15213 The molecular-level coordination of receptor-mediated adhesion with intracellular forces generated by the cytonkeleten creates traction forces which control coil motility. Although previous studies have ustablished the fundamental role of interactions between cell-sortie receptors and substretum*bcend extracellular-matrix ligands in movement, the relative role of intraeellalar interactions between these receptors and the eytoskeleton in the spatiallyregulated transmission of trlJetiou foreus remains unclear. To address this problem we have incorporated the dynamics of reversible binding between adhesion receptors and cytnskdetul components into our previous mathematical model for the dependence of cell speed on the properties of the adhesion roesptor/extracellular ligund interaction (Biophys. L 60: 15, 1991). We predictthat regulation of receptur/cytoskeletal binding (through changes in the cuncantrations and affinities of cytoskelesal linkers) can affect movement over ligand-coated substrata, but that the relative roles of these effects depends on the properties of receptor/extracellular ligand binding. We also demonstrate that the presence of spatial asymmeUy in the strength of the receptor/cytoskelelal interaction -- in the absenceof spatirJ asymmetry in the receptor/substratum interaction -- can generate observable rates of movement. Oar results vcrlfy that modifications in in tracell ularly-based adhesive events can alter motility and provide a basis for interpreting experimental studies in which the numbers and pmpertius of adhesion receptors have been modified through molecular engineering.

Indirect evidence suggests that flow regulation of vascular tone via the endothelium involves mechanochemical signal transductinn via G-proteins. To further investigate this, we incubated plasma membrane vesicles prepared from human umbilical vein endothelial ceils (HUVECs) with a ot32p labeled photot~eactive analogue of OTP, (ot32p) GTP azidoanilide, The membrane vesicles were sheared at 75 dynus/cm 2 (370C) for I rain in a cone-and-plate viscometer. Auroradiographic analysis showed a 42 kDa protein was labeled by GTP in sheared plasma membranes, Incubation with bradykinin (10 p.M) also resulted in similar GTP binding. Adherent monolayers of HUVECs preincubatod with (a32p) GTP azidcenilide were subjected to fluid flow in parallel plate flow chambers, The onset of fluid flow (10 dyues/cm 2 ) for 10 sec elicited a 3 fold increase in GTP binding. In addition to the 42 kDa protein, a 31 kDa proteiu was labeled in whole cells. Incubation with perurssis toxin inhibited the flow induced GTP binding by 85%. Immuneprecipitstiun of GTP bound proteins with anti G u antisera showed that Gqa/ll a and Gi3a/oa were rapidly activated by flow, These resnlts demonstrate the involvement of G-proteins as mediators in mechanochemical signal tranaduetinn,

265

STRUCTURE-FUNCTION RELATIONSHIPS IN T I S S U E E N G I N E E R E D C A R T I L A G E

L.E. Freed and G. Vunjak-Novakovlr Massachusetts Institute of Technology, Cambridge, MA 02139 Tissue engineered cartilage can be grown in vitro using isolated cells and biodegradable polymem and implanted in vivo to repair damaged joints. The goals of the current study were to assess the biomechanical properties of cartilaginous constructs grown under venous in vitro culture conditions and to qualitatively relate tissue structure with function. Constructs grown for tive weeks in rotating bloraactora, spinner flasks and petd dishes and natural cartilage explants were assessed with respect to histological appearance (structural elements and their distributions), biochemical composition (cells, collagen and suffated glycosamlnoglycan, GAG), and biomechanical properties (compression and recovery behaviors). As compared to natural cartilage, tissue engineered constructs from all groups was more compliant due to several compositional differences including higher water contents and lower collagen and GAG contents. As compared to samples from rotating bioreactors, flask samples responded to a step change in normal stress with lower maximum strains, but recovered more slowly and to a lower extent. The ra/at~veb/high stiffness and visoostty of flask constructs was correlated with a structural feature: a capsule consisting of multiple layers of fiat cells and collagen fibitls. Further efforts to quantitate structure-function relationships are needed to design bioraactors for in vitro cartilage cultivation and to develop cdteda forin vivo implantation of tissue engineered constructs.

S-54

Poster Presentations

266

269

CHARACTERIZATION O F AN O R I E N T E D ARTERIAL MEDIA-EQUIVALENT Timothy S. Girton, Daniel L. Monradian*, and Robert T. Tranquillo D e p a r t m e n t s of Chemical Engineering and M a t e r i a l s Science, and *Laboratory Medicine and Pathology, and The Biomedical E n g i n e e r i n g Center, University of Minnesota

CALCULATION OF THE TRANSFER RATE OF LIPOSOMES TO LUNG ENDOTHELIAL CELLS iN A MODEL OF THE MICROVASCULATURE. F.R. Haselton, T.D. Ginrgio and R. Pascual. Center for Intracellulsr Engineering, Departments of Biomedical and Chemical Engineering, Vanderbilt University, Nashville, TN 37235.

Our working hypothesis is t h a t the circumferential orientation observed in the n a t u r a l a r t e r i a l media is crucial to the mechanical s t r e n g t h of n a t u r a l arteries. 9Binartifir media which mimic this orientation will, therefore, confer improved mechanical properties t h a t are prerequisite in a completely bioartificial a r t e r y (i,e. w i t h o u t any synthetic polymer components), C i r c u m f e r e n t i a l orientation of collagen fibrils in a media-equivalent (ME) is achieved in a simple and effective w a y u s i n g t h e o r i e n t i n g effect of a s t r o n g m a g n e t i c field d u r i n g collagen fibrillogeneais when the ME is first created f'l'ranquillo et el, 1995; patent pending) and/or by allowing the nubsequent SMC-induend contraction to occur on t h e rod t h a t forms the lumen (L'Heureux et al, 1993). Circumferential orientation of the entrapped smooth muecle cells (SMC) in achieved subsequently via cell enntaet guidanco, the induced SMC orientation along oriented fibrils. After describing our methods, we provide several lines of evidence t h a t our ME are cireumferentially oriented, including collagen birefringenca and circumferential SMC orientation, and increased M E ntiffnons with reduced eronp in the cireumferontial diroetion, an compared to control MEs. The relative contributions of the SMC and the collagen to the ME mechanical propertien are assessed. We discuss the optimization of our methods in order to better mimic the circumferential orientation and mechanical propertien of a n a t u r a l media. Other applications or magnetically-oriented tienueequivalent tubes, an well as other structures, are indicated.

We have developed methods to estimate the transfer rate of liposemes to endothelial coils. Cell-column elutlen profiles of labeled Iiposemes ware compared with s coInjectnd flow tracer and found to have a constant non-negative extraction ratio suggesting that the transfer of tiposemse to the endothelial cell is Independent of Ilpcoome concentration over the range investigated. Therefore, we estimated ke, the coil-column rate of Ilposeme-endothellsl transfer, from a simple mass balance comparison of the total elution profile given by k, = - i.}--~=,,ln(l-~'~E(t)) where F is the flow rate through the column, I column length, r column radius, s void fraction and ~E(t) total normalized differenco In the elutlon profile cooncontratinns. A similar analysis was used to best fd kr, the local rate of liposeme rapture, based on the retention patterns of liposemes ramalnlng in the coll-cdiumn from the expression c( x ) / c, ffi c~p(-k, ~ - x ) . ( l - exp(-k, ~Ax))

where c(x) Is the retained Iiposemes st a distance x from the entrance to the column, c I is total lipcoome inJestnd, u average velocity, S surface area concentration and a,x column slice thickness. Similar values ware found for ke and kr. However ke values have the greatest utility since each estimate of kr requires a separate cell-column. This quantifiable /n v/b'o approach should prove valuable in optimizing Ilposeme delivery to endothelial cells in vivo.

267

270

PORCINE PI.,&TELET ACTIVATION AND AGGREGATION UNDER THE INFLUENCIt OF TIME VARYING SHEAR FLOW FIELD

MEASUREMENT AND ANALYSIS OF FILOPODIAL DYNAMICS IN NEURONAL G R O W T H Stucy S, Hawkins and Helen M, Buettner, Deparlment of Chemical and Blochem[~tl Engineering, Rutgers, the State University of New Jersey, Plscatuway, NJ 08855.

C.C. GLISMAN~. A.F. HORTONu, S K WANG, NH.C. H W A N o Csrdiov~mlar Fluids Laburstm% Deportment of Biomedicou Engineering, University of Mismi, Coral Gables, Flodds, 33146, USA, uCoulta" Corp.. Mimni, Florida, 33196-2500 USA The edivation and eggrogafion of platelets studied by several investigatorshave shown thatplatelet nggregation (PA) will cr only above a omaln threshold shear stress (60 dyn/cm2) These ~=ligaticos wae ~ hosed on mcssm'ements made en platelets collected fi'~n steady shear fields. However, in cerdiovescouK sysuan, s coontant shusr stress flow field does actually ant exist Using a retw/flow field betwem two cytieden, wizen tbe l~p um bu vmed by shi/ting en onter mtioma3, t~yfiedes~ to so i-mr n:tst~,,gcyl~deL it is pean'ble to g=nernte a time varying shear flow field. Porcine whole blood was diluted l: 4 and subjected this/low field for s t ~ a d of 600 seconds. A88regat/on was ~ by flow c ~ e enalysls (FeM) using a CD61-FITC n~ncelenal antibody. Changes in inte=cellultr fleecoucium were probed Ls m ~ t s of plateletactivation ruing F m Red. Fm'a End intemity, measerement tsken be~'ore end alter expom~ to sbuar slmwad increases in intnrenllouar calcium. BrA23187 (calcium Isno 4) added during FCM revuslad the fraction office calcium available before and after shear. Digitonin collapsed all caloiuro gradients end revealed the calcium equih'brium stetc. The rstico of the free ~ l h i l m - calcium due to shear stre~ and to free calcium caused by BrA23187 gave a relative measure of platelets calelum mobilization potsntial. The flow field induced during the experiment was precisely mapped using Les~ Doppler ~ (IDA) and a clearanalogflui& F M C enalysis r~,'calad that even at the low shear stress of (5 to I 0 dyn/um'= menn end 20 to 30 dyn/cm'~ maximum) a 5 % increase of aggrngetien oveT a period of 600 seconds takes place. The decrease in activation potential of more than 50% througbuut the mine time period, indicatsd that the platelets wese activated.

Quantitative analysis of neudte outgrowth provides key information for the development of new therapeutic strategies for treating nerve injuries. Filopodial dynamics are an important aspect of neuronal growth and guidance. However, modeling and tracking filopodial behavior is very difficult because fihipodia move end deform in apparently random patterns, end filopedia are difficult to image when they move out of the plane of focus or occlude one another. We have developed a set of computer algorithms, Surface Extraction end Labeling for Neufite Tracking (SEaLNeT), which can be described in three separate phases: 1) obtaining growth cone outlines, 2) automatic labeling of growth cone features, end 3) automatic analysis of growth cone features and motion computation. These computer-assisted algorithms allow us to obtain quantitative results in a reasonable amount of time. We have used SEal.NeT to track filepedial rates of extension and retraction, ffiopodlal interaetlons with substrate boundaries, end angular displecements of fihipedia relative to the ~rajectories of the growing neudtes. These measurements have been made for growth cones in chick DRG neurons growing in vitro on laminin. Comparisons of filopedial dynamics in growth cones in homogeneous and inhomogeneous environments will be presented.

268

271

THE SICKLE SHAPE OF THE ERYTHROCYTES D. Grabec, V.lOalj-lgli~, B.2eki Institute of Biophysics, Medical Faculty, University of Ljobljana, Slovenia

STRESS-GROWTH RELATIONSHIP OF TUMOR CELLS Helmlingex G., Netti P.A., Lichtenbeld H.C., Molder It./., and Jain R.K. DpL of Radiation Oncology, Mass. General Hospital, Hasv~ed Medical School, Boston, MA 02114, USA

The aim of this work is to detersune whether, for specified parameters, the sickle shape of erythrecytes is an energetically favourable state of the membrane. The long-range goal is to detenurae more precisely the importance of the shape of the sickin-cell. While the general problem of detefnuniag the cell shape for a given set of cell parameters is not yet solved, solutions have been obtained for axially symmetric shapes and for other shapes

The aim of this study was to cvalnate the effect of solid stress on tumor growth. We thus quantified the in vbro growth kinetics of three tumor cell lines (human colon adennea~inoma, LS 174"1~.murinr Iraanm&-'ycarcinoma, MCOUV:.~ rat r h a b d o m ~ BA-HAN-I. ekinus A.B.C) grown as spheroids in ageroso gels at different gel concentrations (0.3-1.0%) and in free suspension. Agarese gels will support tumor cell growth as anchorage-indepeedont multicolhilas spheroids; they also ave inert, stable, ram.toxic, and do One Ira:sonS s dgnifu:ant diffusion bonier to nutrients. CoUture in agascoe gels dramatically inhibited the growth of spheroids. The growth rates and final sizes of LSI74T splierolds wese similar among the 0.3, 0.5, 0.7, 0.8, and 0.9% gel concenwatims, but were significantly lower when competed to f~e so.~.asion conWols. The must dramatic result wus a marked inhibition of growth and s redused maximum size of spheroids at a 1.0% gel concentrstion; however, the colony growth effielencius were remarkably similar (88 to 95%) for all coUture conditions tested. Avenge spheroid sizus did not denrea~ in a progressive Onmm~ with increusing gel coocemrations t~p to 0.9%. Growth inhibition was observed between the 0.9 and 1.0% gel concentration. Other tumor cell line~ tested.responded in a q~itatively similer manner. The growth rates and final sizes of MCaIV spheroids wese not significantly dlffe=ent among the 0.3, 0.5 and 0.7% gels, and ~ growth inhibition was again observed a t e 1.0% gel concsatration. For the least differcotinted clone (A) of the r h a b d o m ~ cell line (BA-HAN- 1), growth inhibiti~t w ~ again observed at s 1.0% gci concontratico, as compazed to 0.7.0.9~ gels ~ free sospcosim& However, the growth kinetics of the more differentiated clones B and C were similar among all gel concentrations tested (0.7, 0.8, 0.9 and 1.0%), but were lower when compared m free suspensions. Out" results suggest that solid s~ess exerted by an inert gel ~ on tumor" cells is able to restrict spheroid growth. We are ~ n d y evaluating the stress field in the vicinity of a spberoid at different gel concentrations. We pmpuse that the mechanical slxongth of the gel will provide us with an indication of the force exerted by growing aaner cells in a spbe~id. This information will be useful in the estimation of sa'ces fields in growing tonu~ in vivo.

possessinghigh~ .

In this paper we find the solutionsfor several modeled non-a:dally symmetxic shapes, including sickle shapes, using a bilayer couple model of e~Tthrecyte shapes in which cell shape is related to membrane bending energy. In the biisyer couple model of the cell, the cell parameters include its volume, its average membrane area as well as the difference between the areas of the inner and outer membrsne lipid laye~s. The cell, in the shape of 'crescent', is descn'bed as a section of a totes with ellipsoidal end caps. While the number of pofm~Olecell shapes is sovefely restricted by assuming such a shat~ file aaalytical solution of the problem is l~m~ole. Further we show how the analysis can be generalized to obtain ~cal appm~hes for describingthe bending of cells pussoss~g arbitrary shapes and again apply this a~lysis to the cylinder with elllpsoidou end-caps. A comparison of the results with tho~ obtuiucd from the cre~ont model is used to pennis a partial check of onr results. We show that within our models tbe bent shape of e ~ is not the most favoured one. As some of the shape~ included in the models appear to dusely appm:dmate the rigorous solution of the biinyes couple model, we therefore conclude that the bent shape of etythronytes is not solely a consequence of the cell membrane. Since in the sickle cells hemoglobin is polymerin~ we fustherlucdietthat the polymerization of the hemoglobin has the mechanical impact On the cellmembrane. The p ~ of the polymerization is then modeled as the stick gxowing th the coil, which may bund at tbe certain conditions.

Poster Presentations

272

274

Controlled-Delivery of G r o w t h Factors: The Heparin-SepharoseCalcium.AIginate System for Release of basic F i h r o b l a s t Growth Factor S.K. Hobbs" M. Nugcnt', R.K. Jain*, L.G. Cima" "Department of Chemical Engineering, Massachusetts Institute of Technology, Cambridge MA, *Departments of Ophthalmology and Biochemistry, Boston University School of Medicine, Boston MA, "Steele Laboratory, Massachusetts General Hospital, Boston MA

ANGIOGENESIS-ASSOCIATED EXPRESSION OF ADHESION MOLECULES ON ENDOTHELIAL CELLS TREATED WITH bFGF, VEGF, TNFe% AND TGF[!

The Iocai controlled delivery of growth factors is desired to initiate and maintain specific tissue growth responses in a limited area. Systemic delivery of these factors is not feasible primarily due to significant loss of localized biological activity. One example of such a growth factor is basic fibroblast growth factor (bFGF); bFGF is a potent mitogen for a variety of ceils in vitro and an angiogenic factor in rive. Previous in rive studies with bFGF controlled-delivery using conventional materials (i.e. Gelfoam, Elvax, PTFE, and collagen) have established the feasibility of local delivery; however quantitative, reproducible efficacy of such release has not been established. To this end, a thorough investigation of the release mechanisms from these devices mast be undertaken. We have concentrated on the heparin-sepharosealginate system developed by E. Edelman and M. Nugent. This device is composed of heparin-sepharose beads immobilized within a Calcium-Alginate gel. This device exploits the high affinity of bFGF for heparin. We have experimentally investigated and mathematically modeled the heparin-bFGF interaction and the alginate-bFGF interaction.

Koenig, G.C., Mann, L.L, Meldes, RJ., and Jain, R.IC Department of Radiation Oecology, Harvard Medical School and Ma ~-~chusetm General Hospital, Boston, MA 02114 T lymphonytes have been feend to localize and become functionally active in regions of wound repair. These regions of neovaseuinr response also possess increased levels of rations angiogenlc fucto~ Therefore, we hypothesized that the angiogenir factors produce an upregolation o f oell adhesion molecules on the e~tdothelium conducive to T lymphocyte binding. Using the method of Targeted Sampling Fluorometry (TSF), HUVF~ monolaye~ treated with the angiogcaic factors bFGF, VEGF, TNFa. and TGF~ were analyzed for thek expressim of ICAM-I, VCAM-I, E-Selecfin and P~electi~ The results showed increased levels of ICAM-I expression by VEGF and T G I ~ treated monolayers. TNFa upretp,l~t _,y] all of the corresponding adhesion molecules, whereas no upregolatinn of any of the adhesion molecules by bFGF treatment was apparent at 24 h or 48 h time periods. In ~ with lx~vions studies involvinglymphocyte arH~sina to TNFa treatment, increased extxession of ICAM-I may correlate with T lymphocyte binding at low shear stress. This study implicates a plausible mechanism by which T lymphocytes may localize in the regions of neovascuinrization, and reinforce the angiogenic process through the local production of cytokines.

272a

275

EFFECTS OF IONIZING RADIATION ON THE ADHESIVE INTERACTION OF HUMAN TUMOR AND ENDOTHELIAL CELLS IN VITRO Mohammad F. Kizni, Bmee M. Fenton, Lee Ann Spurn, and Dietmar W. Siemann Departments of Radiation Oncology and Medicine, University of Rochester, Rechester, NY 14642

NERVE REGENERATIONTHROUGHCOLLAGENTUBES CONTAININGCOLLAGEN-GAGCOPOLYMER

A contriingation assay was used to determine the effects of ionizing radiation on the force of adhesion between 3,549 lung adcneearcirtoma tumor cells and human umbilical vein endothelial cells. The tumor cells were fluoresccndy labeled and divided into control and irradiated groups. The irradiated groups were exposed to irradiation levels ranging from 5 to 20 Gy using a cesium source. A specified number of these A549 tumor cells were then delivered into each well of 96-well cell culture plates voptaining confluent monolayers of human umbilical cord vein endothelial cells, and were given time to adhere to the endothelial cells. The wells were then sealed and were ezpose~ to an acceleration field varying from I to 50 g for 10 minutes. Finally, the wells were drained, and the number of minor cells adhering to the endothelial monolayer were counted using a fluorescent microscope system. Our results indicate that the irradiation of A549 tumor cells significantly increased their adhesive interaction with endothelial cells (irradiated/controlffi 1, 1.3, 1.9, 2.2 for 0, 5, 10. 20 Gy). hx eont~nst, when endothelial cells were irradiated, rather than tumor cells, adhesive interaction decreased with an increase in the radiation dose(irradiated/control~ 1, 0.9, 0.8, 0.5 for 0, 5, 10, 20 Gy). These findings may have important implications in terms of metastatic potential following the irradiation of either tumor or endothelial cells.

Landstrom, A.E.t, Yannns, I.V.t, Ky, B.2, Hsu, H-P), Norreganrd, T.V. 4, SpeCter, M. 3 Deportments of Mechanicalt and Chemical2 Engineering,Massachusetts Instituteof Technology, Depamnent of Oahopedic Surgery3, Brighamand Women's Hospital, Divisionof Plastic Surgery4, New England Deaconess Hospital Injuries to peripheral nerves innervatinga limb cause paralysis, and can necessitate amputation. The inabilityof the nerves to regenerate spontaneously and the limitationsof autograh procedures led to treatmenta involvingintabatienof the nerve ends. Work from our labomtmT Showedthat enshenthingthe nerve ends in a siliconetube containinga specificporous, resorhable collagen.GAGcopolymer, servingas an analogof cxtracelluisrmalfix,improved zegenetahon. In this study the siliconetube was replecnd withporous and non-porcos collagen tobes m order to produce fully degradabledevices. Collagen-GAGfdlnd collagentubes and centrals (collagen-GAGfdledsitic~e tubes and empty collagenand siliconetubes) were implantedin a 10 mm gap in the rat sciallc nsrve, and tl~ regeneration evaluatedafter six weeks usinghistology and immunollistecbemistry,employing antibodiesto neo.rofilamenta. Lightand fluorescentmica~scopy imagesof cross scetinosof the neawc were digitizedand analyze~ Histograms of the diametersof the axons weregeneratedand compaxed using three nervesper prostheticgroup. The coilagen-GAG rdled porous and non-porans collagen tubes were comparable with the collagen-GAG filledsiliconetubes in promoting axonal regrowth. Confirming earlierwork from o~ laboestory, regeneration through the sificoneand collagen tubes was siguilicendy enhanced hy the coilagen-GAGcopolymer, enmpared to empty tubes. Sinthesare underway to compare long term functionaland neurologicalresults of regrowth for the differentgroups.

273

276

BONE REGENERATIONDEVICESBY FREE FORM FABRICATION W. S. Koegier, L, G. Cima, R. A. Giorde~o, M. J. Cima Department of Chemical Engineering,MIT. Cambridge, MA 02139

THE EFFECTOF A POROUSCOLLAGEN-GAGMATRIXON TENDON HEALING Louic, L.K., Hsu, H,-P,l, Yannns, I.V., SpectoL M.1,2 Mass~haselts Instituteuf Techoology, Program in Polymer Science and Technology tBfighamand Women'sHoapilai,Dept. of Orthopedic Surgery, HarvardMedical School 2Brockwn/West Roxhoi'yVA MeScal Center

A substantialclinical need exists for improveddevices for treating segmental defects in bone. Ideally. these materials would provide the mechanical strength assueiated with bone function and support the growth of bone cells until the device is replaced by the patient's own tissue. Most work in this area has been devotedto finding materials with chemieai properties that will allow or even stimulate bone formation. We propose that architecture may have just as significantan effect as composition in a bone regeneration device. For instance, the healing rate may be strongly related to how fast cells can migrate into the device and that rate will be related to the si7~. shape, end orientation of pores end channels. We are developing aa approach that allows delioentinnof how each of these important variablesaffects bone wound healing. We are using a solid free-form fabrication technology, Three Dimensional Printing (3DP), to gain control over both structure and compositionat three levels: (1} macroscopic shape and compositionof device; (2) intermediate features such as size, orientation and surface chemistry of pores and channels; and (3) microscopic features such as porosity and texnn~. This talk will focus on optimizingthe surface texture and compositionof devices built by the 3DP process with respect to adhesion and migration of celis involvedin bone wonnd-hanling.

S-55

Followinginjuries resulting in a gap, tendons generally do not heal sufficientlydue to either the lack of reparativetissueor the funaional inadequacy of the resulting scar tissue. The objectiveof this study was to evaluatehealing in a uew rabbit model in which porons collagenGAG(CG) copolymeranalogsof extracollularmatrix (ECM) were implanted in a 10 mm defect in the Achilles tendon. Analogs of ECM with chemistry similar to those used in this study have been shown te promote regeuensticoof denmis,peripheralnerve, and meniscus. Ensbeathed in siliconerobber tubes, CG copolymers with averagepore sizes 10, 60 and 120 pm were implanted into the rabbit model. Empty silicone rulfo~ robes were used as controls. The implants weze evaluated histologically at three, six ~ twelve weeks postoperdtatively. The volume of reparative tissue was ganecally greater in the tubes filled with 0 3 copolymef than in the empty control tubes. Aftersix weeks, the disinter of the cylinc~ of reparative tissue in defects filled with copolymeraplxoximatndthe inner diameta of the silicone tube (3.5 ram). Tissue rdling the control tubes had diameters from approximateJy0.1 to I Dam. At three weeks, CG copolymerwas still prer,ent in the majorityof the implants. Less CG was foand by six weeks and by twelve weeks all of the copolymer had resorbed. A fibrin dot va~ fonnd within the CG copolymer and within the empty control tubes throe weeks aria" implantation. At six weeks, tubes filled with the copolymer displayed the presence of fibreblnsts and newly syntigsised collagen disa'ibuted throughout the copolymea'. Spindle shaped fibroblastand coUagenousmatrix could also be fotmdaligned along the long ax~ of the tendon in the control tubes. Theae early findingsdemonstrate the value of the new animal model for investigatingtendon healingand the potential of a CG copolymer for favorablyal~t'ing the healingprocess by filling the defect site with fibroblastsand new collagen.

S-56

Poster Presentations

277

280

DEGRADATION OF POLY(GLYCOLIC ACID) NONWOVEN SCAFFOLDS AND THE ENGINEERING OF MECHANICALLY FUNCTIONAL TISSUES Peter X, Ma and Robert S. Langer MIT, Department of Cbemical Engineering, Cambridge, MA

SEEDING OF COLLAGEN-GAG IMPLANTS WITH ARTICULAR CHONDROCYTES Ramappa, AJ.I. Breinan, H.A. 2, Yannas. LV. s. Olssn, B.R.4, Sledge. C.B.t Spoetns, MJ ~Departmcet of Orthopedic Surgery, Brigham and Women's Hospital; 2Division of Health Sciences and Technology and SDepartmem of Mechanical Engineering, Massachusetts Institute of Technology; '*Department of Cell Biology, Harvard Medical School.

Biodegradable polymeric scaffolds have been used to engineer new tissues or organs intended for transplantation and reconstructive surgeries. The degradation behavior, microstructural changes and mechanical properties of a model poly(glycolic acid) (PGA) scaffold (fibrous nonwoven disks of a diameter of 10 mm and a thickness of 2 nun) were studied in vitro under mixing conditions at 37 oC in tissue culture medium. The mess increased slightly during the first 3 days, presumably due to hydration and adsorption, but it decreased exponentially thereat~r. After 9 weeks, only 12.5% of the mass remained. The melting point of the scaffolds decreased slowly (-0.40 OC/day) during the first 11 days and decreased much faster (2.89 ~ thereafter. The crystallinity doubled during the first 11 days, but it decreased thereafter. The degree of polymer chain orientation increased slightly in the first week, but it decreased sha~ly between week 2 and 3 to nearly random orientation. The compressive modulus decreased ca. 35% and the aggregate modulus decreased less than 10% in the first l l days. The apparent permeability of the scaffolds increased about 5 times during this period. The thickness of the PGA scaffolds incrfased slightly during the tint 3 days, but the scaffolds collapsed into disintegrated short fibers between week 2 and week 3. The PGA scaffolds were used to engineer new cartilage by culturing bovine choedroeytes on them. The compressive modulus of [he PGAchomirecyte constructs increased with time, and re~hed the same order of magnitude as that of normal bovine cartilage at week 9. The aggregate modulus of the PGA-chondrocyte consmicts decreased by 57% over the float two weeks, but increased thereafxer, reaching the same order of magnitude as normal bovine cartilage at week 12. The apparent permeability of the PGAchoudrocyte constructs decraased by 4 orders of magnitude between week 1-3 and thereafter remained the same order of magnitude as that measured for normal cartilage. These results demonstrated that the PGA scaffolds did serve as templates and that mechanically functional tissues could be engineered with the scaffolds,

Articular cartilage has a limited capacity for spontaneous regeneration. We have seeded a porous, rnsorbable eollagan-glycosaminoglycan (CG) matrix with articulm- chondrocytes in an effort to develop an implant to facilitate regeneration of defects in the articular surface. Specifically. choudreeytes isolated from canine articular cartilage were passnged and then cultured on a CG scaffold. Histological studies rcve.a~edinfiltration of the cells into the pores and immediate attachment of the chondrocytes to the wails of the matrices. Cell morphology gradually changed from a spherical to a fiat, fibmblastin shape. Agarose and collagen gels are known to promote chondreganeals by allowing chandsoeytes to assume their natural shape. We separately incorporated passagod canine articular ehondrocytes and cells from a mouse chondroeyte line (MC615) into collagen gels that were then impregnated into the CG scaffolds, The cells were also grown in monolaynr, ageross, and collagen gel cultures. Histological and immunohistochemlcal methods were used to evaluate the chondrocyte infiltration of the matrices cell morphology, matrix.production, and cell phenotype. Cells grown in agarose culture maintained a spherical shape and produced cartilage matrix molecules. Chundroeytes cultured in collagen gels assumed a spherical morphology but later became more fibroblastic. These gels underwent a change in shape after cell seeding, Cellseeded collagen gels could be fully infiltrated into the CG matrices with no appreciable change in matrix architecture. Many of the cells assumed and maintained a spherical morphology, The CG matrices containing ceil-see.rod gels retained their original shape and were easily handled in contrast to the collagen gels that were friable. The CG matrices seeded with chondrocytes alone are promising candidates for implantation into chondsal defects immediately after seeding while .the CG matrices seeded with chondiocytes held within a collagen gel are amenable to growth in cell culture prior to implantation, (Supported in part by the VA Rehabilitation R&D Service.)

278

281

ERYTHROCYTE MECHANICAL INTERACTIONS ENHANCE LEUKOCYIEENDOTHELIAL INTERACTIONS Lance L, Munn, Robert J. Melder and Rakesh K. lain Steele Laboratory of Tumor Biology, Department of Radiation Oneology, Massachusetts General Hospital and Harvard Medical School, Boston, MA 02114

DYNAMIC ANALYSIS AND MECHANICAL MEASUREMENT OF INTRA-BILAYER VISCOSITY AND TRANSBILAYER MOVEMENT OF PHOSPHOLIPID MOLECULES R, M. Raphael and g. E. Waugh Dept. Biophysics, U. of Rochester, School of Medicine and Dentistry, Rochester, NY 14642.

Many factors influence the marglnation, rolling, and adhesion of circulating cells at the vascular wall. Hemodynamics, wall shear stress, receptor-ligand binding kinetics and the resulting intercellular adhesive forces are but a few of the contributors that have been implicated in the localization of passing cells within the vasculature. In the present study, we focus on the mechanical forces imparted by erythrocytee and the Influence that these forces have on the interactions between leukocytes and the endotheliuns. Using a verticallyoriented parallel plate flow chamber to eliminate cell sedimentation toward the monolayer, we examine the adhesion of normal human isolated leukocytes with TNF-ct activated HUVEC monolayers in various flow media. Erytl~ocyte suspensions et various hemalocrtts in saline soluBon constituted the flow medium, and leukocytes suspended in these solutions were perfused over the activated roonolayers at various physiological shear rates. Our results show that erythrocytes greatly enhance the fraction of rolling cells and the binding efBciencies to the monolayer compared to experiments where the suspending medium is saline solution. A mathematical model was developed that describes the evolution of the numbers of bound, rolling and non-interacting cells with time.

The application of an axial force on the order of 15 to 60 pN to a micropipette-aspirated lipid vesicle results in the formation era thin cylindrical strand of membrane, referred to as a tether. The mechanical equilibrium of this system has been used previously to determine the local and nonlocal bending stiffness of the membrane. A dynamic analysis of this system predicts that when a perturbation is made in the force balance, growth of the tether will be limited by interracial viscosity between the leaflets of the bilayer as they move into the tether region at different velocities (Evans and Yeang, Chem.Phys.Lipids 73 (1994) 39-56). When applied to our experimental system, this model predicts that aRer a step change in external forces, the tether will grow exponentially to a new equilibrium length. Consistent with this model, we me~ure an initial exponential phase of tether growth after a step change in membrane tension. However, we fred that the system does not reach equilibrium, but that the tether continues to grow at a constant rate proportional to the magnitude of the change in the boundary tension. This behavior is explained by hypothesizing the movement of phospbolipids from the inner to the outer leaflet to relax differential stress in the membrane. When a term including this dissipative effect is included in the dynamic model, the system is predicted to a reach a steady state in which the tether grows at a constant velocity. From the measured time course oftather growth, we calculate a coefficient of interlayer permeation of .009 s "t, a coefficient of intrabilayer viscosity of 3.107 dyn-s/chl 3 and a value of the nonlocal bending rigidity of 3.10"12 ergs. These measurements imply that in mechanically deformed membranes two mechanisms am available for relaxing differential density gradients: a rapid slip between the leaflets which tends to equalize the area per molecule over the surface and a slower permeation of moleeulee from one leaflet to the other. (Supported by NIH grant no. HL-31524.)

279

282

THE MODULATION OF HEPATOCYTE AGGREGATION BY E X T R A C E L L ~ MATRIX SUBSTRATA Mark J. Powers and Linda Gdffith Cima Department of Chemical Engineering, Massachusetts Institute of Technology, Cambridge, MA 02139 Aggregation of hepotoeytes in culture is an important phenomenon to control in tissue engineering applications. Typically, aggregation will enhance maintenance of differentiated functions but inhibit cell growth, The ability to control ceil aggregation through [he rational design of substrata is a desirable goal in the field of tissue engineering, but at present there exists insufficient information regarding the cellular mechanisms that underly this process to achieve this goal.. In an attempt to determine these mechanisms, we have investigated the role of a variety of phenomena, including cell migration, in the process of hepatocyte aggregation on Matrigel, laminin, and fibroneetin coated surfaces. We have observed that the extent of aggregation depends on both the concentration and type of extranelhilar matrix (ECM) molecules present on the substratum. In addition, it appears that this aggregation is independent of classical single-cell locomotion. Instead, aggregation between cells apparently depends on intercellular contact attained through a different form of cell motility: active cell membrane extensions (i.e., lamellipodia, filopodia) followed by adhesive cell/cell interactions. We also find that the proximity of cells to one another, determined by the initial eell-surfane attachment density and cell spreading areas, is a key parameter in the aggregation process. An implication of these findings is that aggregation may be largely governed by relative strengths of cell/cell versus cell/substratum interactions, a hypothesis which is currently under investigation. These observations could be helpful for improving design strategies of cell Iransplanmtion devices and cell culture substrata.

PLATELET ADHESION TO P L A T E L E T - P R E P A R E D Peter D. Richardson Brown University, Providence RI 02912

SITES

When b l o o d enters a prosthetic b l o o d channel many interactions occur b e t w e e n its c o n s t i t u e n t s and the channel surface. The complex interactions have been difficult to quantify with use of few parameters. In flow chambers it has been seen that, after an initial lag, platelets adhere at a steady rate, reproducible for individuals when re-measured after 6 - 12 mo. The adherent platelets u s u a l l y depart after a short residence, and the sites they leave have enhanced adhesiveness to other p a s s i n g platelets for a few minutes. The p l a t e l e t adhesion to p l a t e l e t - p r e p a r e d sites is not proportional to the initial adhesion rate, and drugs can change it. This paper describes a conditional encounter analysis w h i c h has b e e n developed, with timedependent adhesion enhancement at freshly-available sites. From this, p a r a m e t e r s can b e found to fit experimental results so that adhesion-enhancement can be assessed separately from the initial a d h e s i o n rate. This allows in v i t r o tests to resolve q u a n t i t a t i v e l y the effects of d r u g s on s e c o n d a r y adhesion.

Poster

Presentations

S-57

283

286

NEURAL STIMULAT/ON USING ELECTRICALLY CONDUCTING POLYMERS C. E. Schmidt*, V. R. Shastri', T.-H. Kint +, J. P. Vacanti t, and IL Langer *+ *Dept. of Chemical Engineering, Massachusetts Institute of Technology, Cambridge, /via 02139; fDept, of Surgical Research, Children's Hospital, Boston, MA 02115

BIOPHYSICAL ANALYSIS OF RECEPTOR MEDIATED ADHERENCE BETWEEN SICKLE ERYTHROCYTES AND VASCULAR ENDOTHELIUM IN SICKLE CELL ANEMIA. P.A. Smolinski, J.R. Eckman, and T.M. Wick. Georgia Institute of Technology and Emoty University School of Medicine, Attanta, GA.

Past research has demonstrated an important role for electrical charges in enhancement of neurite extension in vitro and nerve regeneration in vivo. Examples of conditions that stimulate nerve regeneration include piezoelectric materials and electrets, exogenous DC electric fields, pulsed niectromagnetie fields, and direct application of current across the regenerating nerve. The electrically conducting polymer, polypyrrole, has been explored in the past for its use as a substratum f o r multiple cell types, but not for nerve cells. Advantages to using polypyrcole as a tissue culture substratum, besides Its capacity to permit efectron flow, range from the ease of polymer film formation to the flexibilily of modulating the polymer's surface properties by incorporation of different counter anions (dopants). in our current w o r k we h a v e utilized quantitative image analysis to s h o w that polypyrrole (PP) enhances in vitro nanrite extension in the neuronal-like PC-12 cell line and primary chick sympathetic nerves compared to tissue culture polystyrene (TCPS) and polylectic acid (PLA) controls. In addition, PC-12 cells interacted more uniformly with polypyrroin and displayed less cell-cell clumping compared to TCPS. These data suggest enhanced interactions of nerve ceils with polypyrrole. We have aLso explored in vivo tissue response to polypyrrole to find minimal inflanmaatory response and little to no fibrous tissue formation compared to other common FDAapproved matrices such as PLA. Po]ypyrrole is thus a novel candidate material nor stitaulation of regeneration in the peripheral and central nervous system.

The abnormal adherence of sickle red blood ceils (SRBC) to microvascular endothelium is thought to initiate episodic periods of mierovancular occlusion in sickle cell anemia. Several adherence pathways have been biochemically characterized under flow conditions to involve SRBC receptors, endothelial cell (EC) receptors, and adhesive pJasma proteins. In the present studies, biophysical detachment assays were utilized to compare the persistence of SRBC adherence to EC via previously characterized adherence pathways under physiologic shear stress values. SRBC adhered to EC in a parallel plate flow chamber under continuous flow at 1 dyne/cm=. After an initial rinse period, shear stress was incrementally increased from 1 to 10 dyne/cm= and adherence monitored. The tenacity of adherence for the following isolated receptor ligand pathways was then compared: adherence mediated by plasma proteins thrombospondin (TSP) or high molecular weight yon Wilfebrand Factor (vWF), a,I~,-VCAM-1 mediated adherence, and adherence between activated o:,~ on a subpopulation of SRBC and surface bound protein on EC. The tenacity of adherence was assessed by comparing the shear stress required to remove 50% of the initially adherent SRBC. r~. z,, (dyne/cm =) for the adherence pathways studied were as follows: TSP: 1.7_+0.1(n=6); vWF: 1.6d:0.07(n-5); u4i]:VCAM-t: 2.5-+0.1 (n=5); activated a,d3~:2.6_+0.2 (n=6). The most tenacious adherence, as measured by 1:~,was that mediated by the integrin ~[3,, either to VCAM-1 or in its activated form, as compared to the two "bridging" plasma proteins TSP and vWF. These results suggest thai under identical conditions SRBC adherence to vascular EC mediated by the SRBC integrin o.,l~~ is more tenacious than adherence mediated by the ptasma proteins. Thus, ~nder shear stress levels present in post capillary venules in v~vo, ~J3~ mediated SRBC-EC adherence may be more tenacious than adherence mediated by the plasma proteins TSP and vWF under otherwise identical conditions.

284

287

GLYCOSAMINOGLYCANSUBSTITUTIONIN AN ANALOGOF DERMALEXTRACELLULARMATRIX T. J. AlesShafdtz,'tIoannisV. Yanms? "HarvardMedicalScbed, tHarvsngMITDivisionof Heal~ Sciencesand Techndogy Previncs studieshave dx~wnff~at a simple analog of extmea~tdarmatrix dela),s wound conlTact/onend inhib~ scar synthe.~sby inducingpapal regenerationof ff~edermisin animalsand humans. While thesestudies vetoed and eventuallyoptimized ~ physical psramonturs of the matrix material, theydid not provide inforrrBlion on the potential role of the glycosami.noglytan (GAG). . c ~ . t of this coflagan-GAGmatrix. ~ studyfocuseson the effect of s u b s ~ of chondro~l 6-ouffatewi~ otherGAGsand with the correspondingpmteoglycans. pmtecglycan of dermatan sdfa~, .d~dn; has been indepondenlfy identified as an InhibitorofTGF..8. TGF-8,In rum, has beenimplicatedIn ~ syrdhaslsof scarlfssue. Aeco~ngly a studyof a cdlagen-decodn matdxin thiswool healing model should provide fur~er inlormalfon on the machan~unaof regenerationand repair. A cdlagen-aggrecan.ma.trixwas usedas one of the controls:the mawr stncturaldifferencebetweenaggt'scanand chondredm-6-sulisteis the of a proteincore In sggrecan. The detailedprocedurefor sTn~e~sof ECManalogsas graft copofymemof type I cdlagen a~l s GAGh~s beanpublished. Rsctangdar s e c t s (1.5 x 3,0cm) were 9raftedon full-thickness skin wo~s In the dersumel w~lfnHar~eygumanp*~s.~me~idns of ~ grafted wound bed were measuredand phot(~raphudstart~g 24 hoursafterme surgeryand evety 48 hours 6"~emafter,The wound heir-life,defined as the time lor the wound to conbactto or~-half of its original area, is a quantilfea5on of mabixaclfvity. Woundhall-lifeincreaseswith increasingacuity of matdx. Wconds grafted with cdlagen-C6S matdxand coflage~-aggrecan matdx coetracted at a.pproxJmatelytie samerate,yie~ng woundha!f-lirasof 21.2+ 1.4 and 20.5+ 0.6days, respnc~ely. mess haffqivesare con~parableto wound heiNives previouslyobtained by matrixprepamEoesof colla!~m-C,6St~t succeededin reducingscar. Coflagen-decorindatashoweda significantdelayi,1 half4rie,to 23.4+ 13 days. RemaAably,the cofla~n-dermatan sulfatematenal showed a similar delay, 24.0+0ll days, ma'icatin9 thata GAG-spuaficrssclfon in vivo may be the most important deterrr~nantIn rnatnxa~ivity_..Nso, the protein corn of ag~grecandoes not molly significanlfythe observedhealingbehavtor. 111eha]f-Sfeof ungrafledwounDsof ~e samemodeJis 8+1 days. This work was supported by the MIT Slean fund, and by the Huward-MITH,S.T. Division. Aggrecanand deconnweregilts Imm Dr. L C, Rosenberg,MontefioreMedicalCenter,Bronx,NY.

MATHEMATICAL MODEL OF ANISOTROPIC O ~ HAIR CELL WALL A./L Speetor 1, W.E. Brownell2, and A.Pu Popel 1, 1Dept. of Biomedical Engineering, Johns Hopkins University, Baltimore, MD 21205; 2Dept. of Otorhinolaringology and Communicative Science, Baylor College of Medicine, Houston, TX 77030 We propose a mathematlenl model of the composite cochlear outer hair cell (OHC) wall us a three-layer elastic cylindrical shell. The three layers correspond to the distinct elements of the OHC wall: the subsurface cisternae, the plasma meanbeane, and the con:lcni lattice located between them. We model the cisternae and the plasma membrane by two isotroplc shell layers with corresponding elastic constants. We use a model of an orthotrople shell layer for the co~lcni lattice. The coRlcal lattice comista of wound actin filaments and spect~tn crusslinks. One of the a.,0s of onhotropy is tm~ent to the filament direction, the second one is parallel to the local crosslink and the third axis is normal to the filament/cresslink network. The elastic properties of the oortical in~ce in the normal direction ate deter91ined by the system of pillars that attach the lattice to the plasma membrane. The proposed model allows to take into account bending which occtws in nonaxysimmeute (nonhomogeneous) strass-strnin states. There is a considerable bending resistance due to reladvely high tigtdtty of the OHC wall (the effective bending coet~dem for the OHC wall is two orders of magnitude higher than for the red blood cell). We applied the model to the interpretation of the results of the OHC mlcroplpet aspiration experiment. Modeling the pipet effect by a system of prescrthed normal forces and assuming the angle of the acvin winding to be constant, we obtained a full solution in terms of double series. We derive the stiffness parameter of the OHC wall (measured in the micxopipet aspiration experiment) in terms of elastic properties of the composite w a l l its geometry, and the charac~eti~es of the micropipet.

285

288

ADHESION OF LARGE-CELL LYMPHOMA CELLS TO EXTRACELLULAR MATRIX PROTEINS, PEFrlDES, AND ENDOTH~IAL ~ UNDER LAIv~AR FLOW T, W. Smith i, Z. y e n 2 D. G. Meator2, G. L. Nicoison2, and L. V. Mcinthe 1 ICox Laborateq, for Biomedical Engiz~erin~ Rice University 2Uaivaulty of Texas M. D. Andersen Omcer Center

ANIMAL MODELS FOR THE STUDY OF SPINAL CORD REGENERATION Spilker. M.H.. Yann~s. I.V., Hsu. H.-P.*. Norregaard. T.V. tt, SpectoL M. t Me.~sadmsetts Instituteof Technology, Department of Mechanical Engineering *Brigham and Women's Hospital. Deportment of Orthopedic Surgery, Harvard Medical School "New England Deaconess Hospital Department of SurgeD/, Harvard Medical School

Stsblltzaflon of udbeulve inturactioas betwnca tumor cells and codethelial cells or exposed extraceilularmatrix protei~ in the target organ miorovaseainture is an important step in the metastatic imoce~. To ~ d y the relationship bctwcen cell adhesion stabilizationand organspecific metastasis, we investigated the adhesive behavior of metastatic mntine RAW117 largecell lymphoma cells under flow r RAWII7 cells onmist of a poorly metastatic parental line (P), e livor-specific snbline (RI0) and a lung/llvor colonizing subliae 0.,17). Ptevio~ stodfes showed a con'elmion between organ iunfmendni metastasis of RAW117 sublinas and thdr in v~ro adbeulon to organ.deaivedmicrovascotaxeudcthulialceils,and alseshowed that HI0 cells ~ s stnfaco receptor that recognizes the polymet~ lg.ptide, (GRGDS)4 or D4. We have chmacteaSzod tths D6binding n,'cegtor un RAW117 cells as the integha evil3, with s signilltaatly higher level of Gtv~3 receptor l~escat on Ihe HID subtine than for eitbef the P or LIT l i n ~ Flow studtos wm'~ onathg'ted in s perallul-plate finw chamber asing video mlcaosonpy and digitaI image luncosslng. Results were ezpressed as stabilization lime, (he lime requized for the cells to fotm slable edbesines eapehie of withstanding increased shear stresses without detachment. HIO cells stnidlized qnicldy to VN, FN and D4, while P cells sbow~l significantly Ioqor ~abillzafion ilmce, und LI7 cells falkd to stabilize. Stabilization of HI0 cells to VN or D4 was inhibited by anti-~3 MAb, D4, and a monometic RGD paptide. Stabilization of HI0 cells to FN was partinlly inhihited by unti-[33 MAb, anti-FNR polyclonul antibody, D4, and s monomerlc RGD peptlde. A combination of anti-J33 MAb and anti-FNR polyclonal antibody completely blocked stabilization to FN. In contrast, the LIT subline had the shortest stabilization time tO unstimuistcd micxovas~lar endothelial cells of the lung and hapatic siunsoids, followed by HIO cells and P cells.

Many expertmental animal models have been used to study the ~ e n t of spinal cord inju~. Common methods of producing injtmy in the rat have included contusion, incomplete transection (hemlsectioo), and complete transection. While s contusion model is more conducive to non-invasive trenm~ents sucb as drug delivery, hemisection and transection models lend themselves to implantation of graft tissue or biomatertals. Ovecall, the goal of this project is the development of a collagan-giycosaminoglycan (GAG) implant to facilitate regeneration of spinni cord. Previously, a similar implant has been shown capable ofiuducing regeneration of a functional sciatic nerve in the rat over a gap of at least 10 ram. The objective of this specific study was to compare hemisection and compiem transection models for the assesmaent of the healing promoted by the cohagon-GAG implant. Rats am given a spinal hemisectioo or complete Wansectien at tbe level of thoracic 7-8. Evaluation of the models is based on two factors: 1. Repeatability - the precudure must produce an anatomically constant wound bed 2. Recovery - the procedure must not impose excessive trauma on the animals Specific quantitative endpoints include the ntunber of deaths caused by surgical trauma, the time required post-operatively for the animal to regain reflex bladder function, and histological evaluation of the size of each lesion.

S-58

Poster Presentations

289

292

CYTOSKELETON AND THE RHEOLOGICAL PROPERTIES OF HL-60 CELLS M. A. Tsai, and R. E. Waugh Dept of Biophysics, University of Rochester Medical Center, Rochester, NY 14642 Cell rheoiogy plays en important role in leukocyte trafficking in the mlcrovasculature and cell egress from the bone marrow into the peripheral blood. In this study, we investigated the influence of the cytoskeleton on rheologioal properties of proliferating HL-6O cells. Fractions of HL-60 cells of different sizes were prepared by centrifugal elutriation. Previous work has demonstrated that cell size is closely correlated with the cell cycle. Cell subpopulations in the G1 and S phases of the cell cycle were selected, and cellular rheoioglcal properties were evaluated by single-cell micropipette aspiration (Tsa[ et al, Biophys. J., 65: 2078-88,1993). Differently-sized pipettes ware used so that the ratio between the call and pipette radii was similar for all subpopulations. 100 pM coiohlsine and 30 taM dihydrocytochalasin B (DHB) were used to disrupt miorotubules and microfllamants (F-actin) respectively. Tha rheological properties of HL-60 cells were found to be dependent on call cycle. Cells in the S phase were intrthsioally less deformable than cells in G1 phase. Coiohioine and DHB had differential effects on the proliferating HL-60 cells. Treatment with 100 p.M oolchlsine had-no significant effect on G1 cells, but reduced cytoplasrnio vlscaslty of S cells by - 20% at a given shear rate. 30 p.M DHB dramatically reduced tha cytoplasmic viscosity of both G1 and S cells. This effect was more pronounced on the S cells. Tha cheracteristio viscosity of Gt cells was reduced by - 50% whereas the viscosity of S cells was reduced as much as - 70%. These results are consistent with our previous report that F-ac'dn plays the predominate rola in datarmlning leukocyte rhaological properties. Moreover, we provide the first evidence that microtubules may also have a significant influence on rheological properties of proliferating leukocytes and that the contdbution from the miorotubules depends on tha cell cycle. (Supported by NIH grant No. HL-18208 and Am. Heart Assoc. grant No. F93-219).

CELL BEHAVIOR IN RESPONSE TO CYCLIC SUBSTRATE DEFORMATIONS Huicong Wang and Edward Greed Noyes-GiannestrasBiomechanics Labs University of Cincinnati,P.O. Box 210048,Cincinnati, OH 45221-0048

Celt behaviorin responseto substratedeformationsis importantto understandin orderto elucidate signaltransduedonmechanisms.We havedevelopeda new model of cell behaviorthat describes coil orientationresponseto cyclic substratadeformations.We alsotestedthe hypothesisthatcontact guidancecan preventfibroblssts from rcoriantetiun.Basedon existing experimentalobservations, we postulated that:(I) each bipolar cell hasa strain thresholdnod avoidsany orientationon the culture surface where the pesk-to-pesk axial strain is above its threshold; (2) the cell will randomlyorient within a range of directions wherethe penk-to-peskaxial strainsare lessthan its threshold;and (3) the axial strain~ o l d varies from coil to coil, and is normally distributed in a cell population. The colt model, combined with the substrata axial strain, was used to predict cell orientation distributions for different s~etch amplitudes. The model was also tested by stretch experiments, where human melanocytes were grown on einstie rectangular dishes subjected to uniaxlal cyclical stretches, with amplitudes of 0, 4, 8, and 12% at I Hz for 24 hours. No significant differences were found between the predicted and experimental distributions at any stretch amplitude studied. Similar stretch experiments were conducted using the human skin fibroblssts grown on two types of dishes having smooth and grooved culture surfaces, respectively. We found that the cells on the smooth surfaces oriented away from the stretch direction, and the extent of reorientetion was dependent on the stretch amplitude. In contrast, ftbmbissts on grooved surfaces did not reorient even after 8% stretch for 120 hours. This study supports hypotheses that coil reerlentation in smooth culture surfaces is an avoidance response to excessive cell length changes, and that the failure of ftbroblas~ to reorient in-vivo results from contact guidance provided by collagen fibers.

290

293

THERMALLY REVERSIBLE POLYMER GEL FOR BIOHYBRID ARTIFICIAL PANCREAS Vernon BL, Gutowska A*, Bae YH*, Kim SW* Department of Bioengineering, University of Utah, Salt Lake City, Utah, 84112 USA *Department of Phacmaceutics and Pharmaceutical Chemistry, CCCD, University of Utah, Salt Lake City, Utah, 84112 USA

THE WORK TO FORM BILAYER CYLINDERS FROM RED BLOOD CELL MEMBRANE R. E. Wangh and W.C. Hwung Dept. Biophysics, U. of Rochester, School of Medicine and Dentistry, Rochester, NY 14642.

The development of the Biohybrld Artificial Pancreas is presently an exciting and active area of tissue engineering, The concept behind the biohybrid artificial pancreas combines biological material, the insulin secreting islets of Langerhans, and pelymerie membranes for immuonprctectlon of the islets. Many other systems have received consideration in development of the Biohybrid Artificial Pancreas. However, there are several unreselved problems with these approaches, For one, the life span of these implanted islets is limited and therefore a method to replace the islets when they are no longer viable must be developed. Poly(H-isopropylanrylamide) (PHIPAAm) and many of its copetymers exhibit lower critical solution temperatures (LCST) below physiologic temperature. Below the LCST these polymers form an aqueous solution. Above the LCST (approximately 320C for PN'IPAAm) the polymers precipitate, and, at sufficient concentration, form a permeable gel matrix. A liquid islet/polymer solution below the LCST can be injected into an implanted pouch. Within the pouch at 37~ the polymer forms the gel, immobilizing the islets. To replace the islets, the polymer matrix is cooled below the critical temperature causing the gel to redisselve. The spent solution is withdrawn and fresh islet/polymer suspension injected. Islets from Spague-Dawley male rats immobilized in the copelymer of P(HIPAAm-coAcrylic Acid) (98:2 mole ratio) showed longer viability and higher insulin secretion compared to control islets bee in medium. These results indicate the potential of developing a rechargeable biohybrid pancreas using thermally reversible polymer gels. Acknowledgment: This work is supported by HIH grants IR29DK46458-01AI and GM0839304.

291

The physical integrity of a cell membrane depends on maintaining proper a.smciation between the membrane bilayer and the underlying membrane skeleton. The mechanical formation of biiayer cylinders (tethers) from the cell surface provides a way to measure the strength of this association. (The distribution of membrane components after tether formation and the suinhility or tethers in detergent indicate that tethers consist primarily of a hilayer membrane lacking both a membrane skeleton and associated integral proteins.) The reversible work of separation was men.tared by measuring the force of tether formation under quasi-stoady conditions. A red cell was held in the tip of a micropipette and withdrawn from an adherent bead attached to the tip of a calibrated microoantilever. The deflection of the cantilever provided a measure of the tethering force, and the displacement of the cell projection in the pipette was used to calenlate the tether radius. The tether radius ranged from 15 to 35 nm decreasing with increasing tethering force. At a constant velocity of tether formation, the force increased with the length of the tether, but approached a steady value on an exponential time course. The steady'state pulling force increased with the velocity of formation, increasing by approximately 20 pN for an increase in velocity of 0.1 trots. If a formed tether was held at constant length, a relaxation of the tether force was observed. Extrapolation of the steady state pulling force to zero velocity and mezsuremant of the force after stress relaxation provided independent, but consistent measures of the criticni force for tether formation (55-60 pN), From the critical fcrce and the corresponding tether radius (30 rim) the "reversible" work of tether formation was calculated. The work of separation includes ~,vo main parts: the work of membrane bending and the work of separating the biiayer from the cell surface. We find a bending stiffness of the membrane Of~L9 x 1049 J and a reversible work of separation of --0,2 mJ/m.2 (Supperted under NIH grant No.'s HL 31524 and HL 18208,)

294 EFFECTS OF HYDRODYNAMIC FORCES ON NATURAL AND TISSUE ENGINEERED CARTILAGE G. Vun]ak-Novakovhi, L.E. Freed and R. Langer Massechusetts Inetitute of Technology, Cambrldge MA 02139.

Exogenous forces are known to regulate the cellular secretion of extracellular matrix in natural cartilaga. In our previous studies of cartilage constructs based on isolated cells and polymer scaffolds, tissue morphogenesis depended on flow conditions in the tissue oultura environment. Hydrodynamic fomes can modulate the properties of engineered tissues in at least two ways: by direct effects on cell shape and function, and by changes in mass transfer rates. In the present study, we studied the composition and morphology of cartilage explants and engineered constructs from mixed and static cultures, and made the first attempt at distinguishing the effects of hydrodynamic forces on tissue morphogenasis from those of mass transfer. Over 4 weeks of cultivation, cartilage explants and engineered constructs showed analogous responses to hydrodynamic forces. In both groups, mixing resulted in significantly higher amounts of all tissue components (cells, glycosaminoglycan, collagen) when compared to the corresponding static cultures. Mixing also induced the formation of an outer capsule containing a high concentration of flattened cells, the thickness of whioh increased with cultivation time. Tissue morphology end composition could be related to levels and duration of shear and mixing dudng culture. Thase data imply that cartilage tissue bioreactors should provide efficient mass transfer in conjunction with controlled hydrodynamic shear at the tissue surface.

TRANSIENT INTERACTIONS OF SICIG.,E ERYTHROCYTES W1TH ENDOTHELIUM UNDER SHEAR STRESS Richard A. O. Montes,/ames R. Eckman, and Timothy M. Wick School of Chemical Engineering, Georgia histimte of Technology, and Division of Hematology/Oncology, Deparanent of Medicine. Emofy University School of Medicine The pathophysiology of sickle cell v~o-ocelasion is complex and likely the result of multiple cellular and humoral factors. One of the most important is the ahoofrmfl adherence of sickle red blood cells to vascular endctbufiam. We posit that the significant events leading to edhegenen occur dating the esdy COntaCtOfsickle cells with endotbefiam. In order to study the kinetics of sickle cell ~ , we have developed an optically munito~l cone-andplate apparatus to moasme the rste of atltachment of sickle e~Fthmcytes to h ~ Ombilioal vein endothelial cells under shear mess. Washed sickle red cells are ccotlnoooaly recirculated over endothelial cells and the namber of sliding (transiently intemeting) and firmly adbetent cells is quantified over 15 miautes. Then, the shear stress is increased to ~ m ' e the detachment rate of finnty adburent cells. This system ls used to moasme and comp~e the kinetics of sickle attachment and detachment via some of the known pethways mediating sickle cell adhesion, anmely, VCAM- I/VLA-4, thtoml~spo~ltn, and activated VLA4. inlllal reanits show that at 0.8 d y ~ 2, the number of firmly adherent cells linearly increases with time. Beyond tea minutes of recircula~g flow, the slope of adhesion became considerably sleeper. This suggests that either expoenm of endothelium to flow of the repeated enceunler with red bleed cotls peedisposes the endothefium to increased adberence. AS shear stx~e~swas incroasud at 0.4 dyan/cm 2 increments, the adherent red cells initially detached from the endothefium. However, attractive bonds wea'equickly ~establlshed resulting in enhanced sickle cell adhesion even with further increase in shear swess. These data mggest that wansient interactions an: important in the establishment of fwm sickle cell a#~te.rence. Studies such as these will contribute to a ~ complete u n ~ t u n d i n g of the biophysics of sickle cell adherence and microvascuinr occlusion in sickle cab anemia.

Poster Presentations

S-59

295

297

A MATHEMATICAL MODEL OF CALCIUM OSCILLATIONS IN ENDOTHELIAL CELLS RESPONDING TO FLOW Theodore F. Wiesoer, Bradford C. Berk, and Rohet~ M. Net~m Bioengineering Center, Georgia Institute of Technology, Atlanta, OA 30332-0405, USA

DETACHMENT FORCES AND MECHANISMS OF CELL-CELL ADHESION MEDIATED BY F ~ (CD16) I S O F O R M S A T T H E S I N G L E B O N D L E V E L Cheng Zhn, Scott Chesla and Periasamy Selvaraj" School of Mechanical Engineering, Georgia Institute of Technology, Atlanta, OA 30332-0405 *Department of Pathology. Emor/University School of Medicine, Atlanta. GA 30322

Some of the most interesting phenomena in cellolar calcium dynamics are the periodic or chaotic fluctuations in cytesolic free calcium level in response to stimuli. The non-random nature of these oscillations suggests that information conveyed in signal recognition and traneduction processes may he encoded in the frequency and amplitude of the fluctuations. A mathematical model has been developed which simulates calcium transients in endothelial cells as a function of extracelhilar signals. This model was osod to investigate how cellular processes and exlxacellniar signals influence the oscillstory calcium transient. A base model was originally reconciled to a non,.oscillatory benchmark transienL No oscillatory responses resulted when the base model was subjected to relevant physiological ranges of shesr stress and agnnist concentration. A linearized stability analysis revealed that the base model was inherently non-oscillatory. However, selected modifications to the values of file constenLs, without changing the structure of the model, yielded a stable oscillatory solution. Both the amplitude and frequency of the escilintions increased with increasing concentrations of agnnist. The model predicts file effects of drug thapsigargin, which abolishes calcium oscillations, Neither capacitative calcium entry nor calclum-indoced calcium release were sufficient to elicit oscillations in the base model. However, these phenomena facilitated floctuations in the model that was inherently oscillatory. Shear stress increased file amplitude and frequency of the oscillations by increasing file delivery of agonist to the cell. In contrast, by enhancing calcium influx, high levels of shear stress dampen the oscillations, causing file solution to attain a steady state level. The set of constants which produced the oscillations is not necessarily unique. These numerical studies indicate the potential for defining domains for file parameters which guarantee the existence of oscillations. An analytical criterion establishing necessary and sufficient conditions for oscillations may thus be possible,

296 AN IN VITRO CYTOTOXICITYASSAY OFTHREE POLYMERIC, BIOABSORBABLE FILMS

Wyl/e KB, Spodntek GJ, London AP, Tonelll AE. Hudson SM. Gupta BS North Carolina State University, Raleigh, North Carolina Bloabsorbable films have potential use as wound dressings. Experimental bioabsorbable polymerte ~ compc~ed of either chltosan (CH), poly D,L-lactlc acid (PLAt. or poly.t caprolactene (PCL} were evaluated for toxicity in vitro u s i n g human dermal flbroblasts. Toatclty of film extract solutions and direct contact toxicity were Inves~gated. Films were sterilized by ethylene oxkle or,t irradiation. Film extracts were produced by Incubation of the films with culture medium for 72 hours and 21 days at 37"C. Dilutions of these extracts were Incubated with cell monolayers for 72 hours at 37"C. Trypan blue staining was then used to determine viable cell numbers. A statiatically significant decrease In viable cell numbers was not observed In any cell cultures Incubated with "t Irradiated film extracts. The number of viable celts was significantly lower In cultures Incubated with ethylene oxide sterilized CH film extracts (p< .05). Direct contact toatclty was assessed by Incubation of h u m a n dermal llbroblasts with film samples. After Incubation at 37~ for seven days, hiatniogte staining of the films was uaed to qualitatively evaluate cellular adherence and morphology of the adjacent cell monolayer. Monolayer formation was observed on and adjacent to the PLA and PCL indms regardless of sterilization method. Monolayer formation by human dermal llbroblasts was not observed on either "( Irradinted or ethylene oxide sterfl/zed CH films, although a healthy monolayer was observed adjacent to these films. No In vitro toxicity was observed with PLA and PCL films In thls study. Ethylene ox/de sterilized CH film extracts had a toxic effect on h u m a n dermal flbroblasts, A cell monolayer was not observed on CH films regardless of sterlIlzatlon method, This study supports in vlvo evaluation of poly D, L-lactte acid and poly-e caprolactone films as bioabsorbable wound dressIngs and further tn via,o Investigation of direct cellular Interactions with chitosan films.

9Effec.ts of sir~. rural variations in Fc'f receptor HI (CD16) on file detachment foi'ces and mech .antsms revolved m cellular adhesion were investigated at file single orossbridge level. CDI6 ts one of only four known receptors which naturally exist as both la'ensmembrane (TM) and glycosyl phosphatidylinositol (OPI) plgsma membrane anchor isoforms. The strength of individual mo.lecniar interactions between a Chinese hamster ov'~y (CHO) cell transfected with a given CDI6 tsoform (TM or GPI) and a human red blood cell (RBC) opsonized with a given type of CDI6 ligand (low affinity IgG or high affinity anti-CDl6 antibody Fab fragments) was evaluated using a micropipatte technique which uses file highly deformable RBC as an u!tr~, e .n.~trve tunable transducer capable of measuring pieonewtoo range forces: A binomial dlstnlmtton based stochasuc model developed to describe bond formation probability predicts that when the adhesion frequency between individual CHO and RBC cell pairs decreases below 25%, file number of bonds formed in any observed adhesino approaches one. By rep',.at..,-,~_lyb.'5"ngL-.g together individual cell pairs to form a point contact, changes in binding frequency with increasing attampts'(up to 250) were observed. O~r data show that for interactions between CD16 and low affinity ligands (weak bonds), file adhesion freqoency remains constant with increasing attempts, soggesting reversible bond formation and detachment. We interpret this reversible behavior as,detachment at file specific receptor-ligand binding site. For interactions .between CDI6 and htgh affinity llgands (slsong bonds), adhesion frequency decreases with mcrensmg attempts, suggesting an ineversible loss of available binding sites. We interpret this as evidence of bond detachment via receptor anchor uprooting and increasing occupation of liga~ binding sites by these uprooted receptors. The detachment forces range from 2 0 p N to 50 pN, comparable to file single bond and/or membrane anchor strengths published in file literature. (Suppened by NSF grant No. ECS-935037 lo CZ, NIH grant No. AI R29 30631 to PS, Emory/GT grant to CZ and PS, and NIH Training Grant No. GM 08433 to SC.)

S-60

Structure and Flow by Three-Dimensional Echocardiography

298

301

THREE-DiMENSIONAL ULTRASOUND AND FLOW IMAGING OF PROSTHETIC VALVES: IN-VITRO STUDIES Robin Shandas, Jeffrey Kwon, Ole Knudson, Michael Jones, Litliam Valdes-Croz The Children's Hospital, Denver, CO, NHLBI, Bethesda, MD.

EFFECT OF LEAFLET FUNNEL GEOMETRY ON THE COEFFICIENT OF ORIFICE CONTRACTION IN MITRAL STENOSIS D. GIIon, E.G. Cape, M.D. Handsehumacher, L dlang, C. Sears, d. Sothelm, F- Mcrds, J.T. Strebal, A.E. Weyrcan. R. A. Levlne. Massachusetts General Hospital, Boston, MA; University Of Pittsburgh, PA; Santln Englnsedng, Peabody, IdA.

Three-dimensional (3D) reconstruction of echocardiographic images promises to provide clinicians with the ability to visualize cardiac structures in 3D. This investigation examines the utility of 3D ultrasound and ltow imaging of prosthetic valves es a method to comprehensively assess valve behavior both qualitatively by providing muiti-faeeted 3D views of valve motion and flow, and quantitatively by allowing detailed measurements of opening areas. Bio-prosthatic (porcine aortio valves) and mechanical (tilting disk, biteaflet) valves of various sizes were mounted in a pulsatile flow chamber (70 bpm, 60 co stroke volume) that allowed for 3D ultrasound imaging of the valves using an ultrasound scanner interlaced to a 3D reconstruction system. Multiple ultrasound images ecquired by mtationally moving the ultrasound transducer in 1" increments ware processed and displayed in pseudo 3D form. 3D ultrasound imaging provided subedor detection of normal and abnormal valve motion as well as greater recognition Of valve structures such as individual orifice areas for the mechanical valves when compared to 2D echo images. Valve opening areas measured by 3D ultrasound also compared well to true anatomic opening areas measured by digital video planimetry for all valve sizes and types (y---0.93x + 1.204; r=0.94; SEE=O.t6 cm~). 3D reconstruction of transvaivular flow paltems cleady revealed opening characteristics of the valve leaflets including abnormal motion such as leaflet immobility and partial opening. 3D ultrasound and flow imaging of prosthetic valves is superior to conventional 2D echocardiography, providing detailed resolution of valve motion as well as greater accuracy in quantifying opening areas.

Three-dimensional (3D) echo reconstruction can potentially allow u s to a d d r e s s uniquely 8-dlmenslonal questions: for example, the effect Of 3D leaflet geometry proximal to the limiting orifice Of a stenotlc valve on the coefficient Of orifice contraction (Cc). This is Important because: 1) for a given flow rate and anatomic area. a lower Cc givss a higher mmdmal veleolty, pressure gradient, and poealble head loss; and 2) Cc, assumed ec~stant In the Godln equafion, may vary with valve shape. To date, It has not bean possible to study this effsct with the actual 3D shapas OfvaIvse In patlsnts. We therefore used a spark gap system with respiratory and ECG gating to reconstruct leaflet funnel geometries typically seen in patients with mltral stenosls (MS) at maximal leaflet opening. Valve models were then constructed by stereoJifhography (eompute;~zedlaser polymerization) and studied hemodynamlcaity. Results: Cc vaded minimally with flow rates, but prominently with shape: (t) For any anatomic area, Cc was larger for tapered dome geometries, which allow more gradual flow convergence proximal to the limiting odfice than relativaly fiat leaflets (more perpendicular to flow). (2) For each funnel shape, Cc Increased with Increasing orifice to proximal funnel size (more tube-like). Conclusions: Cc, which can alter net head loss, IS Importantly affected by the variations In leaflet geometry seen in patients. Echo stereolithography can allow us to address such untquely 3D questions,

302

299 PROXIMAL CONVERGENCEANALYSIS OF ELLIFTICAL ORIFICES IN THREE DIMENSIONS KA. Powcll, P.M. Vandervcort, S,N. Ganobeik, J.D. Thomas Cleveland Clinic Fonndadon, Cleveland Ohio Introduction: Proximal convergence analysis is a quantitativeteehhique for predicting regurgitantflow rates observed in miual regtwgitation.It is based on hemispheric flow convergence of isovr shellstowards a finiteorifice.It has previously been shown that the assumption of hemispheric flow convergence is valid for flow through a circularorificein a planar surrounding. However. in the case where miU'al regurgitationoriginates stong the entirecoaptation iinc of tic leaflets, the rcgurgitant orifice may be better represented as an 9 The

purposeof this studywas to evaluatethe effectof orifice eccen~cit7 on the shapeof the isovcIocity shells and to testthe ~sump~on of hemisphericaxlsymmee7 of the proximal flow field. Methods: 2-I) color Doppler flow images were obtained for Ig rotational angles at 5~ increments. The ultrasound prct~ was aligned along the centeriineaxis of the orificein the in vitro flow model. Three dimensional flow fields were reonnstrocted from the 2-D images. Data was collected for a set of orifices with varying aspect ratios (1:1, 1:2, 1:3,1:5, I:10) and constant cross-sectional area (0:3 cm 2, l cm2). Flew rates ranged from 80-210 ml/s. Three dimensional fluid dynamic simulations were developed for a similar set of orifice and flow rate conditions. The centeriian velocities (Ve) between 5 and tO% of the orifice velocity were used to predict flow rates(Q=2zr 2' v,). Results and Dlscussion: Tic three dimensional shape of the isovelocityshellstend to flatten out across the orifice as the orifice eccentricity increases. This leads to a decrease in the predicted flow rate from that observed for a circular orifice. This deviation was observed to he on the order of 35%+ 2 for eccentricities of 1:5 and 1:10. Future work will focus on developing correction factors to the hemispheric flow assumption for these eccentric orifices.

300

STRUCTURE AND FLOW BY THREE DIMENSIONAL ECHOCARDIOGRAPHY Steven L Schwartz MD, Natesa G. Pandlan MD Tuf~ University, New England Medical Center, Boston MA A pitfall common to aii forms of cardiac imaging is that the results yield 2dimensional images of a three dimensional structure. The observer is required to mentally reconstruct those images in three dimensions. Recently, a method has been developed which provides volume-randered, dynamic, three dimensional reconstructions from two-dimensional echocardiographlc data. Briefly, serial 2-D images gated to the cardiac and respirator'/cycle are obtained in sequence from trans-thoracic or transesophageal echocardicgrams. Image acquisition employs either serial parallel slices, fan-I~
THREE-DIMENSIONAL ECHOCARDIOGRAPHIC RECONSTRUCTION OF COLOR DOPPLER FLOW JETS: AN IN-VITRO STUDY Alain Delabays MD, Steven L Schwartz MD, Takahiro Shiota MD, Dag Teien MD, David J Sahn MD and Natesa G. Pandian MD Tufts-NEMC, Boston MA and Oregon Hlth Sci Univ, Portland OR. Quantification of intracerdiac blood flow jets using color Doppler imaging has a number of limitations inherent to the Doppler technique. In addition, this appmech provides only 2-dimensional pictures of a 3-dimansional structure. Recently, a method has been developed which provides volume-renderad, dynamic, 3-dimensional reconstructions from 2-dimensional echocardiographic data. To explore the potential of this new technique, we studied the effects of different onfice shapes on the geometry of ragurgitant flow jets generated by a pulsatile flow pump. From the 5 different orifices, 3 were flat: circular (C), rectangular (R) and triangular (3") and 2 were asymmetric mimicking a mitral valve prolapse (MVP) and an aortic valve regurgitation (AR). Ten flow rates (1.3-14 Vmin) were examined for each odfiee. Sequential 2-D color Doppler images were acquired using a rotational scanning and processed in a dedicated computer (TomTec Imaging System), Thrae-dimensional raconstmction of the flows nicely portrayed the geometry of the different components of the jets: flow convergence region (FCR), vena contracta (VC) and body of the jet. Shape of the VC corresponded well with the shape of the orifice. FCR was symmetric for the C valve and asymmetric for all the other odfices with changes in shape during the cardiac cycle. There was a tight correlation between the 3-D derived volume of the FCR and actual flow rates for each odfice (r=0.95-0.98). There was also a good correspondence between the actual peak flow rate and the 3-D derived volume of the jets for the flat onfices (r=0.92-0.94) but not for the asymmetric valves. We conclude that volumerendered 3-D reconstruction of flow jets may overcome some limitations of conventional 2-D color Doppler and improve quantification of ragurgitant lesions.

MRI of Tissue Structure and Function 303

S-61

305 DIFFUSIONTENSOR MRI: A NEW METHOD TO ELUCIDATE TISSUE MICROSTRUCTUREAND TO ASSESS ITS PHYSIOLOGIC STATE Peter J. Basser Biomedical Engineeringand Instrumentation Program, NCRR, NIH

Diffusion Tensor MRI (DTMRI) is a new imaging modality that consists of estimating an effective diffusion tensor, 12, in each voxel, and using it m elecidate tissue microstructurr and to assess its physiologic state. Specifically, we can characterize the degrees of diffusion isotmpy, diffusion anisotropy, structural similarity, and fibor-tmct organization within tissues using quantitative parameters derived from D, all of which arc scalar iuvarisots (i,e., independent of the choice of the laboratory coordinate system), that can be used like quantitative histological or physiological stains. We derive these parameters by decomposing D into its isotropic and anisotmpic tensors: 13 = <1:3>/, + ( D I) . [11

isolxoldc anis,mopic The isotropie tensor is the product of the identity tensor, i, and the scalar mean diffusivity, : < I ~ = Tr(12) Dxx+Dyy+Dzz ~.1+~.2+;k3 3 3 = 3 [2] where kl, ;L2,and ;k3 am the eigenvalans (principal diffusivitius) of D. The anisotropic part of D is the diffusion deviation tensor, D ; it measures how much D deviates from isotropy. Invariant (scalar) measures of tissue isotropy, anisotropy, structural similarity, and fiber-tract organization are constructed from metrics of the teusors appearing in Eq. [l]--in particular, their generalized tensor products. Applications of this method will be presented. These include measuring and monitoring structural and physiological changes (at macromolecular, cellular, tissue, and organ length scales) in development,aging, and disease.

304 MICROSCOPIC NMR TISSUE CHARACTERIZATION BY IMAGING OF ANISOTROPIC DIFFUSION AND CHEMICAL COMPOSITION Paul C. Lauterbur, M. Joan Dawson, Yihong Yang, Su Xu, Vikus Gulani, Joshua S. Shimony, Joseph A. Kmieeik, Carl D. Gregory and Zhi-Pei Liang University of Illinois at Urbana-Champalgn Microscopic diffusional and spectroscopic imaging, at 500 MHz and 300 MHz, have been used to study phantoms and biological specimens. Among applications to be described are physiological studies of the excised rat uterus and cervix by imaging of anisotropic water diffusion, rapid measurements of the water diffusion tensor in excised rat spinal cord, and lactate imaging in phantoms and muscles by use of the GSLIM localization technique with spectral editing methods. Multi-parameter magnetic resonance characterization of tissues and organs at microscopic resolution in physiologically-relevant states is bringing new opportunities to microscopy, to normal and pathophysiology, and to the understanding of clinical MR investigations.

STUDIES OF THE STRUCTURE AND FUNCTION OF CARTILAGE BY MR D. Surstein, M. L. Gray Beth Israel Hospital and Harvard Medical School, Boston, MA, and MIT, Cambridge, MA The ability of cartilage to function as a lead bearing material is related to the structure of the extracellular matrix, which consists of collagen with attached pmteoglycans (PG). Our goal is to measure the content and structure of the matrix with magnetic resonance (MR) techniques. Here, we give the basis for 3 MR techniques, and present studies of tissue degradation: (1) Pmteoglyoans have abundant ionized side chains, conferring a negative fixed charge to the tissue. Therefore, cation concentration is enhanced within the tissue, forming the basis of the measurement of PG content with MR measurements of tissue sodium (Na'), (2) The restricted motion of water associated with macromolecutss is detectable using the MR technique of magnetization transfer (MT). The MT effect is mainly determined by tissue collagen content and st~lctura. (3) MR is inherently sensitive to water diffusion (D). The effective D of water in cartilage decreases as the lime for diffusion is allowed to increase, indicative of restriction to diffusion in the tissue. The absolute value of D at a given diffusing time is dependent on tissue hydration. These MR parameters were measured as cartilage was degraded by trypsin or interleukin-l~. The PG content decreased to approximately 0 in both cases. D of the tissue water increased the same amount, reflecting the increased hydration. Despite similar changes in PG and collagen content and tissue hydration, the MT parameter increased with trypsin, but decreased in II.-1 treated samples, suggesting that MT reflects some differential change in the collagen structure in these samples. Our data continue to support that the combined use of Na, D, and MT MR may provide an early and sensitive indication of cartilage degradative changes, and may provide a means for evaluating therapeutic efficacy.

S-62

Medical Image Reconstruction and Processing

306

309

DYNAMIC MAGNETIC RESONANCE IMAGING BY MOTION MODELING AND ESTIMATION IN (k,t)-SPACE Z.-P. Liang, H. Jiang, and P. C. Lauterbur* Department of Electrical and Computer Engineering *Biomedical Magnetic Resonance Laboratory University of Illinois at Urbana-Champnign Since its inception in early 70's, MR/has revolutionized biomedical imaging over the past two decades. But its application thus far has been mc~tly limited to stationary objects. For time-varying objects such as the heart or abdomen, significant image artifacts arise which often render the image usele~. To overcome this problem, many motion-compensated data acquisition and post-proce~ing methods have been proposed. The mc~t successful ~pproachs thus far are perhaps the adaptive data correction method using navigator echoes and the (k, t)-spa~e method. In spite of this progress, imaging of time-varying objects remains one of the mo~t challenging problems. This paper pre~ents a novel dynamic imaging method using motion modeling and estimation. Specifically, we propose to model the time-varying objects with a generalized harmonic model in (k,t)-space. We have shown that with this model, the desired time-varying image function can be recovered exactly from temporally under-sampled data if certain condition is met. We have also developed a novel MR data acquisition method using phase-encoded navigator signals. This data acquisition scheme covers (k,~)-space effectively so that sufficient data can be collected for estimating the model parameters, making it possible to obtain high spatial and temporal resolution simultaneously. Experimental results from phantom and biological objects will be presented st the conference.

ACCELERATION OF ML-Ebl BLIND DFA:ONVOLUTION FOR 3D LIGHT MICROSCOPY $~losh B hattachaty~l, D. Szal~wski2, p. ~ y 2 ,

J. T m m ~ i~1T. Hohn~ 1'3

1 Biomedical Engineering Del~Ime~t, JEC 7049, RPL Troy NY 12180 Wadaworth Lalmcatodes" New York State Department nf ltealth, Alhany NY 12201-0509 3 Autoquam lmaging Inc. Troy NY 12180 Holmes el a/have hilmduned the ML-I~! itetative blind decouvoletim algorithm fne 31) microscopic h ~ g o reconstmetioe. Thts method r~omm~ts both ~ objea and the l x ~ t spired function (PSF) of the system. "rhe edvantage over othe$ non-blind a l g ~ i ~ is that it eliminates the ledious and compficated protocol nf meusming the PSF and thereby makeg e v e r y usage feasible. The limitation has b e ~ that It require~ maay itemticv~ to ~ a vergo. A pgelimlnmT ~lldy to agcelefale the algorithm has ber p e l f ~ ~ g~eleraflon swategy i n o n t p o ~ a stm~ha'~ lin~ sem'ch ter_hnique. The ~on-acceleraled alg~d~m emally generates a dusited reco~m~ted image at arotmd 300 iteratio.. We obtnined a a~mparable image mc~atmction (judged by object delineation, baekgroond removak edge sharpening, and axial de,bluffing) at around 50 iteration using the accelerated verslau nf,O~ algmlthw. By comparing the iteratica number it appears that we have gained a 6 fold speed increase, although, the tlme taken per itexation has also gone up from -30 seonnd to -.45 secoud (oa an IBM RS6000 SP2 architecu~) due to the addition of the llne sem'ch. ~ o t ~ we estimate an overall increase of 4 times in r speed.

310

307 LOCALLY FOCUSED MR/: APPLICATIONS TO SPECTROSCOPIC, INTER'v~NTIONAL, AND FUNCTIONAL IMAGING Lian Yao, Yue Cao*, and David N. Levi. Department of Radiology, University o f Chicago, Chicago, IL *Departraellt of Neurology, Henry Ford Hospital, Detroit, MI Conventional magnetic resoa~ce images ~ recous~vc(ed by Your/er u~nsformation and have uniform spatial resolution scro~ the entire field of view (FOV). This paI~" describes 9 way of creating MR images which have higher spatial resolution in some areas than in others (1). Such locally focused tmagus can be acquired i. less so.iratime than that requhed to image the entire FOV with tmifcrmly high resolution. A/~er ~he user specifies the spahal resolution desired in each portion of the FOV, the algolSthin ~Wmsfically generates image basis functionswhich oseillxte mo~ rapidly in the zones with l~,hest resolution, Images are reconsla~aeted by summing image projections ~ t o tbese basis fuectio~; these ate calculated from 9 subset of the usual phase~neodod signals ~,qnired to create 9 uniformly well-re~lved tmage. The algorithm also ~ i n u s which pha.~..ear ale optimal for this purpose, and these are usually non-uniformly scattered in k-space. Thus, both dam acquisition and unage r e c o ~ c t i o n are optimized. The new method t~ more efficient thila r Fourier imaging because "resolving power" isdism'butedm tho~ Leas likely to have sharp edges, thereby avoiding tnmcation artifacts. In contrast, Fottrier images have the fame spatial resolution everywhere, thereby "wasting" resolving power on slowly varying regions at the expense of regious which contain sharp edge~ and produce mmcatice artifacts. Funr and interventional imaging may benefit from the locally focused technique, which makes it possible Io acquire a roptd senus of dymmucal images which have high re~olution hl a/eas of expected change and lower resolution elsewhere. Spectro~pic images may be improved by using high resolutionin the neighborhood of sharp edges (e.g. scalplipids) which might otherwise ca~e mmcation artifacts. 1. Cao Y., Levin D.N.,/n "Proc., 1994 IEEE Intern.at Coal on Image Proc.", IEEE Comp. Soc. Press. Los Alamitos, CA, 30-34,1994.

308 TOWARD R E A L - T I M E K - S P A C E S A M P L E S E L E C T I O N IN M R S P E C T R O S C O P I C IMAGING Stanley J. Reeves and Rahmi Eezar Auburn Unlverslty, Department of Electrical Engineering, Auburn, AL 38849

Reconstructed images from MR spectroscopic imaging (SI) can be improved significantly by incorporating a priori informatlon such ~ region of support or smoottLue~ constra.lnte into the reconstruction process. However, ;n the presence of a priori infor~atlon, certain irregularly spaced sets of k-spaue samples will provide significantly more information about the original image than a regularly spaced set. The purpose of this work is to exploit a priori information about the object being imaged to choose the set of k-spaue samples that minimizes the error in the reconstructed image. In previous work on this problem, we developed an e~r opt imisatlon technlque that started with a set of possible candidate k-space locations and eliminated points from the candidate set until the desired number remained. In this work, we develop a new criterion based on a Wiener filter reconstruction that allows us to select rather than eliminate kspace locations. In contra~t to the previous criterion, this crlterion is defined when the number of ~mp|es is less than the number of u~z~now~ in the ira~ge. Therefore, we can use sequential selection rather than elimination. As a result, the imaging device can begin to acquire each selected sample while the next ~ m p l e is being selected. This optimisation strategy allows the possibility of real-time sample selection, since the imaging can begin immediately without waiting for the entire sample selectlon procedure to be executed first. We demonstrate the power of this technique using simulated and phantom MR[ data.

TOWARDS QUANTIFICATION OF MYOCARDIAL DEFORMATION FROM 4D IMAGE D A T A James Duncan, Pengcheng Shi, Todd Constable and Albert Sinusas Yale University, New Haven, CT 06520 The accurate quantification of lraasmuralleft ventricular (LV) regional function is crucial for managing patients with ischemic heart disewse. Previous imaging- b~sed efforts have been hampered by the limitations of conventional two- dimensional imaging and inadequate image analysis methodology (e.g. the inability to handle out-of-plane motion). More recently, these limitations are starting to he o~rcome by three- dimensional (3D) imaging strategies and methods to estimate pointwise myocardial displacements and/or velocities. However, the task of assembling dense 3D sets of image- derived displacement/velocity data into transmural, quantitative measurements of LV function remains a difficult one. We will discuss initial efforts towards the development of an approach aimed at integrating several sources of displacement and velocity information from one of several imaging modalities. The approach uses biomechanical Finite Element models, computer visionrelated strategies based on differential geometric features and mathematical optimization rea.~ning methods to estimate the non-linear, non-rigid deformation of the left ventricle of the heart. Results regarding the estimation of 3D shape- based displacements from cine Maguetis P~sonance Imaging (MRI) and their comparison to gold standard implanted markers and posl movtem T T C staining will be presented, v.s will initial results using a 2D version of the integrated approach that incorporates both shape- based displacement and phase contrast velocity MRI data.

Technical Advances in Magnetic Resonance Imaging

S-63

311

313

MRI OF CORONARY ARTERY FLOW D. Burstein Beth Israel Hospital and Harvard Medical School, Boston, MA

FUNCTIONAL MRI OF THE HEART: DETECTION OF CHANGES 1N CORONARY FLOW VELOCITY AND MYOCARDIAL TISSUE PERFUSION DURING HYPEREMIA B,P. Poncclet, P.Nieml, G.Zervos#, R.M. Weisskoff, TJ. Brady & H. Kantor# Massachusetts General Hospital NMR Center and #Cardiac Unit Boston MA 02214

Recent advances in magnetic resonance coronary angiography have led to successful depiction of at least the proximal sections of the coronary arteries utilizing both 2D (multislice) and 3D imaging techniques. We have been implementing a multislice segmented k-space approach in which multiple phase encoding steps are acquired during a bnef period of diastole dudng each of a series of sucoessive heart beats. Typically, 120 phase encoding steps (120 x 256 matrix) are acquired (8/heart beat for 15 heart beats) with a field of view of 240 x 240 ram, resulting in an in plane spatial resolution of 2.0 x 0.9 mm. Each 2D image is acquired with a 3-4 mm slice thickness and 0-1 mm slice overlap. With this technique, several large clinical studies have documented the sensitivity and specificity of the MR approach as compared to conventional coronary angiogmphy to be in the range of 70 100%, depending on the vessel under study. However, these techniques have still not achieved widespread clinical applicability. Some of the probiams inherent with the current techniques include the difficulty in identifying the coronary arteries in plane in single slice images and the need for significant post-processing to obtain a full angiographic view of the vessels. Previous attempts at development of a true angiographic technique (i.e one in which blood is tagged in the coronary osfia and imaged as it flows into the coronary arteries) have also had limited success, This is due to the small size of the bolus of blood tagged in the osfia, relaxation effects of the tagged blood as it moves within the coronaries, and difficulty of implementation of the techniques on standard clinical systems. We will demonstrate preliminary studies utilizing a stimulated echo flow tagging technique which has the versatility to address these problems. -

Non occlusive coronary artery lesions cause symptoms by limiting the increase in flow necessary to offset an increased myocardial oxygen demand. Current methodology for evaluating this decrease in coronary flow reserve relies on either anatomical measuremant of coronary lesions, or radionuele.ar examination of tracer distribution or wansit (thallium and PET). The purpose of the study is the development of a noninvasive clinical MILl stress test to measure coronary flow reserve (CFR) and detect changes in myocardial tissue perfnsion. Measurements of comna.'-y flow velocity (CFV) changes induced by dlpyridamolr are presented. using a time-of-flight (TOF) method, h correlation to these measurements, are also presented the simultaneous changes in T2* in the myocardium due to changes in blood oxygen content, Short axis studies of the heart were performed with a gated, flow compensated, gradient echo echo planar imaging (EPI) pulse sequence. Series of 25 images were collected in mid-diastole and during breath-hold using the TOF technique. The lust images of each series served as T2*weighUrd image ('rE=29ms/TR~) for myocardial signal intensity meesarernents. The next 24 images served for CFV measurements. After baseline measurements, 0.56 mg/kg of dipyridamole was infused I.V. over 4 mins. The effect was monitored every 2 mias for 10-20 rains, then reversed with 50-75 mg of aminophylline. We examined a group of 10 healthy subjects. CF'V were measured in the left anterior descending coronary artery (LAD). The peak change in diastolic CFV occun~l at post dipyr. infusion times ranging between 3 and t3 mths (7+a miss). The average peak change in diastolic velocity, CF'v'dip/eFVrest was 3.94-1.6 (n=10). The temporal profile of myocardial signal intensity changes closely correlated with the measured CFV. All subjects demonstrated a transient increase in myocardial signal intensity after dipyridamole with a maximum 10-15 miss post infusion. The average increase in signal intensity from baseline to the peak dipyridsmole effect was 14 + 7 % (n=10), producing a change in I/T2* of- 4.6".t:2.0s "1.

312

314

Ultrafast Vascular Imaging using 3.D Segmented Echo Planar Imaging P.A. Wielopolski, P.V. Prasad Dept. Radiology, MRI, Beth Israel and New England Deaconess Hospitals, Boston,/viA.

THREE DIMENSIONAL MAGNETIC RESONANCE IMAGING OF BONE MINERAL Yaotang Wu, *,~ Jerome _L.. Ackerman, a David A. Chesler, a Leonr Garrido, ~ Melvin J. Glimcher? Aiping Jiang, a Hung Jiang,* Jun Li,a Robert M. Neer,= Chandrasekhar Ramanathan,=," Jinxi Wang b "Massachusetts General Hospital, Boston, MA?Children's Hospital, Boston, MA;~Ma~a cimsetts Institute of Technology, Cambridge, MA

Conventional magnetic resonance anglography (MRA) techniques suffer from motion artifacts (ghosting and blurring) and signal loss due to flow related dephnsing. Enhanced performance of new generation magnetic resonance imaging (MRI) scanners allow short echo times (reducing flow related artifacts from turbulent flow) while maintaining high signal-to-noise ratios and resolution. Echo planar imaging (EPI), a technique that permits image encoding with sub second data collection times (30-100 ms) has been considered for MRA, Nonetheless, high rnsolution MRA with single shot EP[ is constrained technically by the speed and maximum gradients of the MR scanner, and physically by neuromuscular stimulation and artifacts arising from flow during the data collection time and signal loss and geometrical distortions from poor magnetic field homogeneity, Recently, new data acquisition strategies that combine the principles of conventional MRA and speed of EPI have permitted a several fold reduction in the image encoding time compared to conventional scanning protocols while keeping the diagnostic quality. We have applied such segmented EPI sequences to perform MRA in regions where motion degrades image quality on conventional scanning, such as in the abdomen and thorax or in studies on uncooperative patients. These techniques when further combined with STAR (Signal Targeting with Alternating Redio-fteqnency), a projection time-of-flight subtraction method, led to angiograms with complete background suppression, Excellent stationary signal suppression improves the conspicuity of small vessels and has been shown particularly advantageous for mapping perfusion in organs such as the brain and the kidneys. With this multi-shot approach, speed and resolution can be balanced adequately to complete scanning in a single breath-hold with low sensitivity to artifacts from flow and magnetic susceptibility. When combined with magnetization preparation, the entire coronary artery tree can he visualized in a single breath-hold. High quality 3D vascular imaging can then be generated with isotropie voxels that permit the reformation of the 3D data set in any desired view using multiplanar reconstructions and maximum intensity projections (MIP).

The mineral phase of bone, like other solids, is generally regarded as beyond the purview of conventional magnetic resonance (MR) imaging or speetrotcopy because of the difficulty of eliciting and acquiring MR signals from nuclei with exceedingly short T2's (spin-spin relaxation times). However, special solid state measurement technique~ we have been developing show great promise for eventual use in human subjects. Imaging of phosphorus yields a direct quantitative (within 1-7 percent of dual energy x-ray absorptiometry) measurement of bone mineral density, and does not rely on assumptions about the x-ray attenuation of a variable mixture of bone, soft tissue and fat. Spectrcecopic analysis or chemically selective imaging can provide information on the chemical composition of bone mineral or synthetic calcium phosphate prosthetic material. This composition may be of interest in evaluating the maturity of the mineral (as reflected in the H P 0 ~ 2 / P O ~ 3 ratio, for example) in a healing fracture or in the rapid growing bones of a child; a second application is in the measurement of the amount of residual synthetic calcium phosphate present in a prosthetic material which is subject to remodeling (this material and the patient's natural bone may be radiographically indistinguishable, but e~tiy distinguishable by MR). Although it cannot compete with methods such as computed tomography or dual energy x-ray absorptiometry in terms of spatial resolution, solid state MR imaging has the advantages that it does not expose the subject to ionizing radiation, provides compositional information unavailable from radiation-based techniques, and is readily performed in three dimensions, affording a unique and rich anatomical perspective.

S-64

Receptor Modeling in P E T / S P E C T / M R I

315

318

I M P O R T A N T CONSIDERATIONS F O R M O D E L I N G O F I N P U T FUNCTIONS FOR M U L T I P L E INJECI'ION S T U D I E S O F RECEPTORS W I T H PET A.A. Bonab, ~ , H. Hsu, B.T. Christian and N.M. Alpert. Radiology, M~sachusctts General Hospital, Boston, Mass. 02114 Dynamic Positron Emission Tomography is used to assay the in vivo concentration, B ' ~ u ~, o f various receptor molecules in tissue..B..e~ase. Rceptors interact with Spr rad_ioligands v i a bimolr saturable binding, the models

A KINETIC MODEL FOR MEASUREMENT OF RECEPTOR CONCENTRATION VIA MAGNETIC RESONANCE IMAGING OF A PARAMAGNETIC CONTRAST AGENT D. 1L Vera, M. H. Bunnocore Department of Radiology; University of Califomin, Davis, Medical Center, Sacramento CA

used to describedynamicdata are necessarilynonlinear.One experimentalstrategy has been to collect PET data during multiple injectionsof radlolabeledtho0 and unlabeled (cold) llgand, where the mass o f injected llgand may exceed trace amounts. DeIforge has proposed a modeling scheme for such experiments which assigns separate state variables to both species. This app.m~.h also requires two input functions - one for hot figand, the other for cold ligand m plasma. I n many cases the hot plasma function must also be corrected for the time-varying percentage o f label which is associated with metabulites. Because cold ligand cannot be assayed directly, its concentration function must be infened from the metabolke-corrected hot function and accurate measures o f specific activity of the injectates. A cursory sensitivity analysis might conclude that errors in constructing the cold pla.~na function would affect primarily the blond-to-brain transfer parameter, K1. However, we find that uncertainty in measures o f plasma concentration or meta'bolite fraction lead to surprisingly large errors in estimates o f B ' _ o . . This is likely due to the heavy dependence o f the concentration o f availa~Te"receptor sites on injection o f large doses o f cold ligand. We explore the impact on the cold input function and on parameter estimates o f common limitations in blood a-ssaying techniques during multiple injection studies. Finally, we speculate on the conditions under which the cold input can be known accurately enough to allow reliable estimates o f B'max, from PET data.

We propose a "kineticmodel for the regional measurement of receptor concentration JR],: via magnetie resonance MR imaging of a paramagnetiu contrast ageat Gd.DTPA-gal.CP. The model uses five compartments: extra-hepatic plasma, total hepatic plasma, hepatic pissma within a user-defined region-of-interest (ROI). receptor-agent complex within the entire liver, end receptor-agent complex within the ROt. Observational equations, represented by fp and f, are functions of proton density N(H), tissue 71 and 72 *, TE and TR, the relaxation rate constants k. as well as the gadolinium concentration. Both fp and f~relate [L]~ [L]~,and [C] to MR signal intensities/It, Y=,or Y3via an equation defined by a FSPGR pulse sequence. Because formation of ligand-receptor complex increases the effective molecular weight and deereeses hydration of the agent, we simulated the model with f, < fr The output was consistant with }'2data in rats. L~andF.lani-Helxl~ Hepatic R~eplor ~ Plasma Complex ~~ ~ ILl. ~ [R],k,, ~ [(71. I Liver Od-D:I'PA~JCp[ [L'I, ~ ~ RO L

Vt T:

Enfire to/l v,,,+I Uver v ' ~ Y3

L, Amount of agent injected [L}, Agent in extrahepatic plasma [L]ht Agent in hepatic plasma ROI [L]Ja Agent in hepatic plasma [Rh F, V~ k~ Y~

Receptor within entire liver Hepatic plasma flow Compartmentplasmavol. Forwardbinding rate const. MR signal intensity

316

319

QUANTIFICATION OF NEUROTRANSMITTER RELEASE WITH PET AND CONSTANT INFUSION Richard E. Carson PET Dcparmaont, National Insamtes of Health

QUANT1TATION OF NEURORECEPTORS WITH SPECT Man: Laruelle, MD Department of Psychiatry, Yale School of Medicine and West Haven VA Medical Center

PET and SPECT imaging stodins with receptor-binding radiopharmaceuticals have been used to quantify regional receptor concentration in human subjects. These studies typically erapiny bolus inj~ioas, dynamic scanning and blood sampling, and kinetic modeling or graphical analysis, For high sl~itic activity UaC,r the restdling estirant~ of volume of diswibolion am linearly related to free receptor o0ncentration. Thus, changes in endogenous ligand levels produced by physiologicalor pharmacologicalinterventionscan be measuredby their effects on radiotrac~ binding levels. F~ bolusstudies,two water adminislzations(control and activation)are required. Recently, bolus plus continuous infusion (B/I) has been employed to permit mes.v.uemerits at true equilibrium. This approach allows direct quantification of V ftorn the dssen:plssma radioeacer ratio during equilibrium. This B/I methodology can be used to dynamically measure receptor occupancy during changes in endogenons ligaod concenlratlon in a single study. This approach is demonstrated using [(2-I 1]-raclopride and PET in rhesus monkeys and human subjects to measure changes in endogenous dopamine following i.v. adminlsuatin, of amphetamine 40 min arias the onset of tracer administration. Comparison of V valana between and post.amphctamir~ equilibrium pexiods Inovido an initial measure of dopamine release. Ill Bddidon, the shape of the displaceraemcurve reflects the kinetics of synapilcdopaminr release and raclopridr disassociation from the:receptor. Combined PET and/n v/vo microdialysis s~mdieshave been performed to characterize this re.sponse in m~'~ detail. Using simulafon studies and simub taaenus analysis of the [C-11]-racloprido and dopamine concentration data, we can address the following questions: I) what characteristics of the dopaminr relensr curve can be estimated from raclopdde displacement data, 2) what u'acr adminisuadon scheme (e.g. bolus or B/l) lxovides the maximum seas/dvity to endogenous ligaod release, and 3) what are the desirable kinetic charactedstics of radioligands that will optimize the m ~ e n t of neurotransmittor release.

317 DISTRIBUTED NONLINEAR MODELING O F PET ADRENOCEPTOR LIGAND K. Kroll, Z. Li, R. B. King. J. C. Caldwell. K. A. Krohn. C. Rhodesand J. B. Bassingthwaighte. Bioengineering and Depts. of Radiology and Cardiology, Universily of Washington & VA Medical Center, Sealtle, WA, & Hammersmith Hospital, London, UK, To obtain increased information from PET receptor ligands, a physically realistic distributed model was developed describing vascular flow, exchange between capillary and interstitial fiuk[, reversible nonequilibrium saturable binding of ligand to an extracellular receptor on parenchymal cells, and spillover of a fraction of blood pool activity into tissue regions. Because the model describes competition for binding between tracer and nontracer ligand, it is suitable for protocols using multiple injections with different specific activities. Myocardial PET image data from the Hammersmith Hospital using two injections of the hydrophilie 13-adrenergie [igand [11C]-CGP-12177in normal human subjects were used to lest the model. Left vemricular blood pool activity and dose specific activities were used to define the model input. Parameter optimization was used to fit the data for both injections jointly, using normal values for flow and capillary permeability. Receptor density was estimated to be 4.4 • 0.4 pmol/g (mean + SD, n = 11 tissue regions), and receptor dissociation constant, KD (equal to ratio of dissociation to association rate constants) was 1.3 • 0. l aM. although individual association and dissociation rate constants were poorly defined. Literature values for receptor density and KD are 4-17 pmol/g and 0.3 nM, respectively. [l was not possible to obtain reliable parameter estimates by fitting either the high or the low specific aclivity injection data alone. These pilot data suggest that myocardial ILadrenoceptor density and dissociation constant can be determined by fitting high resolution PET data using a mechanistic model. (Supported by NIH grants RR01243 and HL50239)

Over the last few years, the availability of high affinity tracers labeled with 123-iodine and of high-resolution brain dedicated SPECT cameras has allowed to use SPECT to visualize and quantify neuroreceptors. .This presentation will address some methodological issues regarding in vlvo neuroreceptor quantitation with SPECT. Kinetic and equilibrium paradigms will first be described within the framework of the MiehaelisMenten equilibrium theory. Various possible outcome measures will be described in terms of volumes of distribution. Several interpretations of the term "equilibrium" will be discussed. The bolus plus constant infusion paradigm, particularly suited for SPECT imaging, will be presented and its advantages and limitations will be discussed. "Pseudo-equilibrium" methods will be critically reviewed. The impact of limitations of SPECT quantitation on the derivation of receptor parameters will be discussed. Finally, strategies to develop and evaluate simplified protocols that can be implemented in clinical routine will be presented. These concepts will be illustrated with studies measuring benzodiazepine receptors using [t23I]iomazenil, dopamine Dz receptors using [123I]IBF, and monoamine transporters using [123I]13-CIT.

Advances in 3D PET and SPECT Image

S-65

320

323

ADVANCESIN3DPET AND SPECTIMAGERECONSTRUCTION JR Votaw, SJ Cullom Department of Radiohigy, Emory University, Atlanta. GA 30322 Recent advances in computing power and algorithmic design have enabled medical image reconstruction to become much more sophisticated. ComPutational tasks that were prohibitive in the recent years are now capable of routine clinical implementation. For example, until recently SPECT images were reconsmmted solely using filtered backprojection (FBP) without attenuation or resolution correction. PET imaging used the same reconstruction algorithm with a measured attenuation correction applied post-reconstrustion. The choice of this algorithm was based on its conceptual simplicity and speed of implementation. Now, with SPECT, more sophisticated algorithms which ntilize information describing the nonhomogencous attenuation map and depth-dependent resolution of the collimated detector system is possible. Dedicated gantries, source hardware, detector design and software now allow simultaneous emission/transmission imaging for SPECT attenuation correction, and are available commercially for clinical use with I D A approval. Nonstatinoary resolution compensation is less developed than attenuation correction with SPECT where both iterative and noniterative comprehensive approaches are proposed. With PET, reconstruction from incomplete three-dimensional datasets provides significantly improved statistics for dynamic, functional, and count poor studies such as in neuroreceptor studies. The use of segmented attenuation maps where attenuation coefficients are assigned to regions of similar tissue composition offers a practical solution to the problem of prolonged imaging time and/or poor statistics of PET transmission imaging and avoids artifacts associated with measured attenuation correction. It has also been shown to provide similar improvements for SPECT. These developments serve to ilhistrate some of the common problems and solutions between these two imaging medalities. In this session, examples of these advances will be presented.

THE A~AFFIVE SEGMENTED ATTENUATION CORRECTION - - MERFF$ AND LIMITS MtngXu~MatiincktodtI~sdtuteof Radiology.SI. Louis.Me Adaptive segmented attenuation COtTection(adaptive SAC) is a non-supervision, ~gmanted attenuation correction (SAC) method for PET whole-body imaging. Adaptive SAC extracts tiss~ta boundary information from low-count (400K transmission counts per slice) nansraission images using the local threshnldiag segmentation (LTS) technique; new body attenuation maps ale formed by filling in each segment desired value of attenuation coefficient. Compared to the emission images using the conventional measured method, images corrected using the adaptive SAC are quantitatively equivalent but smoother. Clinical evaloation on ~ whole body studies further iadicales that dwJe is very tittle difference during the image reading and lesion detection (2/56 disagreement-rate), lmplemealed on a multipmcessor SPARCStation 10MP. the method is so fast that it is possible for on-line cfialeal data processing. Currendy. the adaptive SAC becomes a vital procedure for PET whole-body onr imaging at the Mallinekrodt Institute of Radiology (St. Louis, Me). Using the adaptive SAC, a typical four-bed position (60 cm axial-,r'ov, usually covers torso from neck to pelvis) whole body FDG soldy (10 mCi administrated) on ECAT EXACT scanner can now be finished within one hour (2-man. TR + 10-man. EM per bed) with image quality equivalent to or surpassing those using the conventional method (10-man. TR + 15-man. EM per bed). Although the adaptive SAC was originally designed for PET, its concept shmdd apply to SPECT as well. The general technical challenge comes from the limitations such as input image noise, intermediate coefficient values, and unsupervised quality assurance. Additional challenge is to seal'ch for the most robust technique for clinical implementation with minimum sacrifice u perfoemat~. Nevertheless, possibitities of using the adaptive SAC for quantitative studies ere also sought. Finally. the potential applications for 3-D pET, such &sattenuation correction and ~attcr correction, are currently under active investigation. In conclusion, we believe that the adaptive SAC provides a sound software solution m reduce neise for today's clinical PET and SPEC'r studies. As computing power becomes cheaper tad cheaper m obtain, more sophisticated algorithms can be adopted to deliver the ultimate image quality and accuracy with the maximum scanner throughput.

321

324

NON-1TERATIVE COMPENSATION FOR NON-UNIFORM ATTENUATION AND 3D DE'/'BC~fOR RESPONSE IN SPECT IMAGING. Stephen J. Glick, Michael A. King, and Tin-Su Pan University of Massachusetts Medical Center, Worcester, MA.

QUANTITATIVE S P E C T RECONSTRUCTION M E T H O D S B.M.W, Tsai 1+2,E.C. Frey .'2 and D.S. Lahish' lDepanment of Biomedical Engineering and 2Department of Radiology, The University of North Carolina at Chapel Hill

Accurate compensation for non-uniform photon atteaustion and the 3D distaace~epeadetu detector response can improve the diagnostic accuracy of SPECT imaging. One approach for perfonnlng these corrections is to use iterative reconstruction methods with attenuation and the 3D detector response modeled into the projector and backprojeetor pair. This approach has shown great promise, however, its' implementation requires a heavy computational load, thereby making it unsuitable for routine clinical use. In this paper, we present t non-italative approach for compensation of non-uniform attenuation and the 3D detecter response in reeonstrncting SPECT images. The method involves combining Bellini's attenuation compensation method, with 9 frequency distance principle (FDP) based restoration filter to account for the 3D distaaceMependent camera response. To evaluate the accuracy o f this approach, and to compare it to other methods, simulation using mathematical cardiac and brain phantoms were studied. Using these activity maps, simulated projections modeling attenuation and blurring were obtained using a ray-tracing algorithm. The simulated projections were then reconstructed using; 1) ramp-filtered hackprojection with no attenuation compensation, 2) the Bellinl/FDP filtering method using the true non-uniform attenuation map ned ramp-filtered hackprojection, and 3) 30 and 100 iterations of the maximum-likelihood expectation maximization algorithm modeling non-uniform attenuation and blurting. In summary, combining Bcllini's method with FDP based restoration filtering was observed to be a quantitatively accoram approach for compeusation of non-uni form attenuatinu and detector response blurring in gPEL'I', and can be impiemetued with a s~bsUmtislly reduced computation time as compared to iterJtive reconsmtction metlieds.

Projection data in single photon emission computed toraograpby (SPECT) are affected by a number of instrumentation, physical and patient factors. When conventional imaga reconstruction algorithms are directly applied to SPECT projection data without compensating for these effects, significant degradation in the quality and qnantitalive accatacy of the reconstructed images will occur. We have investigated SPECT recolWffuctian methods which provide quantitatively accurate reconstructed images with high quality. Snico most of the image degrading effects are spatially variant, analytical solutions of the raconstmction problems ~ e difficult to derive. The low detection efficiency of the colllmatar-detectoz ~,s'tem gives lise to high statistical noise in the reconstructed images, lterative reconstruction Mgorithm~ allow accurate compensation of the image degrading effects by modeling the inulging proco~ in the projoctor/baekprajector pair. They have been applied to compensate for the non-uniform attenuation in the chest region in cardiac SPECT sludies, the spatially variant collimatordetector response, and the scatter response which is a complex function of the s o u r ~ location in the patient. We have investigated iterative reconstruction algorithms that are based on the maximum likelihood (ML), weighted least squares (WLS) and maximum a posteriori (MAP) criteria and use the expectation mayanuzation (EM) and conjugate gradient (CG) algorithms+ Results from phantom and patient studies demonstrate the convergence rate and noise handling properties of the different algorithms. The quantitative reconstruction methods are found to provide accurate attenuation compensation. Accurate compensations for collin'tater-detoetor and scatter responses are achieved without concurrent inernase in noise as found in conventional compensation methods. Substantial reduction in computation time has been achieved thrnugh advancos in computer hardware, and efficient implementation of fast iterative algorithms. These advances have brought quantitative SPECT reconstruction methods closer to clinical use.

322 THE DEVELOPMENT AND APPLICATION OF 3D RECONSTRUCTION TECHNIQUES FOR PET~ DW Townsend and PE Kinahan Department of Radiology, University of Pittsburgh. Pittsburgh, PA Until recently, commercial multi-dng scanners were designed with fixed lead shields, or supra, positioned between the detector rings. These septa, while successfully limiting the level of scattered radiation reaching the detectors, also reduced the tree coincidence rate. The demand for PET scanners with increased sensitivity led to the investigation of the 3D, or volume, mode of operation in which the PET scanner is operated with the sopta retracted from the field-of-view. The recent availability of commercial multl-ring scanners with push-button retractable supra has been instrumental in promoting 3D PET as a feasible alternative to the standard, 2D, supra extended mode+ The major advantage of 3D as compared to 2D imaging is an increase in sensitivity, after subtraction of scatter, of approximately a factor of five. In addition to the availability of supra-retractable scanners, key factors in the establishment of 3D PET as a viable clinical research modality have been the development of fully 3D reconstruction algorithms, the achievement of reasonable 3D reconstruction times, the development of procedures for detector efficiency normalization, attenuation and scsttaf correction in 3D, and facilities to handle the factor of ten increase in data volume. This paper will summarize the extensive efforts that have, in the past few years, been devoted to addressing and resolving these issues. In particular, the software and hardware implementation of 3D reconstruction will be described, inciudlng procedures to correct for both attenumian and scatter. It is anticipated that this significant increase in sensitivity at low activity concentrations offered by 3D PET will have wlde-ranging implications in application~ such as cerebral activation studies in single subjects; doselimited studies panienlarly in children; ligand studies for which tracer uptake is intrinsically low; kinetic studies in which the dine course of a tracer is followed over many physical half-lives; and studies with low signal-to-noise, such as small tumor imaging in oncology.

S-66

Imaging Cognitive Neuroscience

325

327

REAL TIME ACTIVATION OF NEURAL NETWORKS DURING WORD ASSOCIATION Yalchin G. Abdullaev and Michael I. Posner University of Oregon

FUNCTIONAL ACTIVATION STUDIES OF COMPLEX AUDITORY PROCESSING. Robert J. Zatorre Montreal Neurological Institute, McGill University

We use high density recording from scalp electrodes in conjunction with positron emission tomography studies to view networks involved in generating frequent and infrequent uses of visually presented words. To isolate generators involved in word associations we subtract reading words aloud from generating a use for the word. When generating a frequent use on the first occasion, we find significant differences in frontal areas starting at 150 msec. The midline frontal area previously related to attention, leads the left lateral frontal area by 40 msec. The left temporal (Wemicke's) area shows increased activation by 650 msec. When subjects practice the same list these activations are reduced. After practice, subjects generate a new novel use for the already well practiced list. The odginal activations return and are joined by a right posterior area (mirror image to Wemicke's area). We discuss these activations in relation to the strategy of isolating generators from scalp electrical activity and their time course is compared with cognitive, lesion and cellular studies in similar tasks.

M u s i c and s p e e c h are p e r h a p s t h e m o s t interesting ways that human cognition makes use of sound. It seems likely that the cognitive processes involved in such complex mental operations would d e m a n d a c o r r e s p o n d i n g l y c o m p l e x set of neural computations. This t a l k will r e p o r t on positron e m i s s i o n t o m o g r a p h y (PET) s t u d i e s of functional activation in normal subjects, using the bolus water method with paired-image subtraction and MRI overlay for anatomical localization. A variety of tasks have been used to investigate speech perception, melodic processing, auditory working memory, and auditory imagery. The findings from these studies provide e v i d e n c e for the e x i s t e n c e of a set of h i g h l y specialized neural networks that include not only superior temporal auditory cortical regions, but a l s o f r o n t a l - l o b e areas, p o s t e r i o r cortex, and subcortical structures. F u n c t i o n a l i n t e r a c t i o n s among these regions are crucial for our ability to perceive, encode, and imagine speech and music.

326 USING THE BRAIN TO STUDY IMRI Benjamin Martin Harvard University, Departmentof Psychology Beth Israel Hospital, Departmentof Neurology To use functional Magnetic Resonance imaging (IMRI) techniques to inter taskdependent blood oxygenationchanges in the brain one must distinguish significant image intensity differences from those due to chance variation or extraneous influences. Conversely,it one can produce reliablechanges in local blood oxygenation, it is possible to measure differences in the sensitivityof fMRI techniques. I discuss methodsfOr choosing an appropriatethreshold for significance in fMRI Intensity maps (Martin eta/., 1995.) I then describe a method of computing the sensitivity (d') of an fMRI data set relative to an a pr/otf spatial localization hypothesis. I discuss an apprcation of this method to data collected using two fMRI techniques: BOLD and EPISTAR(Siowart et aL, In preparation.) References: Martin 8, Wel LJ, Thangaraj V, Warach S (1995) A StatisticalMethodtot Relating Stimulus Variation to Changes in IMRI Signal Intensity,2nd Annual Meeting ot the Cognitive NeuroscienceSociety, San Francisco. Stewart B, Martin B, Darby D, Schlaug G, Thangarej V, Benfield A, Warach S, Edelmen RR (in preparation)Comparing BOLD and EPISTAR Functional Brain Imaging Data Using Signal Detection Theory.

Poster Presentations 328

331

ULTRASONIC EVALUATION OF QUALITY ATI'R/B LITES IN LIVE BEEF ANIMALS USING B-MODE IMAGE PROCESSING AND PATTERN RECOGNITION Vixen Amin, Mereedes I~xluierdo, Doyle Wilson. Gene Rouse, and Run Roberts Centerlot"NondestructiveEvaluationand Departmentof AninudScience IowaStaleUniversity.~ , IowaS00Il

REGIONAL MYOCARDIAL OXYGEN CONSUMPTION USING 150-O AND PET Z. Li, T. Yipintsoi, J.M. Link*, K.A. Krohn'. J,H. Caldwell t. K. Kroil and J.B. Bassingthwaighte Center for Bioengineering, *Dept. of Radiology, and *Division of Cardiohigy, Univ. of Washington and VA Medical Center, Seattle, WA 98195

Ultrasonic evaluation of biological tissues has found its potential use in food animal and agriculture fields. For the beef industry, an objective evaluation of the meat quality is of high priority, This report presents the results of IM %-fat evaluation in live animals using B-mode images. Using a commercially available reel-time ultrasound system and a personal computer, images from Longissimos dorsi (ribeye) muscle across l l t h to 13th ribs of 720 live hulls and steers were digitally acquired over the period of four years. Within 7 days of scanning, animals were slaughtered and a slice of the ribeye from each carcass was collected for determining an actual IM %-fat using an n-hexane extraction procedure. These actual %-fat values had mean of 4,98, standard deviation of 2.12, and range from 1,10 to 14.68, Image processing techniques were used to calculate parameters to quantify the image texture patterns. These techniques included 2~iimeosional Fourier transform, second order statistics, and co-oscurronce and run-length based texture analysis. Selected parameters were used to develop regression and classification models to predict IM %-fat. With one regression model, the accuracy of prediction was very good, with root mean square error (RlvlSE) of 1.13 for the actual %-fat less than or equal to eight (low %-fat group) and 1.51 for the %-fat values more than eight (high %-fat group), Diseriminant analysis was used for classification of the two groups. In conclusion, a method is developed to evaluate an important beef carcass quality attribute, the IM %-fat, in live beef cattle from ultrasound Bmode image p~ameters and pattern recognition techniques. Applications of such an objective method of quality evaluation include sorting feedlot cattle, and making genetic improvements in breeding stock.

329

S-67

An anesthetized ventila:ed~ left thorecotomised dog was studied to test a new method for measuring regional myocardial oxygen consumption (MVO2) using PET and :'SO-oxygen. The goal was to determine the optimal route to introduce 150-oxygen, and the minimal ROI size to estimate MVO 2 with models. 1SO-water was injected into the left atrium (LA) for estimation of regional myocardial blood flows. 150-oxygen was introduced via one of 2 routes: pre-equilibrated in blood (hematocrit 0,4) injected into the LA (1 study), or inhaled via the ventilator (3 studies}. Following each study, PET images were acquired in 35 planes every 2 seconds for I minute and then 10-15 seconds for 4 minutes, The data were reconstructed at 5.4 mm transaxial slices such that the LV consisted of 9 slices totaling 4000 voxels ( 13.2 mm3/voxel). Eight ROls were drawn from each slice as constrained by the *lCO sleady-state image. These 72 ROIs averaged 26.5 :t: 10.4 voxels or 0.35 +0,15 grams. The time activity curves from each ROI and combinations of adjacent ROIs were fitted to a nonlinear model for blood-tissue oxygen transport and metabolism (Li, 1995) to estimate MVO 2. Taking the whole LV as one ROI, MVO 2 ranged from 3.2 to 3.8 ttmol g'lmin-I for the four studies; and 150-waler flow estimated by Bergmann model (1989) was 1.22 ml g'lmin'l. It was found that when combined ROls were of 0.8 g size about half of these showed high coefliciem of variation between model fit and data due to noise. The noise level appeared to be related to the sample size as well as the location of the ROI. noisier over the lateral than o~er the anterior LV wall. The sites with :he good or poor fit appeared stable independent of the method of tracer introduction, suggesting that ISO-oxygen by inhalation can be used clinically for measuring MVO 2. With PET and modeling, myocardial oxygen consumption can be evalualed in regions greater than 0.8 grams. Supported by Nltt grants HL50238. 50239 and RRI243.

332

A SEMI-AUTOMATIC METHOD TO DETERMINE MICROVASCI.ILAR GEOMETRV IN IN VIVO IN SITI.I EXPERIMENTS F. Boschetti, B. Crisaful|i, D.Veaturoli, and G. Miserocchi Istituto di Fisiologia Umuna, Oniversi~ degli Stodi, Via Mangiagalli 32, 1-20133 Milano, Italy.

SEGMENTATION OF MULTIPLANAR IMAGES USING BAYES DECISION THEORY K, Muellerl, E.E. Herderich 2, J.F, Cornhill 1,2,3,and The Pathobinlogieal Determinants of Atherosctezesis in Youth Research Group ('PDAY) tLal~ratory of Vascular Diseas~ 2Biomedical Engianermg Center, The Ohio State University 3~ent of Biomedical Engineering, Research Institute, The Cleveland Clinics Foundatlun

Our aim was to determine microvascuiar diameter and perivnscular interstitlum thickness at the lung surface in in situ in vivo lung. Microscopic images of the lung surface collected through a "pleural window" by a video camera were digitized with a monochrome framegrabber (512x512 pixels, 8 bits/pixel) to be Computer analyzed by image prueesalng techniques. We have sludied the gray level dislribetioos of three objects, namely the microveszel lumen (V), file perivascular interstitium (I), and the lung tissue (L), and we have compared the statistics moments of adjacent objects. The V and I objects are characterized by significantly different means, while the I and L objects are significantly different in their standard deviations but not in their means. Dealing with the properties of mixture of differont distributions, we have proved that the edges between the V and I objects are identified by the maxima of the moving standard deviation, whereas the edges between the I and L objects present a maximum of the moving standard error. A computer program implementing a semi-automatic procedure has been developed which detects the edges of the three objects under study in a selected region of interest. The procedure automatically calculates all the maxima of the moving standard deviation and of the moving standard error, and displays all of them on the image, The user selects a line roughly following the maxima which lie on the edges, and then the procedure discards the maxima on the basis of their distance from the line. The microvessel diameter and interstitial thickness are then calculated and saved in an ASCII file for further processing. The developed technique may provide a useful tool to study lung fluid balance and microvascular reactivity in the in situ lung in the normal state aod in response to a variety of hinctional conditions.

Images are called multiplanar if the pixel intensity vector has a dimensionality greater than unity. Examples of muffiplanar images are color images, sets of MRI images emphasizing different relaxation characteristics, and registezed multi-modal images. Multiplanar images offer a high degree of feature resolution, however, due to the large intensity vector space, it is often not trivial to segment the image into its feaunea by means of unbiased automated methods. This paper presents a statistical method employing Bayes decision theory that reduces a multipianos image to a monopianar image with the pixel values representing the feav.n~ identification tags. The method requires the user to mark a small sample of each feamro class to be identified. Using the statistical characteristics of these samples (mean vector and varianco-covantanea matrix), ever/image pixel is then classified into the feature class to which it has the closest statistical distance. Spike noise in the resulting image is eliminated by subsequent median filtering. The algorithm was tested on a set of 20 randomly selected color images of stained, human cross-scctlunal comnap/arteries to distinguish lumen, and the three tissue layers, in:into, media and adventitia. The sizes of media and intlm& inlpra~ant factors in the study of the development of athm'osclemais, were identifiedby the algorithm with high accuracy. A correlationcoelficinntof 0.98 was obtained when the classificationresultswine compared to manual segmentations. The program is contained in a versatile X-Windows interface and was utilized for various other quantitative studies, such as the measurement of extra- and intracellular lipid in cross-scctiunal arteries. The method is superior to histogram based duesbolding techniques since it accounts for the ststistical variance of the image features. It is easily extended to segment 3D volumes, and the computational independency of the pixels allows an efficient implementation on parallel processors.

330

333

MOTION OF THE RIGHT VENTRICULAR FREE WALL IN PULMONARy HYPERTENSION SS Klein, CH Lomus. and JM Bundy Departments of Biomedical Engineering and Radiology;, Vandmbilt Univ, Nashville, "IN

COMPARISON OF IMAGING MODALITIES FOR EVALUATING STRUCTURAL INTEGRITY OF THE ACETABULUM FOR PROSTHETIC DESIGN T.Sarge, D.Hauser. D,Magid, I.Wenz, W.Scon" L.Rileyo C.Resnik, M,DeHart, A.Brooker. N.Spengier, E,Chao JHML Baltimore, MD, USA

Quantifying the changea from normal in right vuntricolar free wall (RVFW) motion in pulmonary hypertension may be useful in monitoring patient progress. Therefore, tagged eine MP,I, a noninvnsive method of tracking tissue motion, was used to track RVFW motion throughout systole. Measuroments were made on 9 normal (NL) volunteers and 6 pulmousty hy~rtensiun patients (3 pn.--singie lung transplant fiRE), 3 post-transplant (POST') within one year of transplantatiun) on mid-ventricuiar short.-axis slieos. Percent segmental shortening (PSS) was defined as the change of distance between two sequential tag-RVFW in~rzectious at a given time in systole divided by the distance between theae intersections at end-diastole. PSS in PRE was difl'etent than NL (pr NL PSS increased monotonically peaking at end-sysinle with a maximal PSS of 17:t:15%. PRE PSS decreased to -1.0~-5.3% PSS at 25% of systole and then increased to end-systole with a maximal PSS of 6.6~:9.2"/e. POST PSS increased monotonically peaking at end-systole with a ~ PSS of 7.8.~9.2% which was not significantly different than NL. The motion patterns found in the PRE, POST, and NL RVFWs were all significantly different (p
Purpose: Accurate pro-operative assessment of acetabular bone deficiencies is key in detennlniag the method of repair for surgical reconstruction of the hip, especially with the increasing number of custom fit aeetabolar components now being produced for patients with bulk defects. Plain films and computed tomograpby are the standard by which orthopedic surgeons pre-operatively plan for anetsbulas reconstruction. We investigated the consistency and accuracy in observer analysis of acetabular bony defects. Materials and Methods: Twenty-one human cadaverie pelves were reamed, press-fit with a hip prosthesis by an orthopedic surgeon, radiographod and scanned. Fourteen lest cases were lesioned to simulate 2 sizes of 7 defects as classified by the AAOS classification scheme: segmental- and cavitazy- superior, anterior, posterior; and medial wall perforation. The data was independently read as plain radiographs and 3D MPR computer images by 3 radiologists, 3 surgeons, and 1 biomechanical engineer. Statistical analyses to assess sensitivity and specificity were performed. Results: Average diagnostic accuracy of the radiologists, surgeons, and engineer was 20%, 18%, 17% for plain radiography, and 21%, 17%, 17% for CT. Majority reader agreement on defect location occurred for only 57% of plain radiography cases and 52% of CT cases (iaterobserver). Comparing plain radiography to CT diagnoses for each reader, the average reader reported the same diagnosis on only 29% (SDffi9%) of the cases (intranbserver). Conclusion: Detennlning the extent and location of bone loss in the peri-acetabular region is highly observer dependent for radiologists and surgeons, and not reliable when using either plain film or CT data for pre-operative planning, Design engineers of ecetabuiar prosthetlcs should be aware of this clinical diagnosis limitation for acetsbular bone defects.

S-68

Non-compartmental Models of Metabolism

334

337 A SL~VEY OF NON..O:)MPARTMENTAL MODELS tN METABOLISM IC K Nocwich Institute of Biomedkal Engineering, Univerdty of Toronto, Toronto, Ontario,

The mmportment is a convonient mathematical idealization of physical reality - a stmctam wherein sdded mainri~ are inmntanr and uniformly ~ C o m p a n m e ~ are easi~, vismli~d and ~h~ ~ to sdutions ia sua~ of c ~ w~ich a ~ ~ tu h=~dl~ However, often t i n ~ am dmplex or morn a c c u r ~ methods for analyzing ~ c data. (i) For ovm"forty years it hns la~n recognized that whole4xxly clnsrancr ctuves of cemin minera~ ~tns and d m ~ are more accmately desm~o~ by power functions of tbe format ~ or ~amma den~ty fun~ens of the fomt A e ~ t ~ , whJch hsve ~ r ~ to many modr ~ c h ~ random walk and diffusion models. (u') Radioisotopic rotors are present in such quanthiea that both thelr transport and cheadnsl tranfformafion am linear ~ Heno~ linear systems theory caa be used to model inpot-OUtlmt relations (rates of ~ and disaRxarance) by ~ tedmique~ There is no rcquirement for a compartmental s t ~ m r ~ (iii) Circulatory models of whole4x~dymetabolic proc~r have been proposed. In these models, inoperties of circulating blood and specific seriesand parallel configoratiom of organs replace the idealization of the perfect c o n ~ (iv) D ~ t e d models, or convc~ondlffusion-reactJon models attempt to represent with some ~ l i t o d e the physical processes by which metabelit= are produced, transported, and finally biochemically degraded_ Its mathematical vehicle is the wctor calcolns. (v) Stati~cal noncompartmental models pen'nh the cal~aintion o f ~ n quantities such as mean residen~ time and volume of dig,'button, without the use of model paran'te~ (cf non-parameUic vs parametric statistics).

C I R C U L A T O R Y M O D E L S F O R T H E STUDY O F W i t O L E - B O D Y A PHYSIOLOGICAL PERSPECTIVE

KINETICS:

Andrea Marl CNR Institute of Systems Science and BiomedicalEngineering. Padova, Italy Circulatory models represent whole-body (WB) kinetics as the kinetics of the single inletsingle outlet organ encompassing all the body tissues ideally obtained by opening the circulatory loop e.g. at the level of the left heart. In the linear case (e.g. with tracer experiments), the body tissues are characterized by an impulse response (at open circulatory loop), i.e. the artery-vein single-pass impulse response (SPIR). The SPIR can be obtained from the observed WB response, and carries atl information contained in WB kinetic experiments. WB kinetic analysis w~h circuiatory models a i r ~ to interpret the SPIR and to derive physiological parameters from it. Zierlefs organ kinetic methods, appropriately extended, can be used for this purpose and are model-independent. A complete characterization of WB kinetics would then require the artery-uptake, production-vein and production-uptake SPIRs, that are not experimentally observable and not simply related to the artery-vein SPIR. Therefore, the possibility to obtain reliable estimates of quanlities such as production, uptake, WB mass, and exchange rates depends on what information is available on non-observable SPIRs. When sufficient information is available, with circulatory models approximate solutions can be developed that have a physiological basis. This advantage is not shared by the more traditional approaches, such as compartmental analysis. Compartmental analysis hides in apparently sound strategies, such as the placement of losses in peripheral compartments according to a tissuecompartment correspondence, assumptions on the non-observable SPIRs that may be totally unphysiologicaL Consequently, compartmental models are prone to misrepresent WB kinetics, and are unreliable for quantifying WB physiological parameters. An example is given by the analysis of the influence of insulin on glucose distribution space: the present theory shows that the apparent discrepancy between experimental results is in reality due to an unrecognized defect of compartmental analysis.

335

338

CORONARY BLOOD FLOW AND ESTIMATES OF INTERSTITIAL ADENOSINE

TRA~SCAPIIJ..ARy EXCHANGE OF GLUCOSE IN THE HUMAN FOREARM: MEASUREMENT BY A DUAL TRACER EXI~[~MEWr ~ DISTRIBUTED MODELING, p9 VicinL Riccardo Bon~hama (^), Mikko Lehtovitta (*), I.~f Group (#) mid Cl9 Cobelil. Deparnnont of Elecu'onir and laformtgJcs. University of PKlova, Italy. (^) University of Verona School of Medicine, Italy. (*) University of Helsinki. I~mland.(#) Univct,sity of Land, Sweden. The tranwatpill~ e=chaage ('rE) of glucose ((3) fi'om pinsma (P) to the interslitium ('ISF) is 9 amcinl kinetic event of G mc~bolism, but it cannot be measured directly ia vivo, Hcxc, fr the first lime. we me,tem~ G lie ia human muscle from 9 dual U'acexexpefimont and disnlbmed modeling. We applied 9 dittalboted modelof TE to tracer data obtained in the fop.arm skdet9 r~sr of tlx healthy individuals, An imtavur162 (indocyanine green. ICG) and aft r (3H-D-manahol) tracer were rapidly in~taed iuframterially. Their wa~oet curves were sampled in 9 deep low.arm vein by an tatom~tin t~t.~pEng devios crecy 0,9 t,er for 90 ~cond*. The ~xp4~ment ~ tq>eated ia eatdt individual ia the baud and inmlinemic state (420 pM in two mbjec~ and 2000 pM in fut~ mb~ts), uting the G clamp technique, P i n ~ t flow Fn was direcdy measured tad resuRed 2 4 . ~ . 5 0 ml mln~ kg"1 in the bend stata and 25.56"~4,38 ml mln"q kg"1 in the iumllnin~l state, A 20*pathway dis~rthuted model was ut~d to . ~ y z c the data. A P capilinsy volume Vp of 3.5 ml kg"1 and 9 flow heterogeneity lagged not,real dlsmbefion of random dispersion 0.55 and skewness 0.3 were assumed. The inlet functina to the capillaries was obtained by decouvoludon f~om the ICG data and the system Ixansfer function, The ISF distribution vohLme Via( was fLxed in each subject due to its low sensitivity in the 90 u~conds window. The peTm~tbility-mr[9162162 lx~dttct of the ondosheHal gap of the capillary wti] PS2 was e~t~ted in individual with good lxeclsion by fi~ing the 3H-D-manmtoI dilution curve. '~he model wts 9 to =afisftctor fit both indinotr162 dilution curves simultanr Average vtlue6 for PS were 21.87• ml m~n"1 kg-1 in the basedstate and 17-27+-3.64ml mln"1 kg"1 in the inzul[alzed t~ate.~ g was ea~mtad with vesy good precision in each individual (%CV from 3 to 10). PSg q~pears unchanged during liype~nsulinemb~ suggesting that 'rE occumng via ~mple dLffution h not i~flue~ced by insulin. PSg was also very well eoxreJated with Fp (r=0.83, p<0.001), this suggesting thtt high p flow regions are also ch&acte~zed by efficient TIE.In conclusion. TE of G can be measmed in human skeletal muscle by 9 dual tracer experiment and di~lxlbuted medeAing.Our results =how that PSg is not influoneed by insulin and that there is no "luxury pertotion" in the foxearm.

David W, Stepp, Richard Van Ribber, Keith Kroll, and Eric O. Feigl Depts, of Physiol. & Biophys. and Bioengineering, U, of Washingten, Seattle WA Adenosine has long been hypothesized to be an important physiologic mediator of coronary blood flow, This hypothesis has yet to be critically tested, because no accurate measure of adenosine in the interstitinm, the critical compartment, is available, The purpose of the present study was to develop a mathematical model to estimate interstitial adenosine using measurements of coronary blood flow, arterial and venous adenosine concentrations, and parameters of endothelial cell transport and metabolism of adenosine. Studies were performed on the coronary circulation of closed--chest, anesthetized dogs. Multiple indicator dilution studies were done and parameter values of para-r transport (PSg) and endothelial eallular transport (PSecl) and metabolism of adenosine (Go:) were derived ~rom the tracer kinetics, The model was validated in separate experiments using inhibitors of transport (dlpyridamole) and metabolism (iodotubercidin), The predictive ability of the model was tested by comparing venous adenosine measurements with model predictions. In additional experiments, the coronary flow dose-response relation for interstitial adenosine was determined during graded intracoronary adenosine infusion, The dose-response curve was steep, having a Hill coefficient of 4-6 and an ED5o of 233nM. Under control conditions, the interstitial adenosine concentration is normally at a subthreshold concentration of 100riM. It is concluded that a distributed model can accurately estimate interstitial adenosine concentration and that there is an extremely steep relation between interstitial adenosine and coronary flOW.

336 CELL SYSTEMS WITHIN ORGANS: DISTRIBUTED-IN-SPACE AND NONLINEAR. SYSTEMS ARE NOT READILY COMPARTMENTALIZED James B. Bassingthwaighte, Keith Kroll, Zbeng Li, Gary M. Raymond Univ. of Washington, Seattle., WA 98195-7962. Whole organ systems are modeled best for analyzing tracer kinetic experiments by accounting for all the observations simultaneously. The "dala" on a given experiment should be regarded as what you "know" about the system and its behavior. These "data" usually include a representation of the anatomy, the arrangemenls of the structures of the tissue and the list of nrembranes, transporters, enzymes, and cell types, arranged suitably. One can not, for example, make observations on interstitial volume. V,,0 in every tissue region in every experiment, but must use data from previous studies and constrain the relevant parameters to realistic ranges. When the parameters the experiment is designed to elicit are little influenced by I~st. then one fixes it so that the degrees of freedom in the fitting procedure are reduced. By bringing in "data" from other studies, the numbers of unknown parameters are reduced, one hopes to few enough that the model tiring identifies their values. Compartmental models, systems of first-order mixing chambers, are commonly used for parameterization of data because they have an overly reduced number of parameters. It is usual that tbe~ reduced model forms have the same free parameters as the more realistic model forms, except that the assumptions u ~ d for the reduction are overly compromising. An axially-distributed blood-tissue exchat)ge model agowing concentrefion gradients from artery to,vein and having a heterogeneity of flows along parallel paths has the same number and type of parameters for the exchange and reaction events, for example a permeation rate, a reaction rate, a specific dissociation constant on a binding site. Compartmental models do not have fewer free parameters, they just have more incorrect assumptions, e.g., the stirred tank capillary assumes a sharp discontinuity between the concentration in the artery and that in the capillary blood rather than a continuous function of position. The incorrect assumptions tend to produce biases in parameter estimates, and so should be avoided. Model nature, not icons? Supported by NIH grants HL50238. HLIg139 and RR1243.

Modeling Metabolism & Its Control

S-69

339

341

MULTILEVEL CONTROL OF HORMONE ACTION J. J. DiStefano, 1i"I,N. Brown and M. Patterson Biocybemetics l.,aboratory, Engineering and Medicine, UCLA, Los Angeles

ESTIMATION OF ORGAN SECRETION~I3ROOUCTION RATES FROM TIME-SERIES CONCENTRATIONDATA GiovanniSparecine,Giuseppo De Nicoteo,',,and Claud~oCobefli Dip.to di Elettrenica,Universlta' hi Padova,.'~Dip.todi Intormatiue,Universita'di Pavia, italy The control at the organ level of hormone sesre~on/substrataproduction rates is a key regulator/ mechanism in physiologicalsystems. Their assessmentis thus crucial but difficult since they can not be measured directly. By viewing them as system inputs their indirect assessmentcan be posed as an input estimation problem from the measured plasma concenb'afons. If the honnone/subsb'atewhole-body kinetics are linear, such a problem can be attacked by deconvolution if the system impulse response is known. Mathemaficafly,deconvolutionis known to be Ul-cond,'tioned.Dealing with physiologicalsystems adds to the complexity of the problem since signals are often thfrequsntly/nonuniformlysampled. To handle fll.conditioning, a leading approach is Phillips-l'ikhonev reguisdzation based on a trade-off between fidelity to data and adherenceto a priod expectationson the input, e.g. smoothness,tuned by the so-caltedregutadzationparameter. However,this methodleaves some issues open. Rrst, it provides ~ly disorote-time input estimates, e.g. staircase secretion rates are often a poor approximebon, especiallywhen scrupling is infrequent. Second,the reguladzationparameteris usuallychosenby criteria develsped within a deterministic embedding which, lacking a statistical characterizationof bias error, can not provide confidenceintervals, an essential quantity, e.g. to choose the threshold for pulse detection. Here we present an approach which deals with the above issues. The infrequentsampling is addressed by decoupling the input continuous-time from the output data discrete.time domain. This enables the estimation of input profiles continuous up to any desired number of derivatives. To tackle ~e regularization parameter/confidenceintervals issue, deconvolution has been stated into a stochestic context where it is possible to derive new Maximum Likelihood criteria and closed-form expressionsfor confidence intervalsaccountingfor bias. " ~ presentedapproach has been applied to the reconsttuolion of various hormone(Insulin, LH, GH) secretionrates with different release patterns (pulsaiile, esciflato~) in variousoondifions(spontaneous,after stimulus), and aLsoextendedto the case of time-varyingsystems ( r e ~ of hepatic glucose production).

Most hormones are secreted by glandular source organs into the bloodstream, which assists in their transfer to target tissue sites, where they may exert their action, get metabolized or otherwise disposed of, or both. Some (e.g., thyroid hormone) are also produced at multiple tissue sites. Control of these processes is manifested at multiple levels, the most well-established and classical being central control via feedback action at the pituitary, and also via the hypothalamus, a primary component for neuroendocrine communication. A dominant and still poorly understood characteristic of pituitary hormone secretion is that it is pulsatile, and possibly chaotic. Carrier proteins in blood, tissue receptors and transporter proteins, and most notably local cellular mechanisms -generally different in different organs, also participate in local hormonal control An increasingly large part of the recent endocrine literature has been devoted to local cellular mechanisms, both enzymatic and molecular, controlling hormone action, with modeling being primarily qualitative. In addition to experimental explorations of local mechanisms, we have begun the process of mathematizing and simulating these mechanistic theories, focusing on an integrated model of the hierarchical control of production, organ distribution and metabolism, action and excretion of thyroid hormone. We present some of these simulations, and some data, with emphasis on understanding the diversity of coupled mechanisms, responses and actions that appear to serve whole organism homeodynamics well.

340

342

TOPOLOGIES FOR IDENTIFYING LOCAL ORGAN HORMONE CONTROL MECHANISMS. T. Nguyen, S. Hakimian and J. J. DiStefano, Ill Biocybemetics Laboratory, Engineering and Medicine, UCLA, Los Angeles

CONTROL OF GLUCOSEMULTI-STEPDYNAMICSBY INSULININ HUMAN MUSCLE Claudio Cobelli, MariaPISSeccomaniand Ricoa,'do8onedor~a(^), Department of Elestror~csand Intormatics,Universityol Padova, Padova, Italy. (') Division of Metabolic Diseases,Universityof VeronaSchoolof Medicine,Verona,Italy Glucose metabolism in skeletal musole is a complex mutii-step process. At the cellular level glucose crosses the membrane by a faciiitatad diffusion process mediated by specific glucose transporters. Inside the celt glucosecan either be transported back outside the cell or irreversibly phosphorylatedto gtucase-6-phosphateby hexokinase.Thus understandingglucose metabotismand its Controlby insulin in muscle requires assessmentof beth the bidirectional glucose transportacross the membraneand the rata of intrecel]ularglucose phosphorylaifon.Howeverquantitaflonin vivo of these processesis a d~fficulttask in the intact organism, since they are net accessible to direct measuremenLAn indirect model-based method is needed.We frst review available techniques,in particular the PET method based on the model of ~lF.deoxyglucosekinetics.We then present a novel in vivo methodto assessglucose metabolismin human forearmbased on a Viple Vecer(~2C-O-mannitol, not transportable; 3-0-v6C-reetbyt.D-glucose, transportable but not metabolizable; 3-~H-D-glucose, transportable and metabolizable) intraarterial pulse injection and on a multicompartmen~ model to analyze the tracer washoutconcentrationcurves measured in a deep vein of the forearm. The assumptionsof the tracer model, its s~'uC~e, a pridd and a postedod (nomedcal,whenidentified by nenlinear least squares) idenflfiability properties are presented. Having fixed the tracermodel,we then develop and solve the trasee motet. Estimatodparameters indudd glucose transmembrane transport fluxes, inb'aceflular phosphorylation, inlxacellular voluree and concentration.With tNs toolwe have assessedcontrolof glucosemulti-step dynamicsby tnsutinin skeletal muscle of normal man. Physiologic hyperinsulinemia increases glucose transmem~ane influx and intraceflular phosphor'/lation, but had no effect on transmembrane outflux. Intracellu{ar olstribation volumeof glucosewas increasedand intracoflularglucoseconce~aiten was decreasedby insuSn.We also studied non-lnsulin dependent diabetes mellitus suhiects: insulin control of glucose metabelism is defective both at the transmembranetransport and intrecefluisrphosphor'/lafon steps.

Sources for the thyroid hormone triiodothyronine (T3) are widely distributed, because T 3is produc.ed enzymatically from thyroxine (T4) in the cells of many organs. This has made quantification of local organ production rates especially difficult, and how much T 3 made from T 4 in any organ, in any species, is unknown. The problem, of course, is that simultaneous tracing of multiple, distributed sources is impractical, if not infeasible, model orno. Five- and 10-compartment, multiinput/multioutput transient response models and 4-, 8- and 10-peol steady state response models of whole-body T~and T 3 dynamics in blood and tissues are developed and analyzed. Decomposition methods and graph theory based cut set analysis are used to design complex but feasible multiinpuffmultioutput experiments to directly quantify local interconversion rates o f t 4 ~ T~ in liver, kidneys, brain and intestine, and any other organs with direct vascular accessibility. Monte Carlo simulations based on partial data pooled from earlier experiments indicate the potential utility and practicality of projected results.

S-70

Substrate Fluxes in Exercise

343 ASSESSMENT OF NON-STEADY C I R C U L A T O R Y MODELS Andrea Marl

345 SUBSTRATE

FLUXES

WITH

CNR Inslitute of Systems Science and Biomedical Engineering, Padova, Italy Steeis's menocompadmental model is the most widely used model for calculating subatrete fluxes in non-steady state, but has important limitations. Alternative approaches have thus been proposed based on experimental designs that maintain specific activity constant or on two-compartment models. These approaches are eppropriate for production but are still inadequate to calculate uptake. Compartmental models are unsoitable to calculate uptake because the strategy typically used in their development is prone to misrepresent the kinetics of uptake. Circulatory models offer a more rigorous approach, based on an extension of Zieder model-independent theory of organ kinetice. With circulatory models, the body tissues are characterized by an impulse response (at open circutatory loop), i.e. the artery-vein single-pass impulse response (SPIR), that can be determined in the non-steady state with some assumptions from the observed whole-body response. To calculate uptake and production, it would then be necessary to know also the artery-uptake, production-vein and production-uptake SPIRs, that are not directly determinable from the artery-vein SPIR. This difficulty intrinsic to the problem makes the determination of uptake and production in non-steady state subject to approximation. However, the theory of circulatory models formally shows how to calculate physiologically valuable approximations of the fluxes. Situations in which uptake and production occur simultaneously are difficult to solve. When uptake and production take place in distinct organs, as for glucose, in some cases fluxes can be predicted in a way that is free from the struclural assumptions of compartmental analysis. The advantage of this approach is evident with glucose clamp experiments, in which the assumption of equality between uptake and rate of appearance, implicit in Steele's model and other compartmental approaches, can be relaxed and is in fact proved to be unlikely.

344 REGULATION OF GLUCOSE FLUXES DURING EXERCISE David H. Wasserman, Dept. of Molecular Physiology & Biophysics, Vanderbilt U~vemity School of Medicine, Nashville,TN The goals of these studies were to determine the mechanisms of the exardseinduced increases in hepatic glucose production and muscle glucose utilization. Experiments were conducted in postabeorptive, exercising dogs in which hormone levels and neural outflow were controlled surgically or pharmacologically. Liver and ~ l a glucose metabolism were assessed using erterioveneus and isotopic techniques. "rhoea studies showed that the exercise-induced fall in insulin and rise in glucngon control the vast rnajority of the increment in glucose preduction dunng moderate exercise. The fall in insulin stimulates hepatic glycogenolysis while the rise in glucagon stimulates both hepatic glycogenofyais and gluconeogenesis. The effect of the fall in insulin is absent when glucagon is suppressed, suggesting that changes in insulin act by potentiating glucagon action. H was fudhar shown that exercise-induced changes in pancreatic hormones and, therefore, gkmose production ere highly sensitive to metabolic feedback control. Despite the importance of the pancreatic hormones in control of hepatic function during moderate exercise, the increases in muscle glucose uptake and oxidation are primary insulin-independent in healthy subjects. The nature of these insulin-independent mechanism(s) remains to be fully identifmd. They appear to work, however, by increasing the Vrnax for limb gluco6e uptake and oxidation without affecting the Km for these processes, in contrast to the impodance of insulin-independent processes in healthy individuals, in the insulin-doficlant diabetic, glucose utilization is impaired as metabolic abnon'mditias oppose normal mechanisms for glucose uptake. In summery, while the hepatic glu<:ocegulatoryresponse to moderate exercise is controlled by the endocrine system, the increase in glucose uptake is normally driven by med~nisrns that are primarily independent of the action of insulin and other hormones.

EXERCISE, LIPOLYSISo AND F A T I ' Y ACID OXIDATION Robert R. Wolfe and Labros Sidossis University of Texas Medical Branch The response of lipid metabolism in execclseis dependent on both the training stall mzd the exercise intensity. We have used tracer techniques to quantify peripheral and intramuscular lipolysis and oxidation at various exercise inteasitie~ in highly trained subjects. Peripheral lipolysis is stimulated maximally during moderate exercise (25% VOz max), whereas intramuscular lipolysisbecnsaes sctivc only in relative high-intensity exercise (85% VOz max). The rate of fat oxidation is greatest during moderate exercise . . . . . . ^ ,eater (65% VOz max), and decreases stgmficantly during high mtenstty, even ~o[tge~h energy requirements increase. In contrast, total carbohydrate utilization (plasma glucose plus intramuscular glycogen) increase as exercise intensity inc.rease~. Taken together, these results show that fatty acid metabolism is not primarily regulated during exercise, but rather responds secondarily to changes in mediators (cateebolamines and insulin) that respond in order to regulate other physiological functions.

Regulation of Energy Metabolism During Exercise: Cellular & Tissue 346 REGULATION OF CELLULAR ENERGY METABOLISM Maria Erecinska, Department of Pharmacology, University of Pennsylvania, School of Medicine, Philadelphia, PA, 19104 Two pathways produce ATP in mammalian cells: glycolysis and oxidative phosphorylation. The mitochondrial production of ATP, which requires oxygen, is 17 times more efficient than the anaerobic cytoselic catabolism of glucose. The contribution of these pathways to the total energy generated depends on the tissue, with smooth and skeletal muscle and retina relying to a significant extent on glycolysis. The same pertains to tumors and cultured cells. Glycolysis is regulated, to a large extent, at the level of phosphofmetokinase by [ATP], [ADP], [AMP], [Pi], [protons] and [ammonia]. Hexokinase activity can be limited by a decline in [ATP] and pyruvate kinase, by a fall in intracellular [potassium]. Oxidative phosphorylation is controlled by [oxygen], [ATP], [ADP] and [Pi] and the supply of reducing equivalents. Activation by calcium of the sensitive mitochondrial dehydrogenases contributes to a rise in N A D H / N A D and thus stimulates the respiratory chain. It has also been suggested that calcium affects mitochondrial A T P generation via other m e c h a n i s m s (inhibition of pyrophosphatase in the liver and changes in concentrations of proteins which interact directly with the ATP synthase itself in the heart). Under most rest-towork transitions the signal for increase in energy formation is a fall in ATP (rise in ADP, AMP and Pi) and in many tissues a consequent elevation in internal calcium. Rapid changes in [ATP] are buffered in some tissues (muscle, brain, pancreas) by the creatine phosphate/creatine system which, in addition, carries the diffusional flux for ADP. In some instances, there may be a specific requirement by an energy consuming process for A'rP generated by a particular pathway.

347 INFLUENCE OF MITOCHONDRIAL CONTENT ON THE CONTROL OF OXIDATIVE PHOSPHORYLATION A.T. Paganini, J.M. Foley, and R.A. Meyer. Department of Physiology, Michigan State University, East Lansing, MI 48824-110l Previously, a model based on a fast order RC analog circuit was proposed to explain the linear nouequilibrium process of oxidative phasphorylation during submaximal stimulation rates in fast twitch muscle (1). This model states that it is the thermodynamic difference of free energy of ATP hydrolysis between the myofibrils and the mitochondrlal matrix which controls cellular respiration. A prediction of this model is that the exponential time constant for phosphocrentine (PCr) changes is inversely proportional to the muscle's mltochondrlal content. We tested this by comparing PCr time constants in hiodlimb muscle of trained or thyroidectomized vs. control rats during in situ submaximal stimulation. To increase mitochondrial content, one group of adult male Sprague-Dawleyrats uMerwent a progressive interval training program on a raening-wbeel apparatus to a final regimen, after the 6th week, of 37meterstmin, 50mir'./day for 5days/week. To decrease mitochondrial content, another group of male rats drank 0.025% w/v methimazole tainted water ad fibitam. After 8 weeks, rats were anesthetized with pentobarbital end prepared for in situ stimulation in a 7.4cm bore 9Atesla Bruker spectrometer. Phosphorous nuclear magnetic resonance speewa of the superficial white gastrocoemiasmuscle were acquired via a 1.2cm diameter surface coil before, during, and after 8 minutes of 0.33 or 0.75Hz isometric twitch stimulation. The superficial 2-4mm of the gastrocneraios muscle was then clamp-frezen under liquid nitrogen and assayed for the mitochondrial matrix marker citrate synthase. The results conrmn that PCr time constants are inversely proportional to mitochondrial content, and hence support the thermodynamic model for the control of oxidative phosphotylatico by adenine nuulcotides. 1. Meyer. R.A. A linear model of muscle respiration explains monoexponential phosphocreatine changes. Am. L Physiol. 254 (Cell Physiol.23): C548aC533. 1988. (NIH Grant AR38972)

S-71

S-72

Isotopomer Analysis of Metabolic Pathways I

348

350

ISOTOPOMER SPECTRAL ANALYSIS: ENCOUNTERING VARIABLE POOL SIZE AND ENRICHMENT. Joanna K. Kelleher, Department of Physiology, George Washington University Medical Center, Washington, D.C, 20037

INVESTIGATIONS OF CARDIAC METABOLISM USING ISOTOPOMER ANALYSIS BY GAS CHROMATOGRAPHY-MASS SPECTROMETRY (GCMSI. Des Rosiers, C., Laplante A., and Comte, B. Louls-Charles Simard Research Center, Notre-Dame, Hospital and Department of Nutrition, University of Montreal, Montreal, Quebec, Canada.

Isotopomer Spectral Analysis (ISA) is a nonlinear modeling method for estimating parameters of biosynthesis using stable isotope precursors and Gas Chromatography/Mass Spectrometry. The basic form ofthe model contains two unknown parameters and assumes that precursor enrichment and the sampled compartment size are constant over the time course of an experiment. Both of these assumptions are violated in certain real experiments. To address these two issues, more complex ISA models have been developed. Variations in the precursor enrichment have been modeled by adding an additional parameter to the model. The additional parameter allows the precursor enrichment to be represented as linear gradient rather than as a constant, The linear gradient is approximated in the modeling process as a series of 10 or more steps, This procedure increases the number of unknown parameters in the ISA model from two to three. Analysis of covariance was used to evaluate the effect of the additional parameter on the fit of experimental data to ISA models with either two or three parameters. Variations in size of the sampled compartment have been modeled by replacing a metabolic steady state equation with an equation that allows the sampled compartment to grow or shrink in size with time. This model replaces the steady state model where flow into the sampled compartment equals flow out of this compartment with a model where flow in is not equal to flow out, This new equation has been used to evaluate the growth of intracellular triglyceride fatty acids in developing 3T3-L1 adipocytes. USPHS support DK45164.

349 A 13C-NMR STUDY OF G L U T A M A T E LABELING KINETICS IN THE ISOLATED PER.FUSED HEART John C Chatham, John R Forder, Jen-y D Glickson, and Edwin M Chance The Johns Hopkins School of Medicine, Baltimore MD and University College London, England. Metabolism of t3C-labeled substrates in biological systems results in the incorporation of the label into a wide range of metabolic intermediates. One of the principle metabolites observed with C-NMR spectroscopy is glutamate which is in rapid exchange with the TCA cycle via a-ketoghitarate. Analysis of glutamate labeling kinetics could be used, therefore, to calculate TCA cycle and malateaspanate shuttle fluxes, as well as provide a means for the non-invasive determination of oxygen consumption. A network of differential equations describing glycolysis and the TCA cycle, including the malate-aspartate and the glycerol phosphate shuttles was constructed. Using this network we carried out a nonlinear least squares analysis of glutamate labeling kinetics measured by 13CNMR spectroscopy in isolated peffussd hearts. In contrast to earlier studies we also included glutamate exchange between the roitoehondria and the cytosol; we assumed that in the bean this occured primarily via the malate-aspartate shuttle. There was good agreement between the measurements of oxygen consumption and glutamate labeling kinetics and the calculated values of these parameters from the network for all substrates. The sigmoidal curve of the glutamate-C3 kinetics was best fit when the cytnsolic transaminase reaction was out of equilibrium. The contribution of exogenous substrata to the overall TCA cycle flux, 89.6i~6.5% (mean:l:sd) as measured in the tissue extracts compared well with 91.4+4.2% calculated by the model. The ratio of TCA cycle flux to oxygen consumption for acetate, was 2.2:t0.1, indicating that NADH production is principally accounted for by TCA cycle flux. For glucose or g-hydroxybutyrate, this ratio was 2.9i'0.2, consistent with the existence of other NADH producing reactions

We developed protocols using 1=C-labeled isotopes and GCMS to assess in the isolated rat heart perfused under clinically relevant conditions (normoxia, anoxia or low-flow ischemia), substrata fluxes through (i) anaplerutic pyruvate carbox3dation (PC) and (ii) the reversal of the succinate dehydrogenase (SDH). Both reactions may be linked to the cardioprotective effects of pyruvate and fumarate, respectively. We used as tracers, [l-t=C]octanoate combined with (hPC) [U-I=C3]pyruvate + [U~C3]Iactate or (Ii:SDH| [U-~SC~]fumarste. Using GCMS, we determined the isotopomer distribution of tissue and perfuaste citric acid cycle (CAC) intermediates. Oxidation of [1-"C]octanoate yields [5-'=C]cltrate (M+ 1), and [1-1=C]- and [4-1=C]succinate ( M + I ) . (i:PC) DecarboxyIstion of [U-l=C=]pyruvate yields (4,5l=C=]citrate (M+2), whereas its carboxylation yields [2,3,6-'=C~]- and [1,2,313C~]eitrate (M+3). (ii:SDH) The reduction Of [U-l=C,]fumarate yields [U'zC4lsuccinate (M + 4), whereas its oxidation through the CAC yields [1,2-t=C=]- and [3,4-1=C=]succinate (M + 2). Our data reveal that substrata flux through PC in the intact normoxic heart is (i) 50-125% and 10-20% that through pyruvate dehydrogenase and the CAC, respectively, and (ii) is modulated by pyruvate, norepinephrine and fatty acids. Our data also confirm the reduction of fumarate to succlnste in oxygen-deprived rat hearts and show, for the first time, that this reaction OCCURSin the normoxic heart, possibly in a different location than the CAC. This is suggested by the enrichment of effluent succinate at 12-22% in M + 4, but at less than 2% in M + 2 for all perfusions with [U-l=C4]fumarate, including those under normoxia.

Regulation of Energy Metabolism During Exercise: Tissue, Organ & Whole Body 35[

352

REGULATIONOF LACTATEMETABOLISMIN HUMANSKELETALMUSCLE G.J.F. Heigenhauser and C.T. Putman Dept. of Medicine, McMasterUniversity, Box 2000, Hamilton ON Canada L8N 3Z5

MODELING CELLULAR METABOLISM DURING EXERCISE Marco E. Cabrera, Gerald M. Saidel and Sat~sh C Kalhan. Biomedical Engineering and Pediatrics, Case Western Reserve Univ. Cleveland, OH

A fundamental but controversial issue i n s k e l e t a l muscle metabolism i s whether the mitochondria are 0 z - l i m i t e d when skeletal muscle produces l a c t a t e (Lac-) at high exercise intensities. T r a d i t i o n a l l y , increased Lac" production has been explained as "anaerobic' production o f ATP by g l y c o l y s i s to supplement 0z-limited ATP production. However, i t may be due to a mass action e f f e c t secondary to increased pyruvate concentration ( [ P y r ' ] ) r a t h e r than l i m i t a t i o n o f 02 supply to muscle. With increasing muscle contraction i n t e n s i t y , the dominant fuel source becomes endogenous glycogen. Accordingly, increased g l y c o l y t i c f l u x r e s u l t s in increased [Pyr'] and [NADH] f o r Lac" conversion by l a c t a t e dehydrogenase, a n e a r - e q u i l i b r i u m enzyme. Flux through t h i s enzyme is l a r g e l y dependent on substrate concentration. In contrast, pyruvate dehydrogenase (PDH) is a f l u x - g e n e r a t i n g enzyme

regulating the rate of Pyr" conversion to acetylCoA, thus regulating its flux into the Krebs cycle. Thus, i f glycolytic flux is greater than the flux through PDH, the mass action effect of increased [Pyr'] and cytosolic [NADH] will produce Lac'. Our major working hypothesis is that increased Lac- production during intense muscle contraction

is nut due to 0 z l i m i t a t i o n

at the

mitochondria but results from a discrepancy between the rates of Pyr" production and oxidation.

S-73

Inferences about lactate (LA) metabolism and oxygenation state of a tissue are often based on blood or muscle LA measurements. However, the mechanistic basis of the relationship between 02 and arterial or muscle LA concentrations, during submaximal exercise are not fully understood. To provide a framework for analyzing processes involved in metabolic changes associated with energy production during exercise, we developed a mathematical model that links cellular metabolism to whole body responses. Our aim is to examine and quantify the mechanisms that Control LA accumulation during exercise. The model, which incorporates tissues with different metabolic characteristics, is based on dynamic mass balances for glycogen, glucose, pyruvate, LA, Oz, and CO2 By numerical solution, the model simulates responses to moderate exercise. Simulations predict (a) substrate concentration changes in muscle, splanchnic bed, and other tissues; (b) changes in o~her metabolltes (eg, ATP) whose reactions are coupled to the main reactions. Mode| simulations closely predict the pattern of change in substrates and control metabolites found experimentally. A large decrease in muscle O2 concentration did not affect muscle respiration. Redox state decreased to 50% its initial value during exercise. With exercise initiation, LA increased abruptly, most likely as a result of a concurrent increase in pyruvate due to the sudden stimulation of glycolysis induced by the sharp rise in phosphoryiation state. Therefore, during moderate exercise (a) oxygenation at the tissue level is sufficient even during the transient state and (b) increases in LA levels in muscle and arterial blood are mainly due to the sudden increase in the glycoly~ic rate, and not directly related to the decrease in muscle 0 2 concentration and redox state induced by exercise.

Modeling Metabolism in Mammalian Organs

S-74

356

353

T R A N S P O R T , BINDING, A N D R E M O V A L O F D R U G S IN LIVER: MULTIPLE INDICATOR DILUTION STUDIES. K.S. Pang, A.J. Schwab, and C . A . Goresky Faculty of Pharmacy, University of Toronto, Toronto, Ontario, Canada M5S 2S2 and the Montreal General Hospital, Montreal, Quebec, Canada H3G 1A4

Abstract Not Available

The multiple indicator dilution technique has been utilized in rat liver perfusion studies for examination of liver physiology and drug and metabolite processing. A MID dose containing 5~Cr-labeled rod blood cells (RBC), 125Ialbumin and [~SCo]EDTA, D20 and labeled tracer substrate under study, is injected into the inflow system of the liver. By comparison of the outflow dilution profile of labeled substrate to its hypothetical reference (constructed based on vascular binding) and by analyzing the data with the variable transit time model of Goresky, mechanism of influx, efflux, and removal, may be found, Diffusion-limited entry for polar substrates such as enalaprilat and acetaminophen sulfate, rate-limiting carrier-mediated influx for the glutathione conjugate of bromosulfophthalein, and flow-limited entry for salicylamide and acetaminophen have become elucidated. Moreover, for acetaminophen, a red cell carriage effect [slow efflux from RBC] was found to limit entry and therefore sulfation. These studies clearly illustrate the applications of the technique for the study of protein binding, transmembrane transfer, and enzymic removal in defining the handling of drugs and metabolites in the liver. (Supported by NIH)

354

357

FLUX BALANCEMODEL FOR ESCHERICHIACOU GROWTH AND METABOLISM Bernhard O. Palsson and Amit Varma University of Michigan, Ann Arbor, MI

MECHANISM OF PCR RECOVERY DURING UNDERPERFUSION IN RABBIT HEART K. Kroll and D. J. Kinzie. Bioengineering. University of Washington, Seattle. WA 98195

Flux balance models of metabolism use the stoicbiometry of metabolic pathways, metabolic demands of growth, and optimality principles to predict metabolic flux distribution and cellular growth under specified environmental conditions. These stoichiometric models have provided a mechanistic interpretation of systemic metabolic physiology and are useful for metabolic pathway design. Flux balance models provide quantitative experimentally testable predictions about cell growth and metabolic byproduct secretion. In the present report we used independent measurements to determine the model parameters for the wild type Escherichia coil strain (W3110). We experimentally determined the maximum oxygen uptake rate (15 mmol O2/g DW-hr), aerobic glucose uptake rate (10.5 mmol GIc/g DW-hr), anaerobic glucose uptake rate (19.5-mmol GIc/g-DW0hr), non growth associated maintenance requirements (7.6 mmol ATP/DW-hr), growth associated maintenance (13 mmol ATP/g biomass), and estimated the biomass requirements. The llux balance model specified by these parameters was found to quantitatively predict glucose and oxygen uptake rates as well as acetate secretion rates observed in a variety of experiments.

355 QUANTIFICATION OF DIFFERENT SITES OF ADENOSINE PRODUCTION IN THE GUINEA PIG HEART. Andreas Denssen Dept. of Physiology, Heinrich-Heine-Universit~t, MoorenstraBe 5, D-10225 D0sseldorf, FRG. Adenosine (ado), the dephosphorylation product of adenine nucleotides may serve several important functions: ado (1) is a potent corona~' dilator, (2) has negative chronotropic and dromotropic properties, (3) may induce preconditioning and (4) is a sensitive marker of isehemia. Because several metabolic pathways contribute to ado production and removal and due to their compartimentalization quantification of ado flux rates in the heart is hampered. As the ado concentrations in the ~'tosol or in the interstitial space cannot be measured, indexes were used which reflect the cytosolic concentration (rate of accumulation of Sadenosylhomocysteine in the presence of L-homoc)-steine; 200 gM') or the extracellular concentration (~nous-arterial concentration difference of ado). Experimental data were obtained from isolated perfused guinea pig hearts and subjected to a mathematical model anal)sis (5-region axially distributed model of capillar)'otissue exchange and metabolism) in order to obtain esttmates of the intra- and ex'ttacellular ado production rates. Using inhibitors of the ado metabolizing enz).xaes adenosine kinase (ITU, 10 itM) and adenosine deaminasr (EHNA, 5 ItM) the global ado production was determined to be 1.3 nrnnl rain"1 g - I Blockage of the nuclcoside membrane trar~porter (NBTI, 1 ItM) permitted to dissect the intra- from the exaracellular ado production rate, which ~ere l.l and 0.2 prnel rain"1 g-l, respectively. This rather small ex'Wacellular production rate, hm~.'er, may be of groat importance, as only 25 % of the exlracellnlar ado are released with the earonaD- effluent perfit~te, while 75 % are taken up into the ceils. Thus, under control conditions the net transmembrane ado flux is from extrato intracellular and the extmcellnlarly produced ado largely determines the concentration to which ado membrane surface receptors are exposed. In conclusion, measurements of cardiac ado production in the presence of specific enzyme inhibitors and subsequent model analysis prm~de a powerful tool to improve the understanding of cardiac ado metabolism.

The purpose of the present study was Io determine if partial recovery of creatine phosphate (PCr) during sustained coronary underperfusion is due to improved ATP supply/demand balance. Kinetic data on ATP and PCr obtained using 31-P NMR spectroscopy in isolated rabbit hearts were fit with a mass-balance mathematical model describing ATP synthesis and hydrolysis, the effect of pH on the equilibrium constant for creatine kinase, myokinase, the AMP-adenosine metabolic cycle, and membrane transport of ereatine. Pi and adenosine. The NMR data were accurately described using identical rates of ATP synthesis and hydrolysis throughout underperfusion, following a transient imbalance at the onset of decreased flow. Directionally opposite changes in ATP and PCr ~ underperfusion ' I . . . . due to a high rale of cytosolic AMP h~drolysis to ~ 1.0 [ ~ l N -,~ t adenosi . . . . hich relieved the myoki . . . . . . . tion from S 0,8 t I~k.~ ATP p,e" 4 product inhibition via AMP. Even though adenosine "~ ~ 0.6[[ / ~~ " ~ ~ ~ -1 production was a drain on the adenine nucleotide pool

do.ng uoderperfosion., h,un,0d the fa. ,n the

04

1

phospboryladon potemiaL In parallel experiments, it ~ 0.2 was confirmed that total coronary punne eftlu• during ~ 0t . . . . . n - 7I underperfusion was similar in magnitude tn the cellular 0 20 40 60 depletion of ATP. Minutes It is concluded that ATP synthesi~ and hydrolysis appear to be identical during underperfusion, and there is no evidence for improved ATP supply/demand balance. The partial recovery of myocardial PCr during underperfusion is due to mass action effects of adenosine production. r was supported b~ NIH grants HL51152 and RR01243.)

Isotopomer Analysis of Metabolic Pathways II

S-75

358

359

NMR AND PET ANALYSIS OF METABOLIC PROCESSES AJ Fisohman, H Hsu, JA Vogt, JW Babich, DM Yarmush, YM Yu, RG Tompkins, VR Young, JF Burke. Massachusetts General Hospital and Shrinere Burns Institute, Boston MA

PRODUCTION RATESOF VLOL BOUND FATTYACIDS IN HUMANS Asia Aarsiand, David Chinkes, Bruce W. Patterson, Robert R. Wolfe Universiy of Texas, Medical Branch at Galveston, Galveston, Texas Very low density lipopretein tnglyceddes (VLDL TG) are synthesized in the liver pdncipally from two sources of latty acids (FA): FA derived from lipolysis and FA synthesized de novo in the liver. In the present study the rate of secretion of de novo synthesized FA and FA from lipolysis bound to VLDL (VLDL TGFA) was measured i~ vivo. The precursor pool for FA synthesis was labeled by infusing iv. 1,2113C]-acetate. The subsequent precursor pool endchment was deduced based on the isotopomedc distribution of VLDL TGFA. The absolute synthesis rate of VLDL TGFA is calculated based on precursor endchment, rate of incorporation of label into the pool, and VLDL TGFA pool size. Measurements were made in 5 normal male in the basal state, and after 1 and 4 days on a high carbohydrate diet (energy intake -.=2.5times energy expenditure). The rate of secretion of individual VLDL bound FA is shown below (average, pmol/kg/day • SEM, * P~;0.05 statistically different from Day 0, 1" Ps'0.O5 different from Da'/

NMR and PET are important techniques for the detailed evaluation of metabolic processes in vivo using invasive and nonqnvasive methods. We developed a generalized steady-state model for determining tricarboxylic acid cycle fractional fluxes from [13C] NMR data and a kinetic model for calculating protein synthetic rate (PSR) in muscle by PET using [11C] methionine. The NMR model relates measured mole fractions of [13C] glutamate isotopomers to fractional fluxes and predicted mole fractions of isotopomers of oxaloacetate (OAA) and acetyI-CoA. Fractional fluxes are determined by fitting the model equations to NMR parameters by use of nonlinear least squares. The PET model assumes negligible transemination and transmethylation and contains vaSCular space, tissue precursor and protein compartments. A specific metabolic system represented by published [13(3] NMR data from extracts of heads porfused with [130] acetate, [13C] pyruvate (PYR), and [130] acetate plus [13C] PYR was used to test the NMR model. The intensities of predicted [t3C] NMR splitting patterns were compared with observed values, and there was excellent agreement between observed and predicted signal intensities. With this model, the OAA-dedved fraction of inflow to PYR, PYR-derived fraction of inflow to acetyI-CoA, citrate-derived fraction of inflow to OAA, and PYR-dedved fraction of inflow to OAA can be determined. The kinetic model was validated by PET measurements in dogs. The results of these measurements demonstrated that the PSR was similar for paraspinal and hindlimb muscle; 0.171+0.062 vs. 0.207+0.050 nmoles/min/g (p=NS). The value for PSR determined by the stable isotope technique, 0.27+0.050 nmoles/min/g, was similar to the PET measurement.

Palmitste (C16:0) de novo 2.6 • 1.2 40.8 • 20.0" 113.3 • 42.0"1" I lipolysis 118.0 • 32.6 71.8 • 35.0 362.9 • 109.1"1Stearate (C18:0) de novo 0.2 • 0.2 1.6 • 1.2 8.6 • 3.3"1" lipolysis 17.5• 11.1 • 38.0• 1. Oleate (C18:1) :le novo 0.4 • 09 5.2 • 2.7 28.2 • 12.7"1" lipolysis 140.0 • 33.5 81.2 • 35.8 363.0 • 109.0"1" Conclusion: Palmitate is the main product of carbohydrate stimulated de novo lipogenesis. Oleste, the other main VLOL TGFA, is largely derived from lipolysis9 Total VLOL TG secretion rate should be derived from the rate of secretion of the individual VLDL bound FA.

359a

360

CEREBRAL METABOLIC COMPARTMENTATION AS REVEALED BY I3C NMR, 13C-13C A N D 15N-13C ISOTOPOMER ANALYSES

GLUTAMATE METABOUSM IN THE GUT MEASURED WITH I6OTOPOMER ANALYSIS P.Reeds, F. Jahoor, D.Burrin, J. Henry & E.Frazer Children's Nutrition Research Center, Baylor College of Medicine, Houston, TX

Aviva Laoidot The Weizmarm Institute of Science, Rehovot 76100, Israel A method for investigating metabolic compartmentation of neurotransmitter amino acids in rabbit brain, assessed by 13C NMR and isotopomer analysis of glutamate, glutamine, GABA and aspartate carbons, was recently developed in our laboratory (1). Here we present s new approach for determining the metabolic compartmentation of neurotransmitter amino acids in rabbit brain from 15N labeled metabolites following 15N~ administrations by 13C spin coupled to 15N. During the course of this investigation, 15N isotope effects on the 13C chemical shifts of L-[2-15N]glutamate, [2,5-15N] glutamine, [2-15N]alanine and [2-15N]aspartate were noticed. Data of 15N isotope effects on 13C chemical shifts, in conjunction with 13C-15N splitting patterns were used to determine 15N enrichments of several cerebral amino acids. New metabolic pathways to remove cerebral ammonia are suggested. and A. Gopher, I. Biol. Chem. 269, 27198 (1994). ~.9 A.A. Lapidot Lapidot and A. Gopher (to be published).

Fatty acids

Source

Day O

Day 1

Day4

I

Because the intestine receives substrates from the circulation and the lumen it is likely that gut metabolism is compartmentalized. Arterial-portal mass balance measurements suggest that the mucosa extensively metabolize enteral glutamate and arterial glutamine. The objective of this study was to quantify the first pass metabolism of lumenal (dietary) glutamate by using a combination of srtedo-portal sampling and glutamate tracer infusions9 Six piglets (6.7 kg body weight ) received s 6 h constant intragastdc infusion of Ut3C-glutamate (30 pmol/kg/h) and =H-phenylalanine (10 pmol/kg/h). The animals were fed hourly throughout and portal blond flow was continuously measured with an ultrasonic flow probe. The isotopic endchments of all mass Isotopomem of glutamate, glutamine, alanine and proline were measured by NCI GC-MS in simultaneous samples of portal and artedal blond taken over the last h of it, fusion. Isotopic steady state was achieved in all isotopomers by 4 h. of infusion. 72% of the phenylalanine dose appeared in the portal blood confirming tracer exit from the stomach and absorption by the small intestine. Only 5% of the [M+5|~]lutamate dose appeared in the portal blood so that 95% of dietary glutamate (610 pmol/kg/h) was metabolized in flnst pass. Portal [M+5l~lutamine was undetectable in 3 of the 6 animals. 13% of the dose was found in lower mass isotopomere of glutamate. There was a positive portal outflow of [M+3]alanine ( g % dose) and [M+5] prol~ne (3% of dos'e). [M+I] and [M+2| alanine were significantly endched but the low mass isotopomem of proline were not.

S-76

Poster Presentations

361

364

AN IIVIPLANTABLE SENSOR FOR CONTINUOUS MONITORING OF LACTATE Dale A. Bakes Department of Bioengineering, University of California, San Diego La Julia, CA 92093

ANAPLEROSIS THROUGH PYRUVATE CARBOXYLATION IN PERFUSED RAT HEART: FLUX MEASUREMENTS USING ISOTOPOMER ANALYSIS OF CITRATE. Comte, B., JettY, M., Cordeau, S., Vincent, G., and Des Rosters, C. Louis-Charles Simard Research Center, Notre-Dame Hospital, Montreal University, Montreal, Qudbec, Canada.

Lactate is an important metabolite for which the development of a continuously operating implantable sensor would be advantageous. A sensor which could be placed at specific sites in the body may be the most desirable means of monitoring lactate, a metabolite whose concentration can vary substantially over a pedod of a few minutes. Monitoring lactate concentration has important consequences for health. An implantable lactate sensor would have potential value in the study and diagnosis of acidosis, acute circulatory shuck, heart disease, ns well as for continuous monitoring in surgery, exercise physiology studies and nconalology. Sensors reported in the literature for lactate have been based oa several different detection mechanisms. However, most of these sensors were not intended to be implantable nor designed to continuously monitor lactate concentration/n vivo. The sensor to be described here is designed to be implantable and is capable of continuous in vivo lactate monitoring. It includes a lactate electrode based on immobilized lactate oxidase coupled to a stable oxygen electrode and an oxygen reference electrode to account for the role of local oxygen. This design renders the response essentially independent of oxygen concentration. In vitro characterizatioa demonstrated that the sensor responds specifically to lactate over a broad concentration range. This response is stable for mort than one week during continuous in vitro operation at body temperature. During acute implantation in the canine right atrium, the sensor pmdaced a continuous recording of lactate concentration that was in good agreement with conventional, discrete blood assays. This implantable sensor for continuous monitoring of lactate is being used in studies of regional myocardial ischemia in the canine heart. These investigations of cardiac lactate metabolism and cardiac biomechanics will be discussed along with pertinent lactate sensing issues.

Pyruvate is an efficient energy substrate in that it can supply both acetyI-CoA (oxidation) and oxaloacetate (anaplerosis) to the citric acid cycle (CAC). To study factors which regulate the metabolic fate of pyruvate in the intact haart, wa have validated an approach to assess relative fluxes through anaplerotic pyruvate carboxylation (PCI, pyruvate decarboxylation (PDe) and the CAC (acetyl-CoA formed from POC and fatty acids oxidation). Our strategy is to perfuse rat hearts with a combination of 1=C-traders, [U-IsCs]pyruvate (PYR], [U-l~Cs]lactste (LAC} and [1-1aCtoctanoate (OCT), which are metabolized to different citrate isotopomera. Decarboxylation of [U-I=Cs]PYR yields [4,5JSCz]citrate, whereas its carboxylation yields [1,2,3-1SCat- or [2,3,4-1SCs]citrate. [1JsCIOCT is p-oxidized to [5-13C]citrate. The isotopic enrichment of these citrate isotopomers are determined by gas chromatography-mass spectrometry. With 0.2 mM PYR, 1 mM LAC and 0.2 mM OCT, PC flux expressed relative to PDC was 1.25 • 0.09 [n = 5}; it is decreased to 0.79 • 0.05 by 1 pM norepinephrine and to 0.59 :t: 0.06 by lowering LAC and PYR concentrations to 0.5 mM and 0.05 mM respectively. Expressed relative to the CAC, PC flux was 0 . 1 8 • at the higher LAC + PYR concentrations and decreased to 0 . 1 2 + 0 . 0 1 in the other two conditions. The activity of CAC is modulated by NE and concentrations of LAC and PYR : 3 . 9 9 • ,umol/min.g vs 3 . 6 5 + 0 . 5 and 2 . 6 • with lower concentrations of LAC + PYR. These data reveal an active PC in the intact rat heart and physiological importance of this anaplerotic flux could be related to the cardioprotective role of pyruvate.

362

365

DEVELOPMENT OF A METABOLIC GAS ANALYZER SYSTEM FOR USE WITHIN THE U.S.-RU$SIAN SPACE MISSIONS

REVERSAL OF SUCCINATE DEHYDROGENASE IN NORMOXIC AND OXYGENDEPRIVED RAT HEARTS. FLUX MEASUREMENTS FROM ISOTOPOMER ANALYSIS OF SUCClNATE. Laplante, A., Vincent, G., and Des Rosters, C. Louis-Charles Simard Research Center, Notre-Dame Hospital, University of Montreal, Department of Biochemistry, Montreal, Quebec, Canada.

Daniel W. Barincau Lockheed Martin Corp. JSC, Houston, Texas

Floyd BouLes NASA/SE2 JSC, Houston, Texas

Steven D. James Medical Graphics Corporation St. Paul, Minnesota

An effort to effectively munitor on-orbit parameters for cardiac output and resting metabolic nutritional balances led to the development of a new device by the Johnson Space Center 0SC) Life Sciences Project Division (LSPD) and Medical Graphics Corporation (MGC). The device, called MGAS (Metabolic Gas Analyzer System), was developed for the U.S.-Russion space missions and is based on a onmmercial system (called CPX-D) made by MGC. MGAS has the capability to determine breath-by-breath values for Oxygen, Carbon Dioxide, and volumetric flow. During 1995, MGAS will fly on the MIR space station as well as on Space Shuttle flight STS-71. The MGAS unit has all interfaces on the front panel, a volume of I fls, requires approximately 110 watts of steady state power, and weighs 21 lbs. The analyzer interfaces with a laptop computer for control and data display, and a gas tank module for calibration of the gas sensors. Gas samples, and volumetric flow measurements are obtained at the patoated MGC prevent | pneumotach, which provides a low dead space, non-bulky subject inteafface. The MGAS seftwam uses aa icoo oriented environment, that rims in either English or Russian, and lends the operator through the steps of hardware setup, subject selection, hardware cstibration, experiment performance, and shutdown operations. MGAS is used for measuring a wide number of physiological parameters including oxygen uptake ("/02), respiratory exchange ratio (RER), heart rate, ananrobic threshold (AT), and noa-invasive cardiac output(Qt). There are 50+ physiological variables that can be obtained/calculated by the MGAS hardware/anftware. The MGAS can also control an exercise device (treadmill or bicycle ergnmeter) to run graded or step-function exercise protocols.

The beneficial effect of fumarate in ischemic hearts may be linked to its reduction to succinate via the reversal of the succinate dehydrogenase. We quantitated the activity of this pathway in isolated rat hearts perfused with 0.4 mM [U13C,]fumarate, 11 mM glucose, 1 mM lactate, 0.2 mM pyruvate and 0.2 mM octanoate. The reduction of [U-'SC4]fumarate, a M + 4 isotopomer, yields M + 4 succinate, whereas its oxidation through the citric acid cycle (CAC) yields M + 2 succinate. The isotopomer distribution and concentration of succinate and other CAC metabolites extracted from effluent perfusates and heart homogenates were determined by gas chromatography-mess spectrometry. Effluent succinaro isolated from hearts perfused under normoxia (30 min), anoxia (30 rain) or low-flow ischemia {LFh120 rain at 1 ml/min) was enriched in M + 4 isotopomars at 22%, 12% and 15%. respectively. Less than 3% M + 2 succinate was detected under normoxia despite normal CAC activity, and none under anoxia or LFL Release of M + 4 succinate was 1.5 nmo[/min under normoxis, and increased 6- and 2-fold under anoxia and LFI, respectively. Total succinate release reached 9, 117 and 40 nmol/min, respectively. In all conditions, tissue succinate was less enriched (< 2% M + 4 ) than effluent succinate. Other CAC metabolites (fumarate, matate, oxaloacetate, citrate and a-ketoglutarate) were also 2 to 10 times less l~C-enriched in tissue than in the perfusate. These data confirm the reduction of fumararo to succinate in oxygen-deprived rat hearts. They show, for the first time, that this reaction occurs in the normoxic heart, possibly in a different location than the CAC.

363

366

GLUCOSE TRANSPORT ACROSS THE BLOOD BRAIN BARRIER IN SEPTIC SHEEP D. Chinkes, E. deMelo, X.-J. Zhang, D.L. Traber, and R.R. Wolfe. University of Texas Medical Branch, Galveston, TX 77555

EFFECTS OF HYPOXIC-RELATED METABOLIC CHANGES ON NEURONAL ACTIVITY; COMPARISON BETWEEN HYPOXIC RESPONSES OF RESPIRATORY MEDULLARy NEURONS AND DENTATE GRANULE NEURONS. M.M. Patti *I , D. Durand z, M.A. Haxhiu 2, T.S. Whltingham s Dept. Biomedical Engg. 1 Dept. of Medicine 2 Dept. of NeurosurgerySCane Western Reserve University, Cleveland OH. * Current address, Dept. of Psychdiogy t , Harvard Univet'sity, Gambridge, MA,

The purpose of this study was to determine the alterations in the exchange of glucose between plasma and the brain that occur during sepsis using a multiple tracer dilution method. Sheep were studied before and 24 hours after administering endotoxin, During both periods a bolus of 1-SH L-glucose, 3-O-D- 14C methyl glucose, and 6,6 ZH glucose was given systemically and samples were taken rapidly from an artery and the sagittal sinus. We confirmed previous observations that the brain contains essentially no interstitial space. In the control sheep, we found that transport into brain and phosphorylation in the brain occur with half lives less than 15 seconds. On the other hand, transport of glucose out of the brain is a relatively slow process with a half life of seven minutes. Sepsis caused a 50% reduction in arterial glucose concentration. During sepsis the rate of clearance of glucose was doubled by the brain to compensate for the drop in arterial glucose concentration. In conclusion, glucose transport across the blood brain barrier is altered in sepsis to maintain a constant uptake of glucose.

Decrease in oxygen supply to the brain, such as at high altitudes or during physiological stress, causes selective vulnerability to hypoxic insult in the central nervous system. This research investigated the effects of oxygen deprivation on the neural activity, synaptic transmission and passive electrotonic parameters of medullary parapyramidal neurons involved in respiratory control and dentate g r a n u l e neurons, using intracellular and extracellular electrophysiological recordings. Energy metabolism is directly affected by a decrease in oxygen consumption, and hence, cellular mechanisms related to energy depletion, such as function of ATP-dependent N a + / K + p u m p and variations in intracellular pH, measured spectrophotometrically, were investigated. The parapyramidal neurons (n=11) were found to survive anoxic insults significantly longer than the dentate gyrus neurons (n=29). The anoxic l~sponses of the two groups of neurons were compared in order to determine possible mechanisms that may resist the damaging effects of anoxia. It is speculated that parapyramidal neurons maintain tissue ATP levels better than other cortical neurons by decreasing their neuronal activity, thereby aiding in prolonged survival during hypoxic insults. This work was supported by NIH grant # HL25830.

Poster Presentations 367 MINIATURIZED BIOSENSORS FOR THE MONITORING OF CLINICALLY IMPORTANT BIOCHEMICALS IN VIVO Michael V. Pishko I and Adam Heller2 1Department of Chemical Engineering, Massachusetts lnstituta of Technology, 45 Carlton St. E25-342, Cambridge, MA 02139 2Department of Chemical Engineering, University of Texas at Austin, Austin, 'IX 78712 In a healthy individual, biochemical concentrations in tissues, blood, cerebrospinal fluid, and other bodily fluids generally remain within well-definod concentration ranges. When disease or injury occurs, however, the concentration of certain bioohemicals such as glucose or lactate may rise above or drop below normal physiological levels. Thus, the continuous in vivo monitoring of bioobemicals would provide physicians with a powerful tool for patient treatment and diagnosis. This seminar will discuss the development of implantable and potentially implantable biosensors to continuously monitor the concentrations of clinically important biocbemicals that serve as indicators of disease or injury, These sensors are based on the electronic "wiring" of oxidoreductase enzymes to metal electrodes. This "wiring" is based on the electrostatic complexation and electron transfer between the oxidoreductase and Os(bis-bipyridine)2Cl-containing redox polymers of poly(4-vinyl pyridine) and poly(vinyl imidazole). Such "wiring" results in amperometric sensors that are selective, rapid, and rniniaturizable to a dimension of 10~m. Using this technology, amperometric sensors for glucose and lactate have been developed. When tested in vivo in subcutaneous tissue or in the blood stream, the glucose sensors accurately monitored blood glucose concentrations. Such devices have potential applications in diabetes therapy, monitoring during surgery and in CCUs, and in exercise physiology.

S-77

S-78

Reverse Engineering the Cochlea

368 SYNTHESIS O F COCHLEAR MECHANICAL FUNCTION USING MODELS A l l y . E. Hubbard Boston University College of Engineering While some investigators struggle to solve the mysteries o f the cochlea using models, others seek to implement and utilize models that capture the essentials of cochlear signal processing. T h e g o a l s o f b i o p h y s i c a l l y based m o d e l i n g are the v a l i d a t i o n o f e x i s t i n g hypotheses, interpretation of data, and challenging the overall scientific understanding of the field. The majority of cochlear models are biophysically based. The two generic types, m a c r o m e c h a n i c a l and m i c r o m e c h a n i e a l , a l l u d e to the c o m p l e x i t y o f the mechanisms that are represented. Substantive understanding of the cochlea has not yet been achieved. Non-biophysical models have the goal of efficiently mimicking cochlear-like responses for use in applications such as speech processing or cochlear prosthetics. Signal-processing models of the cochlea use filtering constructs from engineering signal processing. There are two common formulations of signal-processing models of the onehlea. One formulation is a serially-arranged cascade of filters, such that each filter has a lower tuned frequency than the one preceding, Alternatively, the input signal can be fed simultaneously to a parallel bank of filters, each tuned to a different frequency. CMOS computer chips can be used to implement either type of signal-processing, albeit, with considerable difficulty at the present time.

371 ANALOG VLSI MODELS OF COCHLEAR FUNCTION AND THEIR ~PPLICATIONS IN SPEECH RECOGNITION Andreas G. Andreou Sensory Co~unications Laboratory Electrical and Computer Engineering Johns Hopkins University Cochlear modeling research has been pursued by different groups over the last decades. Such work can be divided into two broad classes. Some modeling work is aimed at understanding the function of the auditory periphery and the role of the different physical structures while the goal of others is the engineering of a better acoustic processor for speech processing tasks. We review and discuss work in our laboratory aimed at the engineering of a small, low power, cochlear based, analog VLSI acoustic processor for auditory processing tasks including speech recognition. We also discuss the utility of modeling at the circuit level and take the view that this is an important tool at the interface between the mathematical models of signal processing and mathematical models of physical function.

(Funded by the Office of Naval Research)

369

372

MODELING THE TEMPORAL CODING PROPERTIES OF AUDITORY NERVE FIBERS David C. Mountain, Sanjai Singh. Socrates G. Deligncrges Dept. of Biomedical Engineering. Boston Univ.. 44 Cummington St.. Boston. MA 02215

FOCUSING FIELDS FOR COCHLEAR IMPLANTS F,A. Spelman *§ B.M. Chipton*. S. Corbett ^, J. Marlynittk^, L Swanson ^, C.N. Jolly*, A.H. Vain*, T. Clary*Center for Bioengineering, University of Washington; *Regional Primate Research Center, University of Washington, ^PI Medical, Portland, Oregon; -Clary lntemationui, Issaquah, Washington

Accurate reproduction of the temporal ceding propenies of auditory nerve (AN) fibea-s by auditory models is important in applications ranging from cochlear ~osthetic| to automatic speech recognition systems. Biophysicaily-based models of cochlear transduction and sysaptic transmission have been proposed which adequately capture the temporal coding preperties of the AN fibers, but are computationaily intensive and are difficult to implement in hardware. Signalprocessing models, on the other hand are computatinnally more efficient, but these models generally fail to capture all of the features ef AN temporal coding. Models of cochlear transduction and synaptic transmission can he divided into three subsystems. The first subsystem represents the coupling between cochlear mechanics and inner hair cell (IHC) transdnction. The second subsystem is the IHC tnmsduction process which results is an IHC receptor potential and the third ~absystem represents the synaptic transmission hetwoen the DIC and the AN fiber. Several different models of each of these hinphysicai processes will be reviewed along with simplifying assumptions which have allowed as to develop a anw unified signal-processing model o f AN temporal coding propenles. The model's arehitanture lends itself naturally to a hardware implementation based on an envelope detector followed by a cascade of two automatic gain control stages. The new model has been tested with tone burs~s, amplitude-modulated stimuli, and complex sequences of tone bursts. The model results have been evaluated by comparing to data from physiological experiments and in aimas~ all cases the model results correspond closely to the experimental data,

Cochlear implants are vr successful, so successful that they are recommended as aids for the severely deaf an well as for the profoundly deaf. Despite their worldwide success, implants have electrode arrays which excite neurons over a relatively great spatial extent which precludes the simultaneous, non-interactive excitation of separate popuiatioas of ceils of the auditor).- nerve. We suggest the use of closely-spaced triads of electrodes, placed near I Rosenthal's canal, which should excite small groups of oclls. Models and measurements in j the ears of guinea pigs and monkeys support the bypothasis that focusing of field potentials is possible in the implanted ear. We have designed, built, and tested an electrode array that incorporates the potential benefits of focused excitation. The array allows an electrode density exceeding that of present devices by at least a factor of two. This design permits threedimensional focusing. We have tested the array in vitro, where its potential fields are predictable with straighffoi'waxd models, and have begun tests in vivo to measure the actual spread of excitation. This ~ork was supported by STIR Number R41-DCO2424 from NIDCD of the National Institutes of Health

(Funded by the Office of Naval Research)

370

373

A FEATURE-BASED AND EVENT-BASED BROAD-CLASS SPEECH RECOGNITION SYSTEM Carol Y. Espy-Wilson, Boston University, 44 Cummington St~ Boston, MA 02215

SINGLE AUDITORY NERVE FIBER RESPONSES TO AMPLITUDE-MODULATED ~ C PULSE TRAINS Shuwan Xue, Chris van den Honett Div. of Otolatyngology-HNS, Box 3550, Duke University Mud. Center, Durham, NC 27710

In this paper, we discuss l) a signal representation of speech that consists of acoustic correlates of the asticulatory-based linguistic features that comprise a phonological desoriptioe of speech sounds and 2) an acoustic-event-oriented strategy for speech recognition. Using a linguistic-based signal representation is a way of explichly extracting only the massage-hearing components of the speech wavefurra. Such a representation is in contrast to cepstra or capstral-based coefficients used in most recognition systems today which contain both linguistic and extrsiingnistic information. At present, we focus on the manncr-of-articuimion features sonoranh syllabic, fricated and continuant and combine them for re.cognition of the brood classes: vowel, so.ernst consonant, stop, and fricative, in addition to silence. The recognition strategy employed is based on a feature hierarchy and is implemented in an event-seeking mode where landmarks associated with the acoustic correlates are detected for recognition. Consequently. speech is not recognized on a frame-by-frame basis or by segmentation into phoneme-sized pieces to which inbr are assigned. Recognition results obtained with the TIMIT database show that our brood-class recognition system (BCR) has comparable performance to that of a frame-based Hidden Marker Model (H/diM) system utilizing 39 coefficients including Mel-freqoency based cepstra, normalized energy, and their first and second order derivatives.

Understanding the encoding of slim"ins envelope in the temporal discharge pattmns of the auditory nerve fibers is important for the d~ign and parameter determirmtinn of cochlear implants and speech processors that use ampfitode-modulated puise trains. The narrow dynamic range of single-pulse I/O functions of the anditory-ncn, e fiber would predict no spikes elicited when pulse intensities drop below "threshold", and faint ou eve'/y pulse when intensities increased by a few dB. To test this hypothesis unit responses from anditory nerve were recorded with AM trains of 100 ps/phasa biphasic pulses with 50 Hz sinusoidal modulation. Modulation depth was 100% or 25% with a I000 Hz carrier pulse rote, and 100% with a 500 Hz cartier. Results differed from prediction of the I/O function in two ways. Flrstiy respOUS~ ShOwed a much broader dynamic range for modulation than was predicted by the I/O ~nction. Behavior at low intensity was ~ n a b l y well predicted by the I/O function spikes were first observed when the peak intensity pulses t~sched the single-paise threshold, and were absent elsewhere int he modulation cycle. But as intensity was increased spike probability fell consistently below I/O function predictions, and modulation of the reaponsu was preserved at high levels where saturation was expected. Clipping was not observed, even with intensity maxima as much as 6 dB above the saturation point of the I/O function; and firing ceased during modulation minima even when the individual puises were weft above the single-pulse threshold (in the 25% modulation case). 'I~ese data have significant ramifications for the operation of prosthuses which encode spectral energy within a paasb~d in the pulse train envelope. They suggest that such envelope information can he conveyed over a wider dynamic range than previnusly predicted,

Vision Prosthetics I

S-79

374

376

CORTICALLY BASED SENSORY PROSTHESES: PHYSIOLOGICAL INVESTIGATIONS. P.L Rousche, C.T. Norditansco, E.M. Maynard, ICS. GniUory, and R.A. Normann. The John Moran Laboratories, Department of Bioengineering, University of Utah, Salt Lake City, LIT.

MONITORING THE DIMENSIONAL A N D ELECTROCHEMICAL STABILITY OF IR AND ACTIVATED 1R MICROELECTRODES DESIGNED FOR NEURAL PROSTHETICS. L.S. Robblee, R,B. Jones, and G.S. Jones. EIC Laboratories, Inc.

The concept of cortically based sensory pro~-theses is based upon a nttmber of premises. One general premise is that sensory percepts can Ix: evoked by current injunctions into sensory cortex. While this has been demonstrated with electrodes that penetrate human visual cortex (Baket al., 1990), it has not been validated with penetrating eloctxodas in other sensory modalities. Another premise is that quality maps exist from primary sensory transducers to higher senso W centers in the brain (ie, tonotopy, retinotopy, and sematotopy). We have investigated these issues in a series of chronic and acute physiological experiments in feline sensory cortex. To study the mapping of vlsual and auditory cortex, the 100 electrode, "Utah Intracortical Electrode Array" COIEA), was implanted acutely and chronically. The visual maps (made within 6 degrees of asea centralis) were spatially non-fiocar: we could predict the receptive field location from the site of the recording in VI only tO within 0.5 degree. For monocular ceils, the aggregate cortical magnification w ~ 1.5 turn/degree. Over the spatial ex'tent accessed by the UIEA, the maps of auditory cortex revealed an approximately logarithmic tonotopicity from 5 kHz to 35 kHz_

The currents required to evoke sensory percepts were studied in cats chronically implanted with the UIEA. These cals had been trained to lever press in response to auditory stimulation. We found that behavioral responses could be evoked by injection of currents that ranged from 20-120 uamps, (250 Hz, 150 usoc, biphasic, cathodic first, 150 usoc biphasic interval). These data suggest that current injection via the UIEA can evoke sensory percepts. However the non-liocarities of the retJnotopic and tonotopic maps may require signal processing circuitry to remap sensory input onto primary sensor: cortex in a more conformal fashion.

375

Abstract N o t Available

Cyclic voltammetric techniques are used to estimate the electrochemically active surface area of lr microeloc~odes and to identify the presence of defects in the seal at the metal/i~ulator junction. Microelectrades with a good seal at the junction will have better long term stability than those with a defective seal which allows penetration of electrolyte between the metal and the insulator. Cyclic voltammograms of solution redox species such as [Ru(NH3)J§ at microclectrodes have a characteristic sigmoidal shape at slow scan rates under conditions of radial diffusion and steady state mass transport. Distortion of the sigmoidal shape is evidence of a defective seal at the metal/insulator junctinn and indicates that the microelectrode is unsuitable for chronic implantation. If the voltammogram is normal, then the background-corrected limiting current for the reduction of the [Ru(NH;)~]"j species is used for approximating values of the apparent radius of inlaid disk mir or the radius and length of conical microelectredes. A~er activation of lr to produce surface oxide, cyclic voltammetry is performed in electrolyte without the redox species to determine electrochemical area based on apparent capacitance and to evaluate the stability of the oxide charge capacity during long term soaking. Results will be presented which illustrate the application of these cyclic vnltammetric methods for monitoring microclectrode integrity during long term in vitro soaking in physiological saline. The difference in voltammetrio behavior over time between "good" and "poor" microelectrodes will be demonstrated. Changes in electrode dimensions and surface areas determined by cyclic voltammetry appear to correlate better with impedances and phase angles measured at low frequency, e.g. < 50 Hz, than they do with values measured at 1 Id-lz. This work was supported by contracts from NIH-NINDS.

S-80

Neural Modeling

377

380

A NUMERICAL MODEL OF PROPAGATION IN SPIRAL GANGLION CELLS L. A. Curtee, C. C. Finley, P. A. Leake, and G. T. Hradek Research Triangle Institute, Research Triangle Park, NC An anatomieally-basad model of cochlear neuroo eleetrophysiology has been developed and used to investigate the role of anatomical feanues of the spiral ganglion cell on action potential conduction and failure in the neuron. As a means of mimicking activation at the hair cell, excitation was initiated at the first node of the peripheral process, The peripheral process consisted of five l tan nodes with a 200 gm intemodal spacing, typical values for auditory fibers. The central process was composed of l 0 nodes of the same size. For the simplified cell body geometries with nomlnal dimensions which have been studied, propagation across the cell body region was always successful when the action potential was initiated on a resting neuron. If, however, the membrane was refractory due to an initial excitation, the second excitation of a pair could exhibit conduction failure. Propagation of the second action potential in response to a stimulus pair was more likely to fail as the interpulse interval decreased and the second excitation increasingly entered the relatively re fructo~ period of the first excitation. Success or failure of conduction was also dependent on cell body geometry with constrictions in the axonal processes adjacent to the cell body having a greater influence on the failure of conduction than the cell body diameter. As the length of the peripheral process was decreased, the likelihood of propagation failure increased. This rmding may have implications for intracochlear stimulation since the site of excitation is not at the hair cell terminus but may be on or near the cell body. Ulti-nstructural examination of spiral ganglion ceils has indicated the presence of large unmyelinated regions on the axou adjacent to the cell body intemode. The influence of these regions on propagation is currently being investigated, and prelimma~ results will be presented.

378

T H E BIDOMAIN CONCEPT IN NERVE STIMULATION MODb':LING. TIME DEPENDENCE AND T H E "EQUIVALENT FIBER" K.W. Airman, M.D., Ph.D. Dept. Otorhinolaryngology, Hospital o f the Univ. of P e n n s y l v a n i a Modeling electrical stimualtion of nerve requires mathematical descriptions o f the stimulus as well as the target tissue. The latter contains multiple fibers, so the b i d o m a l n formulation is employed to include c u r r e n t pathways i n both interstitial and i n t r a c e l l u l a r space separated by a membrane. These models have previously explored the ateady-stste r e s p o n s e of a target f i b e r to an instantaneous a p p l i e d electrical field. T h e model is now adapted to include the complex membrane impedance as a function o f time, permitting examination o f the response to a t i m e - v a r y i n g stimulator current source o f arbitrary wavnshape. The t i m e - d e p e n d e n t interstitial potential field is described in response to a step pulse in order to quantify the influence o f the membrane on the applied f i e l d o f target fibers, F u r t h e r m o r e , the effect o f fiber diameter and depth on the observed s p a c e and time constants o f an " e q u i v a l e n t fiber" (a myelinated nerve fiber whose segmental m e m b r a n e properties are spread o v e r the e n t i r e length o f the axon) are d e t e r m i n e d . R e s u l t s i n d i c a t e the i m p o r t a n c e o f i n c l u d i n g the m u l t i - f i b e r b u n d l e in a model for e l e c t r i c a l n e r v e stimulation, conclude that l o w stimulator frequencies are necessary for a steady-state ( e l e c t r o t o n i c ) approximation, and d e m o n s t r a t e the usefulness o f the model to examine the tissue response to complicated waveshapes and electrode g e o m e t r i e s .

381 MAGNETIC STIMULATION OF EXCITABLETISSUES Peter J. Basset Biomedical Engineering and Instrumentation Program. NCRR, NIH

Maenetie stimulation is the term that describes the stimulation of excitable tissue using a time-varying magnetic field. A decade ago, this phenomenon was just a curiosity whose underlying physical mechanisms had not been elucidated, Today. magnetic stimulation is used widely ns a noninv~ive and painless method to diagnose disorders of the central and peripheral nervous systems. Moreover, its mechanisms of action are better understood. We dnscrihe a model of magnetic stimulation that predicts a) the current distribution in the stimulating coil from its circuit characteristics, b) the net electric field distribution produced in tissue by the time-varying current in the coil and by charge accumulation within the tissue, and c) the response of excitable tissue to the net electric field, This model, derived from Maxwell's equations and from a cable equation of an axon, has been useful in establishing the physical bases of magnetic stimulation, in making testable predictions about there possible mechanisms, in designing and interpreting experiments, in optimizing stimulation regimens, and in designing novel stimulating coils. Moreover, this model draws a fundamental connection between magnetic and electrical nerve stimulation. We will discuss the concepts of an activating function, volume of stimulation, virtual anode, and virtual cathode for magnetic stimulation. The phenomenology of stimulating curved nerve fibers will also be considered. Subthreshold and suprathmshold stimulation will be considered for both myelinated and unmyelinated axons. While excitation of nerves and muscles is desirable in a neurological evaluation, it is both undesirable and potentially dangerous if it occurs as the imaging gradients are switched on and off during an MRI. Issues associated with the stimulation of excitable tissues during MRJ will also be discussed. Finally, some limitations and possible improvements of the existing models of magnetic stimulation will be described.

379

STOCHASTIC MODEUNG OF SPIRALGANGLION CELLS J.T. Rublnsteln Department of Otolaryngology, The University of Iowa College of Medicine

A biophysical model of the cat type 1 spiral ganglion cell has been developed based on single ionicchannd open-dose kinetics. The model is implemented on the Cray 1390 at the San Diego Supercomputer Center. It Includes 24 nodes of Ranvier, myelinated Intemodes and a myellnated cell body. Each Ranvler node includes between 1000 and 4000 voltage sensitive sodium channels each with one inactivation and three activation particles. When electrically stimulated with 100 microsecond monophasie current pulses from a point source extracellular electrode, the modal behaves much like electrically stimulated cat spiral ganglion cells In vivo. The relative spread of threshold (RS) is similarto that observed experimentally. The latency distribution (litter) is stgn~cant/ygreater for cathodat then for anodal stimuli. This dichotomy is also observed in vivo and is shown in the model to be due to the passive properties of the cell body, Because the PST histograms produced by the model closely resemble those obtained in vivo, it should be possible to reconstruct and simulate tha electrically evoked whole-nerve action potential.

382

EFFECTOF FINITE NUMBERSOF CHANNELS ON INTRINSIC OSCILLATIONS IN I.AYER II CELLSOF TI'IE ENTORHINALCORTEX J o h n A. White" a n d Alan IL Kay" 9 D e p L of Blomed. Eng., B o s t o n Urdv. Boston, MA 0 2 2 1 5 + Dept. ofBtoL SCL. Univ. o f Iowa. Iowa City. IA 5 2 2 4 2

INFLUENCE O F NODAL-PARANODAL PROPERTIES ON CONDUCTION IN MYELINATED NERVE FIBERS J.A. Halter I and B. Zupan z *Division of Restorative Neurology and Human Neurobiology, Baylor College of Medicine, Houston, ~ 77030-3498,jah~bcm.tmc.edu and 2At Laboratory, J. Stefan Institute, Jamova 39, Slovenia, [email protected]

T h e t o r h y t h m a p p e a r s to play a n i=nportant b u t til-deflned role in a rn~mrrml'n active exploration of its e n v l r o n m e n L Layer It stellate n e u r o n s of the e n t o r h l n a l cortex exhJblt a n i n t r i n s i c r h y t h m l c t t y t h a t p r o b a b l y c o n t r t h u t e s to the g e n e r a U o n of t h e t a rhythm: w h e n depolarized from r e s t . t h e s e cells s h o w s u s t a i n e d b u t i r r e g u l a r oscfllaUons w i t h f u n d a m e n t a l frequencies in t h e t h e t a r a n g e . OscfllaUons arine in the s u b t h r e s h o l d regime a n d are still evident w i t h s u p r a t h r e s h o l d stimuli. We h a v e s h o w n t h a t s u c h oscillations m a y arise t h r o u g h t h e i n t e r a c t i o n of a p e m l s t e n t i n w a r d c u r r e n t a n d a slow o u t w a r d c u r r e n t . I n this s t u d y , we analyzed t h e effect of finite c h a n n e l n u m b e r s on t h e d y n a m i c s of t h e oscfllaUon t h r o u g h s t o c h a s t i c nimulaUons. P r e l / m i n m ~ r e s u l t s i n d i c a t e t h a t the s t o c h a s t i c m o d e l c a n e x p l a i n s e v e r a l f e a t u r e s of the d a t a n o t a c c o u n t e d for by deterministic m o d e l s (e.g., t h e trregularlty of the r h y t l u ~ a n d o c c a s i o n a l s k i p p e d beats). E x p e r i m e n t a l d e t e r m i n a t i o n of t h e a p p r o x l m a t e n u m b e r of c h a n n e l s involved In the g e n e r a t i o n of t h e i n t r l n s l c oscflJation of layer It cells, t o g e t h e r w i t h the s t o c h a s t i c slmuJaUons, indicate t h a t the s y s t e m is far from t h e determJrdstic llmR. Supported by ]he WNtaker Found~on (JAW). and ONR and NIH (ARK).

Myeliuated nerve fibers exhibit a complex anatomy in the nodal region which includes a marked nodal-paranodal constriction and an intricate paranodal structure where the myelin sheath is separated from the axon by a narrow periaxonal space. There is also a complex character to the biophysical properties of the axonal membrane in this region. For example, there is a very high ratio of the density of sodium channels to potassium channels in the node, with an inverse relationship in the paranodal and internodal regions which have relatively low ratio. The relative influence of these properties on the conduction of action potentials was studied using a detailed distributod-parametar model of the myulinated nerve fiber. The conduction properties and the character of the individual action potentials is strongly influenced by changes in the nodal-paranodal properties. The conduction velocity (CV) is yew sensitive to minor changes in the thickness of the periaxonal space in this region. CV can increase when the nodal-paranodal axonal radius is reduced to 22% of that present in the internodal region. CV is also significantly affected by the properties of the axonal ionic channels. For example, small shifts in the sodium channel activation properties affect CV as well as the firing threshold of the node of Ranvier. These nodai-paranodal properties are also possible contributors to activity-dependent neural plasticity. Supported by the Whit=kea"Foutldatioo,the Viv~n L SmithFoundationfor Re=loratistcN9 the W.M.Keck Center for CsmpmationaI Biology,the DevelopmentS~pport program of the InformationTechnoloW program at BaylorCollegeof Medicine(JAH) av.dthe Ministryof Sdencr and Technolol~fof the Repttblir of $lovenla(BZ).

Cochlear Nucleus Circuitry 383

386

SPEECH P R Y I N G IN THE PERIPHERAL AUDITORY SYSTEM Murray B. Sachs Johns Hopkins University School of Medicine

TEMPORAL CODING OF PERIODICITY PITCH IN INTERSPIKE INTERVAL DISTRIBUTIONS OF COCHLEAR NUaJEUS POPULATIONS Peter Cariani Eaton Peabody Laboratory, Massachusetts Eye and Ear Infirmary and Depatmumt of Otolngy and Laryngology, Harvard Medical Schnol

The representation of speech in the discharge patterns of the population of audltory-nerve fibers (ANFs) has beast the subject of numerous studies over the past two decades. Although great emphasis has been placed on repnnamtatiom base on the tmaporal or "phase-locked" aspects of single ANFs, for most purposes a robust representation can be obtained by considering only the average rates of ff'e ANFs. In this presentation, we will review the major features of mt~mpresantatiom and consider the processing of thuse represantetloea by populatioea of ceils in the vmtral cochlear nudeus. Plots of average d ~ rate versus best frequency (BF) of ANTs provide ao:urate representations of the formant structure of vowels in quiet and in the presence of background noise. However, at high stimulus levels or in high levels of background noise, the representations across fibers with high spontaneous rates (and low lix'mholds) deteriorate berause of rate saturation effects. Under these conditions robust repraseatations depend on the activity of the population of low spontaneonus rate (high thtashold) fibers. These fibers saturate at higher levels than do the high spontenoous rate fd3ers asu:l because of the effects of two.tune suppression they maintain a more robust representation in high levels of background noise. We will discuss evidence that certain populations of cells in the vnetral cochlear "listen selectively" to their low spontaneous rate ANF inputs u n d e ~ appropriate stimulus conditions.

Complex periodic Sounds am per~ived by human listane~ m have low, "periodicity pitches" at their fundamental frequencies, even when there is no anargy present at those frequencies ('missing fundamentals'). Both physiological and modelling studies (Cariank Delgutte, & IGang, Soc Neurosci Abstr 1992 18(1):383; Meddis & Hewltl, J Aconst Soc Am 89:28832894) suggest that strong negraI correlates of periodicity pitch eats be found in population intersplke hiterval distributions of the auditory nerve. For a wide range of stimuti the most frequent interval present in the auditory nen, e con~peeds to the low p i t h which is heard. Iu the present study the responses of single cochlear nucleus units ia Dial-annsthe33zed cats were obscured in order to deten'~e whether such a "population interval code" for periodicity pitch might aL~o hold for differeut neural popelatiofls there. Units of different physiological response types (Prl, Pri-N, Chop-S, Chop-T, On, Pause) were pr-~euted with s set of osrapiex stimuli (click trains, AM tones and noise, synthetic vowels). For each stimulus and within each response-type, all-order interval histograms of most responding units showed prominent peaks near the pitch period or its sabharmonics. In weakly responding units pitch periods could often be inferred from patterns of longer intervals. Exceptions to these rules were a few units with intervals related to "chopping" (Chop-S) or to characteristic freque~y (Prl). The data am consistent with a popolatiou interval representation of periodicity pitch iu each response-type population. Such distributed, temporal codes may provide an alternative to central representations for pitch based on populations of tuned "pitch-detectnss." Supported by NIDCD Grants DC02356 & DL-"~119.

384

387

"PITCH-TAGGED" SPECTRAL REPRESENTATION IN THE COCHLEAK NUCLEUS S E Ke;IsgII; V.M. Richards (U. of Penn.); B.T. Wyman and E.D. Young (Johns Hopkins) Loyola College, Baltimore, MD 21210

OPEN-SET SPEECH RECOGNITION FROM MULTICHANNEL ELECTRICAL STIMULATIOI~ OF THE HU1MAN COCHLEAR NUCLEU~ R.V. Shannon , S.R. Otto , S. Staller , C. Menapace 1House Ear Institute, Los Angeles, CA 90057 and 2Cochlear Corporation, Englewood, CO 80112

Chopper uoits of the ventral cochlear nucleus provide a robust rate representation of the spectra of complex sounds, such as vowels. Choppers with best frequencies (BF) above 0.50.7 kHz also phase-lock to the fundamental frequency (f,) of the vowel. The result is a spectral representation where units code for spectrum by discharge rate and f, by temporal pattern. This could be useful in segregating multiple speech sources, where differences in fundamental frequency of voicing are a contributing cue. Single unit responses to concurrently presented pairs of speech stimuli, with different f, s were studied. The stimuli were CVC syllables of the form/b/N/l/, where the vowel V ranges over I I English vowels, chosen to have formant frequencies that s p ~ the frequency range of interest. Three versions of each syllable were synthesized, with fixed f, s of 88, 100, and 112 Hz. We describe responses to the steady-state vowel portion of the syllables only. Attar units were classified, pairs of CVCs were presented. Vowels were chosen so that one had a formant frequency slightly above BF and the other slightly below BF. By changing the sampling rate of the DIA converter used to generate the stimuli, formant peaks were shifted relative to the unit's BF. This provided a range of stimuli with variations in f, and spectral dominance relative to the unit's tuning curve. Examination of responses from chopper units to these stimuli supports the idea of a "pitch-tagged" representation. As the vowel spectra are shifted relative to BF, the unit's discharge rate changes as the total energy within its tuning curve changes. The relative synchronization of the unit's responses to the two vowel f, s also changes to favor the vowel whose harmonics of fo dominate the unit's response.

385 RESPONSES OF VENTRAL COCHLEAR NUCLEUS UNITS TO LOW FREQUENCY. TWO-TONE AMPLITUDE MODULATED STIMULI. W. P. Sbofaer, S. Sheft, and S. J. Guzman Parmly Hearing Institute, Loyola University Chicago, 6525 N. Sheridan Rd., Chicago, IL 60626 One of the goals of an auditory prosthetic device is to provide heating-impaired listeners with the ability to understand continuous speech. The temporal envelope of continuous speech has a complex wavaform that contains meaningful energy at low frequencies (<50 Hz), In order to investigate the encoding of complex, low frequency amplitude modulation in the auditorj system, we have studied the temporal discharge properties of ventral cochlear nucleus (VCN) units of the chinchilla in response to best frequency tones that were amplitude modulated by two-tone complexes using low frequency modulators (fmodl and fraud2). In particular, neither the stimulus waveform nor the halfwave rectified stimulus contain spectral energy at the envelope beat frequency (fmod2-fnmdl). Halfwavr rectification yields spectral peaks at fmedl and fraud2. Spectral analysis of post-stimuhis time histograms of VCN unit discharge was carried out. The amplitudes of the peaks in the neural spectra am measures of phase-locking to each frequency component. Neural spectra obtained' from VCN unit discharge show peaks at fmodl, fraud2 and fraod2-fmod 1. These results indicate that VCN units encode the two modulation frequencies and the envelope beat frequency in their temporal discharge patterns, The temporal response to the envelope beat frequency can be accounted for by compressive nonlinearities in the auditory system. In addition, there exists an hierarchy among unit types in terms of the magnitudes of their temporal responses to the modulation frequencies and the envelope beat frequencies. (Supported by a Program Project Grant from NIDCD)

Significant open-set speech recognition has been obtained from two patients with s multiehannel auditory brainstem implant IABll. Seventeen patients have received the multinhannal ABI at the House Ear Clinic in Los Angeles. The ABI consists of eight, 1-mm platinum disks on a Silastlc substrate with a Dacron fabric backing for stability. The electrodes are driven by an implanted receiver/stimulator similar to the one used in Cochlear Corp.'s cochlear implant system. The electrode is positioned in the lateral recess of the IV Ventricle, adjacent to the dorsal cochlear nucleus, during a surgical procedure to remove a vestibular schwannomas. Patients receive an average of 27.5% improvement in sentence recognition when the the ABI is combined with lipreading (compared to lip-reading alone). Recently, two patients demonstrated significant recognition of words in sentences with only the sound from the ABI. In one case, performance was as high as 86% correct on CID everyday sentences presented live voice. Recognition of words in taped sentences for this patient ranged from 24% correct to 58% correct. Interestingly, this patient also had the largest perceptual range of pitch across the different electrodes of any ABI patient. A second patient consistently recognized a more modest, but significant 14% correct of words in CUNY sentences in the sound-only condition, with a high score of 22% correct. These results demonstrate that significant speech recognition is possible with prosthetic electrical stimulation of the human cochlear nucleus.

S-81

S-82

Vision Prosthetics II

388

389

VISUAL PERCEPTIONS IN BLIND PATIENTS FROM FOCAL ELECTRICAL STIMULATION OF THE RETINA E. de Juan~Jr., M. Humayun, R. Greenbergand G. Dagnelis. Departmentof Ophthalmology,Johns Hopkins Medical School.

STEPS TOWARD THE DEVELOPMENT OF AN IMPLANTABLE RETINAL PROSTHESIS FOR THE BLIND. Joseph F. Rizzo, and John Wyatt. The Massachusetts Eye and Ear Infirmary and the Massachusetts Institute of Technology. Degeneration of the retina with loss of rods and cones is a relatively common cause o f blindness. In these conditions the inner retina, which is topographically connected to the brain, remains relatively healthy. Our goal is to develop an implantable prosthesis to stimulate the inner retina to restore some vision to blind patients. We face two very significant problems: design of a prosthesis that will not damage the eye and protection of the prosthesis from the salt within the eye. Our prototype minimizes the internal electronics. An external infrared laser provides both power and signal to the internal components that include a photodiode alray, signal diode, stimulator chip and cantilever with metal electrodes. The five year collaboration has resulted in the design, building and testing of a photodiode array and a stimulator chip, detemination of thresholds for stimulation of retinal ganglion cells, recording of cortical responses following focal stimulation of the rabbit retina, demonstration of biocompatibility of the materials for up to one year, and development of surgical methods to implant and attach the device to the retina. The project still faces the most severe challenge -- the hazards of the neuralelectronic interface.

Objective:. Retinal degenerationssuch as retinitis pigmentosa destroy the retinal photoraseptoreand pigment epithelium, but largely spare the ether retinal layers. Currently, there is no cure for the lost photoreceptorfunction. Herein, we evaluate the fees~ility of bypassingdamaged photoreceptorsand electricallystimulating the remaining viable retinal layers as a means to provide limited visual input to patients blind due to severe photoreceptordegeneration. Methods: In the operating room under local anesthesia,focal electricalstimulation of the retinal surfacewith brief biphasiopulses was performedusing small probee inserted through the sclara. The procedurewas performedin five subjects with little or no light perception. Three subjects had retinitispigmentosa, one had age-ralated macular degeneration, and the remaining one had an unspecified retinal degeneration from birth. Results: Stimulationelicited 9 visual'perceptionof a spot of light (phosphane). Subjects who previouslyhad useful vision accuratelylocalizedthe phosphenee according to the retinal area stimulated. Two subjects could track the movementof the stimulating electrodeby reporting movement of the elic'dedphosphene, and could perceivetwo simultaneous phosphenes upon independent stimulation with two electrodes. In a resolutiontest, one of the subjects resolvedphosphenesat 1.75" center-to-centerdistance (i.e., 4/2000 visual acuity). Conclusions: These findings demonstratethat local electricalstimulation of the retinal surfaces in patients blind from outer retinal disease results in focal light perception seeming to arise from the stimulated area. Such findings in acute experiments warrant further researchinto the possibility of prolonged retinal stimulation and improved resolution.

Neural Systems I 390

393 AN ENGINEERING APPROACH TO THE SENSORY R E C E P T O R W. Wung und K.K Norwich Institute of Biomedical E ~ & University of Toronto

APPROXIMATION OF INPUT/OUTPUT RELATION OF A BIOLOGI C A L N E U R A L S Y S T E M B Y F E E D F O R W A R D N E U R A L N E T S M. A. Saglam, V. Z. Marmaralis and T. W. Berger Dept. of Biomedical Englneerin9, Uni-

versitp o/Southern California, Los Angeles, CA, 90089 Every seaso~ event begins at the micrescopic level. ~ tbe senmry evem is mediated by photons impinging on the nnina, or gas molr interacting with the olfactory epithdium, the Idological responss begins at the level of the sensuW reccplor and its associated prima~ afferunt n~xon. Alth~gh the reoeptors of the wariens sense~ ntedalilies, m~h us light, ~uell, and t a n ~ differ in their anatomical c o - - o n , their responses are ve~ similar. Fer crumple, ~ to a steedystimufus,or the messured pregresslvedecreasein ruing rote,us nxx,nled in the neuron to a macrusoopically constant s~nuing is a proFe~ shared by almost all the modalilie~ The prope~ that stimuli of higher intensities elicit higher neural fmng frequencies and the mathematical laws that link the frequencies to the intenfities a ~ likewise sham:l by the various sensory medalitius. We propose a single equation of four pararaetets based on the entsopic pnncipln of Bolt~aann which is capable of accounting for almost all sensory phenomena, ompifical laws, and mies of thumb connecting the firing rate of a primary afferent neuron to the intensity of the stimulus and the duration for which the stimulus is applied. This equation is capable of handling lsgh timr and steedy sensory inpots. Thus we have, perhaps for the first time, a "tranffer function" capable of predicting nr outputs from general sensory inputs and valid for a wide range of serLsory rectors. The function has been tested extensively on expedmemal data. It suggests ",hat a neural network might be ennslog'ted using such Bolt,maann-nenmm.

The traditional approach to characterize biological systems is the use of Volterra series. The nonlinear input/output properties of hippocampat dentate gyrus is characterized experimentally by stimulating the perforant path fibers by random impulse train and recording the evoked population spikes. The relation between population spike amplitude and intervals of the random pulse train was approximated by a three-layer perceptron. The structure is a universni approximator, the nigmoidai activation functions were used in the hidden units and the output was linear. This approach has certain advantages over the previously utilized non-parametfic modeling method. The main advantage is the significant reduction in the required data length. We show a three layer net with as few as 2 input and 3 hidden units trained with 50 pairs has the prediction ability of the kernels up to third order in terms of mean squ~red error (MSE}. Preprocessing of the training data increases prediction accuracy. Higher order nonlinearities also can be captured easily by adding more input units to the structure. Having more than three units in the input layer did not increase the accuracy of the prediction, which might be suggesting that higher order nonlinearities are not that strong in the dentate granule cell layer. Achieving better MSE values with small size of training sets also suggeats that there might be some erroneous points (or noise) in the data. The algorithms that are robust against noise are tried. This r--,:~earchwas supported by ONR, NIH Biomedical Simulations Resourse, and NIMH.

391

394

A C O M P U T A T I O N A L STUDY OF S T R U C T U R A L C O R R E L A T E OF SYNAPTIC TRANSMISSION. I.-S. Liaw*. M. Baudrv. G.A. ChauveL and T.W~ Bereer. Dept. of Biomedlcni Eng.. Program in Nenroscienen, University of Southern C.alifomia, Los Angeles. CA 90089-2520

C H A R A C T E R I Z A T I O N OF EVENT I N P U T - CONTINUOUS O U T P U T NONLINEAR NEURAL SYSTEMS USING A VOLTERRA S E R I E S . W. A. Seafinee*. V. Z. Marmarelis. T. W. Bereex.

We have developed 9 computational model of 9 central synapse to study the role of synoptic gcomettT in neural transmission. In particular, we examine a form of synoptic plasticity known us short-team potentiation (STP) and the differences in dic response mediated by two types of receptor-channeL AMPA and NMDA. It has been demonstrated that during the expression of SI"P, the evoked responsus of both AMPA and NMDA components are greatly increased. Furthermore, sensitivity to micronjention of exogenous agonist is selectively enhanced for NMDA; responses to micruejcction of AMPA are unchanged (Xie & Borges Soc. Neurosci. Ahstr.. 1994). We investigated how the relative positions of presynaptic release site and postsynaptlcrecoptor-chunneh may influence the magnitude and time com'se of the exeitoto~ post-synalxiC emTents (EPSCs). Oor simulations show that AMPA-mediated EPSC is 42% higher while NMDA-medlatod EPSC is 17% higher when the receptors are fight underneath the release site than when they are 40 nm away. Forthennore, the rising and decaying phases of AMPA-mediated EPSC br faster, consistentwith expodmental data. NMDA w.ceptor has a higher affinity and is thus less sensitive to the relative position. If gintarnate is delivered to the synapse by means of peffusion, then its distribution becomes more uniform and the influence of receptor location should be reduced for beth AMPA and NMDA-mediated EPSCs. Tho simulations strongly support the hypothesis that the potentiation of AMPA-mediatod EPSC is due to the relocation of the receptor on the postsynaptie membrane while changes in the kinetic parameters underlies the potentiation of NMDAmediated EPSC. NRSA fellowship. ONR, NIMI-I (MH51772. MH00343, MH52194), NIH Simulations Resumce, and the Human Frontiers Science Organization.

392 APPLICATION OF A NONLINEAR LEAST SQUARES TECHNIQUE FOR EEI'I/vIATING ELECTROTONIC PARAMETERS OF A COMPARTMENTAL NEURON MODEL Jinng Ding and Herbert F. Voigt Department of Biomedical Engineering, Boston University, Boston, IdA 02215 Electrotonic parameters (ETPs) of neurons, such as the membrane time constants (~..), ales~Jotonic length (L), somatic shunt ratio (e), sure.tic conductance (go), dendritic input condilctatlcr (ga), and dendritic domlnAnen ratio (p), reflect passive membraac propestius Contributing to spatial and temporal processing of subthreshold synoptic inputs. ~ arc usually estimated from several r ~ r arid coeltlcicnts ~ from the mcmbrune's voltage responses to hypctpolstlzlng otmrent injections (e.g. Rail 1977; Durand 1984; White r al. 1994). In this study, see use 9 least squares approach to ~r directly from impulse responses of 9 corapomnontal neuron model consisting of 9 multisegment dendritic equivalent cylinder and a lumped parameter soma. The model output is 9 nonlinear function of these E T ~ and is obtained by solving a set of state-spaue di~eaenco squations recursively. A modified ~ n b ~ g - M a r q u a r d t algorithm (Mort nt al. 1980) is applied to minimize the sum of the squared errors between the data and model prediction. This approach was tested by titling the model to 38 impulse respous~ simulat~ using various combinations of ETPs (%: 6.--17 ms;, L: 0.5.-.2.0; e: 0.01~L0; go: 1.1-2.9 CA'I't; g,: 1.8.,-86 G~'l; p: 0.9--29). White Ganasina noise (o'=0.I mV) was added to these respons~ to mimic real expor~entol conditions. Results show that the average error for individual ETP estimates ranges from 4% (L) to 17"/*(e) when I
Dept. of Biomedical Engineering and Program in Neuroscience. University of Southern California, Los Angeles. CA 90089.1451. Volterra (1930) developed a general functional power series (FPS) representation for dynamic nonlinear systems such that the kernels of the series have 9 simple and useful interpretafiun. For systems whose input is a sequenoc of disc~te events, the Volierra series characterizes the dynamic nonlinearities of the system in terms of the syue~istic effects of multiple input events on the system response. The fu'st order kernel specifies the response to a single input event, the second order kernel specifies the change in the response due to the synergistic effects of pairs of input events, ete. Thus. the Volterea series provides a useful characterization of the nonlinearities of the system us wall as a useful predictive model. In this work, we investigate the cross-non'elation technique for estimating the kernels of the Voltcwa series for systeans with discrete event input and continuous OUtlmL The precedu~ uses not'put data measta~l while applying a Poissun random sequence of events to the input. Krausz (1975) developed an orthogonal FPS model for this Poissun test input and showed how 1o estimate the kernels of the series using output-input crosscon'elatlons. We show that these kernels ar~ different from the Volterra kernels in they contain components involving all higher order nonlinearities in the system. However, the difference between the two sets of keruels is small if either the mean event rate or the area under the higher order kemeis is small. Thus, cross-correlation is a useful method for kernel estimation. We describe practical techniques for estimating the Volterm kernels using cross-correlation and we demonstrate diese techniques using simulated data and expr data from the rabbit ha,in. Supported by ONP,. NIH, BMSR, and NIMH.

S-83

S-84

Neural Systems II

395

398 DYNAMIC C O N T R O L P R O P E R T I E S O F SACCADE WHEN COMBINED W I T H VERGENCE EYE M O V E M E N T Huimin Zbu and George K. Hung Dept. Biomedical Engineering, Rutgers University, Piscataway, NJ 08855-0909

EVALUATING COHESIVENESS OF SOME PATHWAYS IN THE HIPPOCAMPAL DENTATE GYRUS USING MULTISITE RECORDING ELECTRODES. C.Y. Perron* & T.W. Beree..~.~.Dept. of Biomedical Engineering and Program of Neuroscience, University of

Southern California. LosAngeles. CA 90089-1451,

Purpose. An improved data processing method was developed to properly isolate the vergence and saceadic components in the eye movement response to a cembined stimulus. The dynamic cenlrol properties of ssccudic components were compared to those of pure saceadic responsu. Method. Binocular eye movements were recorded for pure saceades of 2 to 6 dug and saccedes of I tO 6 deg combined with convergence or divergence of 2 to 10 dug with sar stimulus delayed by 0 to 300 msee in five subjects. The vergence component during the ssr interval ~as interpolated by a third order polynomial curve fitting obtained from the vergance velocity on beth side of saccudic interval,. Saceadic cempononta with vergense portion removed were ealculated from the original and vergenee eye movements. Saneadin amplitude was represented by the saecadic difference of both eyes normalized by the vergenee stimulus. Main sequence were plotted for each subjects. Results. The main sequence data of ssecade combined with vergence are quite close to those of the pure saecade, having slopes of 32 to 36 (deg/ser The peak velocities of both eyes am cinsely correlated. The mean of the normalized ssecudic amplitude difference is close to zero, As the occurrence of vergeuce and saecade become more distinctly separated, the deviation of saecadic inequality decreases towards that of the normal pure saecade. The inequality of saccades is bi-directinnal. Conetasiun. Vergenee is not censistently facilitated by saecude. The dynarmc control properties of saceade is not affected by its interaction with vergence. The progranmung of these two systems am relatively independent.

Curront-source density (CSD) analysis, and paired pulse stimulation (PPS), were used to investigate the electrophyslologieal characteristtcs o f the ipsilateral longitudinal associational projection o f the dentate gyms, in v/vo. Stimulating electroaes were placed into the hilar region of the dentate gyrus, as well as, into the angular bundle o f halothano-anesthesizedqqe'w Zealand white rabbits. Recording electrodes were placed more rostrally in the granule cell layer. Characteristics o f the associational input to granule cells were compared to those o f the perforant path (PP). C S D analysis provides an improvement in the resolutinn o f sinks and sources o f currents by poff6i'ming a secondorder spatial partial derivative along an array o f recordings equaliy spaced. PPS analysis on the other hand allows the study o f the temporal interaction o f inputs. By recording from vertical arrays o f electrodes, the topology o f synaptic mnervation for each stimulating impulse can be determined with the C S D analysis, while PPS analysis shows the amplitude variation o f those sxamptie inputs as a function o f the interstinauli intervals (ISI). This allows for the evaluation o f the cohesiveness o f these synaptie inputs by comparing their response magrotude to different ISI values, as well as~ to different stimulus mtensities. The latency and spatial distribution relative to the anatomical topography can also be determined. Discrimination o f multiple pathways (or divergent pathways) is crucial when stimulating a location that leads to multiple target areas. Tills is particularly true when stimulating in the hilus, since the hilns contains cells x~th axons that proJect bilaterally throughout the hippoeampns, The results o f these experiments show that the four C S D components with the shortest lateneies for a hilar stimulation are an inner molecular layer excitatory synaptie sink (5 ms latency), a hilar excitatory synaptie sink (8.5 ms latency), a CA3 somatic source (9.5 ms latency), and a CA 1 stratum radiatum sink (10 ms latency). The inner molecular layer sink could be clearly delienated from all the others by its PPS analysis pattern and by its response to different stimulus intensities. Supported by ObllL NI]-I BMSIL NIMH and MRCC (CYP sludentship).

396

399

I M P L E M E N T A T I O N OF S Y S T E M A L G E B R A T O C H A R A C T E R I Z E UNOBSERVABLE NEURAL SUBSYSTEMS. M.T. Chian,, V.Z. Marmarelis, ILl. Scinbassi, T.W. Betger Department of Biomedical

MULTIUNIT RECORDINGS IN THE DORSAL COCHLEAR NUCLEUS (DCN) OF GEREIL~ USING MULTI$1TE MICROELE~rRODES John R. Lovetl, Kenneth E. Htnr and Herbert F. Voigt Department of Biomedical Engineering, Bo*ton University, Boston, MA 02215

Engineering, University of Southern California, Los Angeles CA 90089 The nonlinear characteristicsof subsystems of real neurobinlngical networks may be represented by expressing the inpol/ontput relation as a goneralized functional power series. The kernels of these functional series can be expressed as the n th order impUlSe responses which are found d m m g h the recoiling of alectrophysiological responses to random impulse train stimulations. This method has bean applied to hipponampni regions of the brain. Since many elements of these regions may not be directly re.un~ed from, it is necesss~ to represent subsystems matbemafically in order to indirectly deduce the inpueuntpat proper'desof an ualmown subsystem from the InOpelliesof other known

manipulations in the frequency domain. This method may be used to characterizeand combine many feedback and fnodforwatd systems in the hippoeampal formation of the brain. In a proven fer.dbeckmodel the deatate gyrus of the hippocampes may be expressed as the feedtbsongh suheystem while the basket cells of this region may be expt~se.d as the inhibitory feedback subsystem. Although the basket ceils are difficult to record from. it is possible to identify the feedback subsystem's inlmffonttmt ~ t e d s t i e s from the known properties of the overall system and the fcedthrongh subsystem. The feedthrough system properties may be isolated by pharraasological.ly blocking the unknown feedback portion, making it possible to mathematically decompose the unknown subsystem when the other system eleme.ta are known. The plausibility of this quantitative approach has initiated the representation of more physiulogicatly realizable systems allowing the cheraeterization of elements otherwise not available through direct measm'emenL Supported by ONR, NIH, BMSIL NIMIL

This project investigated the use of multisite mlcrouleetrod~ u 9 tool for reoording funr interactions between un~ in the DCN of gerblis. In past reaea~h both tingle site metal and mullhi~ microeinctmdes have been used to record unit par1 in the DCN, but have yielded r I~ulta. By investigating the recording charaotorlstir of mukiskn cinr the possibility of electrode bias affecting recordings of the functional intetactfuns in the DCN was te~uL Ten experiments were performed using anesthetized female Mongolian gerbils. Two types of multinite microelectrodeA, fabricated by the University of Michlgan ~mter for Integrated Sensort rod CircuS, were used--G2 probes with onc shaak aad five sites and DEVI'H 2 probca with sixteen sites and four shanks. The DCN was reachod by inserting electrodes through the bulla on llu: lefr side of the sknU and into the cerebullurabeneath which lles the DeN. Isolation charactcrlstlesof muRialte elsctrodes were investigated and digitized waveforms at various gimuhil levein were ~.arded. Background activity that increased with increasing levels of acoustin stimulation prevented the isolation of individual unit's action potentials. Cross-correingrams were computed from activity derived from triggering on background noise in vivo and in vitro to investigate propagation delays within the interstitial tissue. In comparison to meal electrodes, both G2 and DEPTH 2 multhite miemeleO--odea were dcterminod to have poor isolation characteristics. It was also shown that propagation deinys in the interstitinI tissue wete sufficient to create artifactual excitatory peaks. Multisite minmelectrode= were shown to obtain driven activity on separate sites with low best frequency ratios, indinsting increased probabilities of recording carrelated activity, Overall the use of these psiXiculsr multialte micmelectrodes to record from pairs of units in the gerbil DCN was ineifeetive. Background driving to tonal gln~uintion could be observed and the.so electrodes may be useful for the study of evoked field poasRiais 9 the relationship between best frequency and site separation. However, these muRisite mierouleetrodes should not be used to record in the gerbil DCN until the isolation obaracterlstles are improved. (Work supported by NIH, The Whitaker Fouodatinnl

397

400 O N T O G E N Y OF T H E PAIRED-PULSE I N D E X : A M E A S U R E OF DENTATE G R A N U L E CELL M O D U L A T I O N J.FL Blaise and I.D. Bronzino I)el~ of Engineering and Comptiter Science, Trinity College. 300 Summit S t . Hartford, CT 06106

- The present study was undennken to examine age-related changes in the faciinatoryfmhibit0ry modulation of hippecampal dentate granule cell modulation in freelymoving tats at 15, 30, and 90 days of age. Dentate grannie cell population responses evoked by paired-pulse stimulation of the perforant pathway were reonfded across a range of inteq~ulse intervals. Population spike amplitude (PSA) measures calculated from these responses were used to construct a Paired-Pulse Index (PPI), inideative of Re relative level of inhibitory and/~ fanllitatory modulation of granule cell. excitability. Using this meastae, three phases of grannie cell modulation have been identified in the adult: 1) an early inhibitory phase which resnlts in depression of excitability at interpulso intervals (IPls) bet~ean 10 - 40 mzec.; 2) a period of enhanced cellular excitability (facilitation). at IPls in the range of 50 - 200 msecs., and; 3) a podod of late-onset inhibition eccunlng at IPIs betw~n 300 - 1000 resets. Results indieata an age.depe.udent rise in inhibitory modulation, with 15 day old animals exhibiting a markedly lower level of early inhibition, less facilitation, and a lack of late inhibitory modulation compa~d to adult animals. The PPI obtained from the 30 day old group fell between thnse of the 15 and 90 day old groups, These data suggest that although the population of inhibltm7 interueerons is in place at the time of birth, functionally mature connections between these intemeurons and the granule cell population of the dentate gyros develop over the first thice- four weeks of postnatal life. The results also indicate that the PPI can be effectively used to examine the impact of perinaud insults on the functiunal development of hippoeampal neural circuitry in behaving animals as they mature, Abstract

THE METHOD OF COINCIDENT FIRliqG ANALYSIS MULTICHANNELNEURONAL AC'IIVITY Peter D, Perepelkin Pavinv lest. of Physiol,,St.Petershurg, Russia, Whitaker Center, Arizona State Univ.. Tempe, AZ 85287 In reni~ae the investigslionof the procegsesof synchronization,of single dischargesof remotely anangod neurons or neuronal ensembles, during different mental or motor activity, can be realizedthrough various raethods.In most cases,astralmethods for analysis use several netn'on combinationsof two neurons, le this case h~foanafon willexlx~entiallyincrease and its value will decrease, so diete may be interactionbetween rc~'e than two neurons.Thus, rn~-einfon~alve and mr reliable is the method using the spike conjunctions in all ~'l~steeednoa~s - SigkeSpaceSyndnonization (SSS), Because use of the correct theoretical formula is very r a simpler method, the fiinoroialdistribution formula, w~ investigated.E x ~ t a l conaul shows than withincceasodtime of analysis and amount of registered channels, the approximatinnof the pr~isine disw3~6o~~ binomialdDx~xulonis consideredaccurate for practical use. If experimentaland theoretical distributions are different, we examine every bin of the histogram. The difference between histogramsmay be e/dietin,=casesof synchionirati~ in nil bins, or on increased synch~oulzationin the same bin. Thus, this method deals with the determinate reaction neuronal system In the case of the probability nm~0~ it elements,which,as know,w~ be rno~ dtserminationwith increase elements. This technique may be used with the purpose of gstisfical analysis of the data obtained for such summary synchronizationof all neuron ensembles for such time synchronizationof concrete neuron ensembles in eonaete time tasks. This method deals with sing'rural-functionalrelationshipsbetween analyzed neuronal populations,and so neuronalsauenaes influence thr appearance of synchrony potential activity in remotely ananged neuronal ensembles. This method was used In the clinic for analysis of neuronal activity (14 channels) from deep mu~ues of die humanbrain and at the laboratory for analysis of interactionof r r,euronal ensembles (10 channels) in monkeys. It showed: 1) die existence, reliability and the importanceof SSS. 2) there was a nodular character dependency between discharge frequency changes and spike synchsonization.3) that Sgg com.istodwithdefmilephasesof the monkeys behavior.4) there were concrete SSS for definite phases of ps~huloginaltes~ 5) the interactionof neuronal processes, which have been investigated,demonstrated that s~chspaceintsraczionsas SSS could be a special mechanism for quicL short phases of integral mental processes in the I~mate brain.

Neuroelectrodes for Recording and Stimulation

S-85

401

404

MULTICHANNEL CMOS INTRACORTICAL RECORDING ELECTRODE ARRAYS Jennifer L. Lund and Keasall D. Wise Center for Integrated Sensors and Circuits,The University of Michigan, Ann Arbor, MI 4g 109

ELECTRICAL AND HISTOPATHOLOGICAL EVALUATION OF CNS STIMULATION USING MICRO-MACHINED, THIN-FILM, IRIDIUM ELECTRODES. J. D. Weiinnd and DJ. Anderson Graduate Program in Bioengineering and Biosystems Lab, Dept. of Elect. Engineering and Compat~ Science., Univ. of Michigan, Ann Arbor, MI 48109

The complex interactions of cells in the central nervous system can be most effectively studied via the recording of extracelhilar potentials generated by large populations of individual neurons. A structure capable of making such recordings must have numerous electrodes whose size and spacing can be carefully controlled, yet which has as few leads as possible connecting it to the outside world in order to reduce the damaging tethering forces of those leads. Additionally, to successfully record rnicrovolt signals, it is desirable to provide amplification as close to the electrodes as possible to minimize degradation over the high impedance signal pathway. The must advanced structures for this ptmpose developed to date are muldchannst, micromachinnd silicon electrode arrays with integrated CMOS circuitry. These devices feature 32 electrodes, require only three interconnect leeds, and am batch fabricated using an industry standard 3gin p-wall CMOS process combined with hnron etch-stop technology for substrate definition. Eleetmdes are formed by inlaying gold or iridium into photolithographically defined sites on polysilicon conductors. The dieleetrinally insulated polysilicon conductors conneet the recording sites wi~ signal processing eirouitry fabricated in an undiffused well at the base of the same substrate. 2 combmaticos of eight electrodes may be electronically selected from among the 32 available sites; each selected channel is then connected to a high impedance CMOS preamplifier which has an in-baed (100Hz-10kHz) AC gum of 300 and a DC gain of less than one while tolerating offsets of +100mV. The amplified signals are multiplexed onto a single bidirectional data line at a nominal rate of 22kHz per channel and then buffered for transmission to the external world. Active probes also contain circuitry for I/O control and self-test. They am powered from a single 5V supply and have a total power dissipation of less than 2roW. Multichannel recordings have been successfully obtained from rat and guinea pig using active silicon microprobes, Signal.to-noise ratios were as good as or better than those observed from passive devices. Performance issues associated with these structures, including noise sensitivity, bias stability, and chronic reliability, will be discussed. The applicability of these devices for cell localization, mapping, and migration studies will be demonstrated, and an easily portable user interface and data acquisition system for active probes will be described. This work has been supported by the Neural Prosthesis Program, NIH NINDS N01-NS-4-2303

Activated hidharn has several characteristics that make it an attractive stimulating eleelmda mateaist. It has a high charge storage capacity, is biocompatible, and can be patterned on silicon wafers with solid state circuits to form an integrated, mnitiple elecm0da prosthetic device. This papar repor~ the results of e x p e ~ e n t s where passive (i.e., no eleeuonies), mlcro-machlaod silicondevices with multipleIridiumeloctrode siteswero used to chronicallystimulatethe C N S of gttirteapig. In ritro experiments wea~ also conducted. T w o electrodeswere used in experiments, differingonly in theirsitessizes, Both electrodeshad four shanks. One eleetrode had eight sites of 1000 ttrn a distributed two on each shank. The second electrode had site sizes of 400, 800, 1200, and 1600 p.m =. The second design grouped one pair of like size electrodes on each of its four shanks. The site spacing was 200 trm along the shank. Shanks were spaced 200 ttm apart. Biphasic current pulses (50 pA, 0.1 ms per phase, 1000 Hz and 250 Hz) wese presanted four hours daily for five days. Two electrode sites were used to form a bipolar pair. Eleettical measurements of the electrode.ti~ue system wese recorded during the stimulation period to monitor any change in electrode or tissue impedance (measured at 1 kHz) or electrode charge eapaciry, Stimulated tissue and the implanted device are mcovel'ed for histology. The electrodes maintained consistent electrical parameters for the week of testing. Charge storage capacity, of activated Iridium is known to increase if the electrode is anodicaRy biased between pulses, This effect was demonstrated in vivo. Histology revealed no tissue damage directly related to stimulation. SEM analysis of the recovered po3bes revealed the pt~ence of biological material adhering specifically to eleetrode sites that were used fee stimulation. This work was supported by NIH hINDS N01-NS-2-2379.

402

405

SILICON SUBSTRATE MICROELECTRODE ARRAYS: DESIGN VERSATILITY AND FABRICATION L F. Hetke, K.D, Wise and DJ. Anderson Center for Neural Communication Technology and Center for IntegratedSensors and Circuits, Dept. of Elect.Engineering and Computer Science,Univ. of Michigan, Ann Arbor, MI 48109

THE USE OF LOW STRESS SILICON NITRIDE AS A BIOCOMPAT[BLE INSULATOR FOR NEUROPROSTHETIC APPLICATIONS. Kris James(l), Todd Whimhurst(2), Grsgory Kovacs(2), and RIchard A. Normann(1). (l) The John Moran Labs, Universityof Utah. (2) Center for Integrated Systems, Stanford University.

Since the mid-1960s, several efforts have focused on the development of neural electrodes fabricatedusing integratedcircuitprocessing techniques. Photolithographyand micromachining permit the realizationof muhichannel probes with reproduciblefeaturesthatapproach the sizeof the cellsof interest. Probes 15pro thick, 1.2mm m 2.5cm long and with shanks as narrow as 30pro.have been fabricatedwith up to 64 sites. Virtuallyany two-dimensional substrateshape can be achieved and can include innovative features such as holes, barbs, tabs and hooks, Recording probes with hnih-insiliconcables have maintained stableimpedances in the I-TM,Q range and remained functional in guinea pig CNS for up to one year. The intended application for a probe greatly influences its shape and features. For example, probes designed for laminar recording in structures such as cortex or inferior collicuhis have single shanks with up to 31 sites spaced at 100-5001Xm to sample activity from different layers. Muhlshankod devices aae also an option and often have shanks on center separations of 200-300p.m. When interaction with a tissue volume is of interest, it is possible to assemble thsee-dimensional arrays using multiple two-dimensional, multishanked probes inserted through a micromachined silicon platform. The current 3-D structure consists of a four-by-four array inserted into a 1.2mmdiameter platform. More recently, extensions of the process technology have resulted in probes with microchannels buried within the substrate which permit simultaneous multlpoim chemical delivery and muhichannel re~ording/stimulation. Separations between electrode sites and outlet ports can be as small as 2.5p.m. The range of shapes and structures achievable with this technology will be illustrated by examining several recent probe structures and their physiological applications. This work supported by NIH NCRR P41-RR09754-02, NIH NINDS N01-NS-4-2303, NIH NINDS N01 -NS-2-2379.

Int reduction: The use of silicon based device~ as active and passive implantable Systems has increased the need for biocompafible insulating materials. One recent application is silicon based mieroelectrode arrays that contain active integrated circuitry. These devices must be insulated with a material that can be easily applied, have long term adhesion to the silicon substrate, provide an absolute ion barrier, and be biocompatible. We have tested low stress, PECVD silicon nitride as a candldate matedal for short term applications and found it to perform adequately in all areas examined. Methods: We have built passive micreeleetrode arrays and active silicon VLSI circuits and insulated them with low-stress silicon nitride deposited using a dual-frequency (13.56 and 0A87 MHz) PECVD system. We have evalusted the insulating properties of theso coatings in vitro and are co~renfly testing them in feline cortical llssocs for a potiod of 3 months. Results: The adhesion of the silicon nRridr was tested on planar test structures soaked for up to 3 months in 75C saline. An adhesion test showed 0 =lLq-offs" in 10,283 samples. Unbalsed resistivityofthesefilmswas 10^14 ohm-cm. While most activecircoits fui|oddtm to the silastlc potting over the bond pads, microscopic examinations oftbe silicon nitride coatings showed no evidence of failure. 3 silicon nitride coated, silicon microelectrode arrays ~r implanted in the cortex of 3 cats for a period of 3 months. Cortical histology will be performed when the animals arc sacrificed. Conclusions: Low stress silicon nitride can be used as a biocompatible insulating material, at least over the short term that these experiments were performed. We suggest that it can be used in acute biological applications as well as in short term, chronic applications.

403 ACTIVATED SPUTTERED IRIDIUM FOR NEURAL STIMULATION AND RECORDING. L. S. Robblee, R.B. Jones, G.S. Jones, and S.F. Cogan. EIC Laboratories,Inc.

BIORESISTANT MICROELECTRONICS [:OR IMPLANTABLE APPLICATIONS D.J. Edell, LP. Devaney, E.F. Gleason, K.K. Gleason, S.J.H. Limb, J.R. Mann, T. Stems C.V. Thompson, C.M. Vanaria, M.S. Wrighton (MIT, Wext Roxbury VA, NIH NOt-NS2301

Thin film electrodes of sputtered lr manufactured by silicon wafer technology are increasingly used for neural stimulation and recording of neural signals. Sputtered Ir is activated by repetitive poteatiodynarnin cycling in the same manner as is bulk lr metal to form a mukilayered oxide. The maltilayered Ir oxide film has advantages over bare lr for recording and stimulation: 1) it has a lower interface impedance, beneficial for power conservation and neural recording, and 2) it allows the use of higher stimulation charge densities of the order of 3 - I0 mC/cm= without the occurrence of water electrolysis or corrosion. Obtaining these properties requires precis: control of the Ir film deposition variables, the electrochemical activation protocol, the quality control test procedures, and the stimulation waveform, and may require periodically poising the oxide at its highest stable oxidation state to maintain a low interface impedance. Film deposition conditions should be such that a mildly compressive, high density, low porosity Ir film is obtained. The film thickness should be limited to <200 nm to avoid delamination, and for depositions onto polysilicon, a thin adhesion layer of 30-40 nm Ti is typically used. An effective activation protocol is a I Hz potential squarewavebetween -0. 8 V and +0.75 V vs. SCE for -300 complete cycles, ending at the positive potential. The preferred activation electrolyte is 0.3M Na=HPOo pH -9.1. The use of a standard reference electrode is necessary for control of interface potential. Testprocedures should NOT include bubble testing since that may dlsmpt the multilayerod oxide, and cathodic voltages will reduce the oxide to its lower valence state with -104 decrease in electronic conductivity. The highest stimulation charge densities arc obtained with anodic first pulses, or with cathodic pulses from an anodically biased electrode. Experimental results will be presented which illustrate these features. This work was supported by contractsfi'omNIH-NINDS.

Interconnecting and protecting integrated circuit devices that are designed tor long term implantation within biological systems requires thin encapsulation layers that are biocompatible and biorasistant. Implants should be functional for decades for useful clinical applications. Long term testing of a variety of materials is in progress. In order to define assembly procedures and materials, a variety of test devices and methods are being developed to detect subtle degradation of encapsulating materials. Devices are tested both in saline soak chambers as well as in animals. The focus of soak chamber testing is on screening likely materials and techniques to be further tested in animals as well as identification of possible temperature dependent failure modes. The focus of animal testing is on degradation of materials from the biological environment, contamination of the materials by the biological environment, and degradation of electrical properties of the materials. Observations of biocompatibility will be used to determine if the materials under test should be considered lor clinical applications as well as to identify mechanisms of material degradation. After several years of observations, silicones and fluoropolymem emerged as the most promising classes of materials yet tested. Dow Corning silicones can provide insulation of bond pad areas and metal interconnects at the level of IO~SD./E]or better sudace resistivity for at least 5 years under saline soak. PTFE Teflon protects fine coated wire with resistivities approaching the bulk material level of 101Z~-cm for at least 6 years under saline soak. Since Dupont and Dow Coming can no longer sell products for medial implant research, if was necessary to identify new suppliers or processes. While several silicone companies are willing In take over Dow Corning's market, no manufacturers of PTFE were found that were willing to provide PTFE materials. A plasma deposition technique has been under development for producing high quality PTFE like coatings using a pulsed plasma technique.

S-86

Multi-unit Recording and Analysis

407

410

HIPPOCAMPALPLACE CELLENCODINGOF THE ENVIRONMENT:INSIGHTS FROM SIMULTANEOUSRECORDINGOF MULTIPLESINGLEUNITS E.J. MarkusI, Y. Quin=,B.L. McNaughton2. and CA. Barnes2. IPsychologyDept., Univ. of Connr Storrs,CT 06269; ZNeuralSystemsMemow & Aging Div., Univ. of Arizona, Tucson, AZ 85724.

CROSS-CORRELATION AND MICROSTIMULATION: COMPLEMENTARY TOOLS IN THE EXTRACELLULARANALYSISOF SYNAPTIC INTERACTIONS H. A. Swadiowand IC Lnka~h Depa~ment of Psychology, The UniversRyof Connecticut, Storrs, CT and Departmentof Psychiatt7 and Human Behavior,Brown UniversityMedical School, Pmvidenen,RI.

O'Keefe and Destrovsky(1971) showed that Idppocampulcomplex spike ceils fire only when the animal is in a esrtain locations (place fields) in thn environment. Examiningthe factors that can.,u:a given hippecampat neuron to fire in the environmentprovidesdetails on the type of informationintegratedby that neuron. The mechanisms underlying hippocampal representation and function, however,am better revealedthrough simultaneousphysiologicalrecordings of multiple"place cells" in freelybehavingmrs. Hippocampal place cells were reoorded simultaneouslyin male rata as they searched for chocolate in two different manners in the same enviroamenL The animals wouldfirst scorch for randomly scattered food, then withoutremovingthe animal from the apparatus, the seareh pat~ra was changed by placing the food sequentiallyin 4 specificlocations. The data indicate that hippecmnpal represuntation of a given eavirenmenlis rearranged when the behavioraldemands of the task are silered. This recading of the environmentis rapid, coincides~ t h the change in the animal's bohavior,and seems to occur simultaneouslyin a/l the cells that exi'tibitaa alterstion of their place tields. The data alsu indicate that the proportion of hippocampalcells that recode the environmentis related to the degree of motor-behavioralmodificationrequired by the change in task. Thus the hippocumpal neurons encode both space as t~ellas its eotuexlual importance. In addition, the hippocampa!units seem to respond in unison, as an ensemble which reflects a distributed representation of a given envirnnment.

The utility of extracellular microstimulation as an ancillary tool to compliment cross-correlation analysis is explored (Swadlow. I. Neurophysiul., 1995). This technique involves activating the "reference" neuron with currant pulses (1-10 p.A) delivered through the recording micreelectrode and observing synaptlc spikes in the "target" neuron. Data are presented from antidromieally identified descending cortlcefugal neurons and putative inhibitory intemeurons of $1 "barrel" codex, and from simultaneously studied neurons In the topographicaily aligned 'l~arreloid" of VPM thalamus. A very strong relationship was seen between brief, statistically significant peaks in the cross
408

411

ANALYSIS OF POPULATION CODES EMBEDDED wrIHIN SIMULTANEOUSLY RECORDED NEURONAL ENSEMBLES IN B E H A V I N G AN/MAI.;S. J,K C . l ~ . ~ n * ~ D e p t of Anatomy and Neurobiology, no. L,ot Penn. ~, Mannemann U., Phila, PA 19102 and Dept. Neurobiology, Duke U., Durham, NC 27710.

BRAIN NEURONAL ACTIVITY IN BEHAVIORAL STATE CONTROL Subimal Datta

Laboratory of Neurophysiology, Departmento/Psychiatry, HarvardMedical School, 74 FenwcodRoad, Boston, MA 02115

the ~ l e l dis~ibuted wocessing (PDP) mechanisms of Ix*din it is neoassan/to record simultaneously from numbers .or neurons in functionally associated neuronal networks. V~ have develop~l and !m~emented L-=ch~lues for concun~nt recoeding of up to 128 single nsufon,s 0vougn meroelecu'o~e ~ chronically implanted at mu~ple Wo:es.~ng eves of the ~ . . o s e n s o ~ system *n awake behaving rats and monkeys. We also !mplemented mun~-neuronal" data analysis techniques appropriate for defining J~on .p~ing at the neural ensemble level. TI-P,=se involved use of mulWadate sta~stcal techniques to define and reconsltuct the infomla~on contained within the n e u ~ ensembles. Principal oomponents and factor analysis were used to .msol..~. pop~a~on vectors enco~tng diffe~'3t dimensions of information emb~:l...ded wtthin. NxN neuronal autocovariar~e marines. Similarly canonical anaJy=s was used to resolve nfoemalJonembedded in the covadances between interconnected neuronal ensembles, such as those in the ~lamus and cortex. Discdminant analysis was used to predict behavioral events based simultaneous ac~vity of the neuronal populations. Such analyses of neural p~.., ation, responses showed that s o m a t o ~ r y ceding at higher levels is higNy olstfiDUtpO, S..UO]. ~ ev.~ sJr~le .neurons code for multiple dimensions of mformatio~ purmermore, this codtng IS dynamic, apparently involving reverberetory interac~ons between ~e thalamus and codex. Thus, the overall spatial and dyfd~jjnamiccccal.p~ .e~ts of ac'~.vityin simulta..neon.ly recorded neuronal populations were r~ enecwe in predic~ng sensoey stimuh and/or subsequent behawor in these animals. (SuppoSedby grants NS23722 and ONR NO00I4-95-t- 0246 to JKC.)

Extensive studies have ascribed a role to the brainstem peribrachial (PBL) cholinergic and non-cholinergic neurons in the generation of REM sleep, ponto-geniculo-occlpital (PGO) waves, and cortical activation. Most of these data, however, derive from experiments employing methods that are indirect or nonspecific such as systemic or central d r u g administration, measurement of neurotransmitter release or metabolism, and electrical stimulation or lesioning of the brain. To test the hypothesis that the PBL is involved in the generation of REM sleep, ~ wave and cortical activation, extracellular single neuronal activity of 1006 PBL cells were carefully reviewed. All of these cells were recorded, in chronically prepared behaving cats, with a simultaneous recording of cortical EEG, thalamic field potential (e.g. PGO), EMG of neck muscles, and EOG. The discharge patterns of these PBL neurons showed a marked heterogeneity but they could be operationally divided into six categories: (1) wake and REM sleep-on; (2) wake-on; (3) REM sleep-on; (4) FGO-state-on; (5) PGO-state-off; (6) state-independent. These cells are distributed throughout the rostro-caudal extent of the PBL with a specific distribution pattern. These findings are consistent with the hypothesis that the PBL neuronal activity is important in the generation of REM sleep, PGO waves, and cortical activation.

409

412

9

In . , ~

to e l s e

]n~onnauo? processing m ~

INTERLAMINARPROCESSINGOF TACTILESPATIAL FORM IN AREA3B OF THE SOMATOSENSORY SYSTEM. Stevan Hsiao, Jim DiCarln, and Kenneth Jolu~on. The Kxieger Mind/Brain Institute and the Dept. of Neuroscience The Johns Hopkins University BaltimoreMD. 21218. Neurophysiologicalrecordingswere made fromneurons in differentlaminae in area 3b of the primary sumatosensury cortex in the awake behaving monkey (Manana mulatta). In these experiments, seven microulectredes (400 mu spacing), precuated with flourescent dyes and arranged so that the electrode tips formed a fight angle to the central sulcus, are driven down to approximately the same depth into the postceotrsi gyrus. The stimuliconsisted of emhnssed spatialpatterns (letters of the alphabet, random dots and oriented bars) that were scanned across the animal'sdistal fingerpadswhilerecordingsimultaneouslyfromthree to fiveneurons that have similarreceptivefieldcenters but are located in differentlaminae. At the end of the experiment, the exact anatomicallocationof each neuron is foundby reconstructingthe flourescenttracks lefl by each electrode penetration. "['hepreliminaxyresults indicatethat withina cortical column, informationfows from the central layers (granular) to the supra- and (perhaps) infragranularlayers. Neurons in the central layersshow responses that tend to be isomorphicto the stimulus,neurons in the superficial layer tended to be nonisomorphicbut highly structured, neurons in the deepest layers showed uniform random responses and neurons with weak structure were distributedevenly across all laminae. An analysisof correlated activitybetween simultaneouslyrecorded neurons indicutes that there is a monosynapticexcitatoWconnection between the eenesl layers and upper and lower layers. Together,these resuRs indicatethat in area 3b a significanttransformatinnof spatisi form occurs between the central and supergrnnular layers.

THALAMOCORTICAL INTERACTIONSIN THE SOMATOSENSORY SYSTEM Kevin AIIoway and Martha Johnson

Department of Neurosclence & Anatomy, MS. Hershey Medical Center, PennsylvaniaState University,Hershey, PA, 17033 Tbe vnntroposterolateral (VPL) nudcos oftbe thalamus projects directly to middle layers IIlb and IV of primary eomatosensory (SO cortex. To understand the relationship between thalamus and cortical cohinms, we simultaneously recorded thaiamic activity and neuronal responsesin diffnr~at layers of somatoscnsorycorthx.

Thalamic and cortical neurons with overlapping reenptivefields in tic hairy skin were activated by computer-controUcd air jets. Cress-correlation aaalysis ofspontancons and stimulus-evoked activity was used to characterize thalamonorticaI interactions. A total of 107 VPL and 218 SI cortical neurons were recorded in halothane anesthetized cats. There were 136 thaiantoonnical pairs available for cress-correlation analysis and 67 of these showed significant interactions in the thaiamonurficai direction. The majority of significant interactions involved layers IV and IIlb where mean efficacy was 0.047 (n = 21) and 0.037 (n = 22) respectively. Supragranular layers II and lIIa had mean efficacy values of 0.029 (n = 6) and 0.023 (n = 7), respectively. Infragranular layers V and VI had mean efficacy values of 0.029 (n = 7) and 0.019 (n = 4). The strongest functional connections betwesn thalamus and cortex appcm"to be with cortical neurons in the middle layers (liFo, IV). Thalamononical connections with supra- and infragranular neurons were equally weak and suggests that, after activating layers IV and Ll]b, sensory infom~tion is passed by parallel routes into the superficial and deep layers of neonortex.

Poster Presentations 413

416

THE EFFECT OF ACOUSTIC TRAUMA ON COCHLEAR ELECTROMOTILITY H.H. Nakajimal; E.S. Oison'; D.C. Mountainta; A.E. Hubbard t'z~. Depts. of tBiomndical Engineering; 2Electrical, Computer & Systems Engineering; 3Otoluryngology; Boston Univ. ~Physics Dept., Rutgers Univ.

TEMPORAL INTERACTION OF VERGENCE AND SACCADIC EYE MOVEMENT UNDER COMBINED STIMULUS CONDITIONS Hulmin Zhu Dept. Biomedical Enginuefing, Rutgers University, Pisoataway, NJ 08855-0909

The sansitivity of the cochlea is greatly compromised by acoustic trauma. Cochlear sensitivity is hypothesized to rely on outer hair cell (OHC) electromotility. The local mechanical response at a particular location of the cochlear partition is amplified by an active OHC feedback system resulting in sharp tuning to s specific characteristic frequency (CF). The feedback loop is hypothesized to consist of two processes in series: a mechano~ectrlc transduetion process and on electmmechanlcal transduefion process. We have proposed that the feedback produced by the OHCs is negative fur frequencies less than CF and, due to frequencydependent phase shifts ia the loop gain, the feedback becomes positive for frequencies near the CF. To study the effect of traumatizing tones on the OHC transduetion process, we measured the cochlear microphonic (CM) and the electrically-evoknd otoncoustic emission (EEOE). The CM is the extracellutar ac voltage response produced by the OHCs to tones and is a measure of the mechanoelectric transduetion process. The EEOE is the sound measured in the ear canal produced by electrically stimulating the OHCs and is a measure of the electromechanical transduetion process. For stimulus frequencies below the CF of the cochlear location under study, trauma produced an increase in the EEOE and produced a deere.~se in the CM. For stimulus frequencies near the CF, the trauma produced a decrease in the EEOE. These results support the hypothesis that the electromecbanical feedback from OHCs is negative for frequencies below CF and positive for frequencies near CF and that OHC mechanoelectric transduetion, rather than electromecbanicst transduction, is the process that is more vulnerable to acoustic trauma.

Puroose. The combined vergonce and saccadic stimuli were used to investigate the temporal interaction of vergance and sacoadlc eye movement systems. Latencies of these two components were compared to those of pure responses. Method. Binoculur eye movements were recorded for pure sacrade of 2 to 6 deg, pure vergonco of 2 to 16 deg and saccade of I to 6 deg combined with convergence or divergence of 2 to 10 deg, with sacoadic stimulus delayed by 0 to 300 msue in five subjects. An imprmed data processing method was used to properly isolate the vergeuee and saccadic onmpouents. The latencies of vergeuee and saccadic components for different stimulus conditions were detected and plotted. Remits. The latency of vergeuee is not affected by its interaction with sacoade, whereas the latency of saccede can be delayed when combined with ~ergonon response. The saccadic latency approaches that of pure saccade and its variation decreases when vergonce and saccade become more distinctly separated in time. Conclusion. The progranuning of vergenco is not affected by the on-going saccadic stimulus. The saccadic control output can be delayed due to either prolonged or re-triggered programming even though the combined saccade generated has the same dynamic properties us the pure saccade. Since programmings of vergence and saccadic systems are relatively independent, the interaction of vergonce and saccade may occur in the common neural pathways for both eyes after their primary control commands have been generated.

414

417

THE EFFECTS OF STIMULUS DUKATION ON THE KESPONSE PROPERTIES OF DORSAL COCHLEAR NUCLEUS (DCN) UNITS IN ANESTHETIZED MONGOLIAN GERBILS. lrlna Sigsiovsky, Kenneth E. Hanoock ~A ~ F. Voigt Depadncnt of Biomedical Engineering, Boston University, Breton, MA 02215

MULTICHANNEL INTRANEURAL ELECTRICAL STIMULATION FOR PROSTHETIC SENSORY FEEDBACK

The purpose of this study was to exam~ the effeO.~ of stlmnius din'aden on DCN unit respa~e pa)perfies. 'fhe~.arc two comm,m ways to classify the physiological responses of sinsle unita-periosthnntm time histograms (PSTH) and response maps (PJd). There are four umjor classesof PgTHs ob~n~edin DCN: psuser,chopper,buildup and onset R.Ms,on the otherhand, plot regions of unit excitation and inhibition as a function of fiequeuey and sound preesm'e level (SPL). Responses to 500 ms tones of various fiequmcies and SP~ were recorded from 19 DCH units in anesthetized gerbils (Meriones ungulculatu.v) using pisfinum-lrld;um microelestrndes. RMs were constructed using three analysis rate windows (50 ms, 200 ms and 500 ms) for estimating driven rates. In general, unit threshold and ruing rate estimates of buildup (n'=6), peu.unr (n=8). choppsr (of3) and onset (nffi2)unite depend on sthnulus duration to various degrees while best fie.queucy (frequency to which unit is the most sensitive) estimates are minimally affected. Increasing stimulus dm'ation from 50 ms to 200 ms resulta in l) decreasing threshold estimates for buildup unite (6/6) and 2) decreasingmaximura firing rate for chopper units (3/3). Increasing sSmutus duration had ~ o u s effects on pauser units;threshold estimates increased (1/8), (3/8) of stay the same (4/8). Like,rise maxhuum firing rate estlmates increased (3/8). decreased (2/8) or remainnd the same (3/8). The sizes of these changes were generally less than ~ seen for buildup and choppsr tmlts. Finally, the changes in threshold and firing rate estimates for pauser and buildup uni~ occur pt~)~ity within the first 200 ms of the stimulm while chopper unit respome property estimatescontinueto changeafter 200 ms. [work supportedby H]H and the Whiteker Foundation]

415

LOW-POWER, LOW-NOISE ELECTRONICS FOR NEURAL PROSTHETICS Bruce C. Larson, David J. Edell, James R. Mann, and Antonio M Soares West Roxbury VAMC, and the MIT division of Health Sciences and Technology, depadment of Electrical Engineedng and Computer Science, and Lincoln Laboratory The need for an improved control interface for functional electrical stimulation, prosthetic, and assistive devices for the spinal cord injured has prompted research toward developing the technology behind a direct neuroelectric interlace. A system that could chronically sense many independent channels of neural information could provide an abundant source of motor control information if placed in the appropriate areas of the nervous system. Since a neumaiectric interface based on a microelectrode array implanted in the human brain would probably have to be wireless (fixed only to the brain) for mechanical stability, if would require its own amplifiers, multiplexers, signal transmitter, and power supply. Because power supplies of sufficiently small size and mass will be limited in their supply capabilities, the implant's electronics must be of low power design. With limited power, if becomes difficult to detect the small extracellular neural potentials that are anticipated. Hence a thorough characterization of the available circuit elements is necessary to optimize the electronics for the application. A preliminary study of the current-voifage relationships and noise characteristics of various transistors fabricated on a commercial low-noise analog N-well CMOS process is presented in this paper, and low-noise, low-power circuit design guidelines are developed. For applications where high-speed electronics are not required (such as the neural infom~tion sensor which prompted this study), subthreshold transistor operation has distinct advantages in terms of minimizing noise and maximizing signal amplification per unit power. A method for designing circuits which will meet noise performance specifications and operate in subthreshold is presented as a means to optimize power consumption.

Daniel DiLorenzo 1,2 David Edell1,3, Ron Riso4, Mark Koris5, Lisa Devaney3 1Harvard_MiT Division of Health Sciences and Technology, 2MIT Department of Mechanical Engineering, 3MIT Lincoln Laboratories, 4Case Western Reserve University, 5West Roxbury VA Medical Center The functionality of prosthetic limbs is restricted by the limited availability of sensory feedback. This research aims to develop a technology to allow the presentation of muitichannel sensory information directly to the sensory afferent neurons of the transected peripheral nerve in the stump of the amputee. Intraneural implants of several designs were developed and implanted in the sciatic nerve of rabbit animal models and monitored for chronic functionality. Both neurephysibtogical and behavioral techniques were used to detect activation of the implanted nerve. Cortical evoked potentials in the somatosensory cortex were monitored via chronically implored epiudurai electrodes. Behavioral conditioning consisted of delivering an electrical stimulus to the implanted nerve followed by an airpuff to the eye and the resultant eye-blink reflex. After several training sessions, the rabbit assocaited the two stimuli such that electrical stimulation of the sciatic nerve, in the absence of an airpuff, elicited an eye-blink. Animal studies to date demonstrate single channel electrode tunctionality of up to 129 days after implantation. Development of evaluation techniques for the discrimination of multichannel information are under way.

S-87

S-88

Nonlinear Systems Analysis in Biomedical Engineering

418

421

N O N L I N E A R D Y N A M I C MODELS O F T H E PHOTORECEPTOR-LMC SYNAPSE IN T H E FLY C O M P O U N D EYE A.S. French, M. Juusola, M. Weckstr6m, R. Uusitalo and M.J. Korenberg. Department of Physiology and Biophysics, Dalbousie University, Halifax, Canada, Department of Physiology, University of Oulu, Finland, and Department of Electrical Engineering, Queen's University, Kingston, Canada.

PARAMETRIC NONUNEAR SYSTEM IDENTiFiCATION, STRUCTURE DETERMINATtON AND MODEL VALIDATION: APPLICATIONS TO LUNG MECHANICS.

Membrane potential responses to randomly modulated light stimuli were used to characterize the fwst synapse in the fly compound eye. The parallel cascade method was used to obtain zero-, f'wst- and second-order Volterra kernels between input light fluctuations at eight different levels of light adaptation, and responses in photoreceptors and large monopolar neurons (LMCs). Photoreeeptor responses were approximately linear functions of the input, but LMC responses were clearly nonlinear. Synaptic input-output relationships were obtained by passing the measured light stimuli to LMCs through the linear photoreeeptor characteristics, to estimate the synaptic input. The resulting synaptic functions were well approximated by second-order Volterra series. They could not be modelled by a linear dynamic-nonlinear static-linear dynamic cascade, but were well-fitted by a nonlinear static-linear dynamicnonlinear static cascade. The two models were both used to predict the synaptic response to brief pulsatile changes in receptor membrane potential. Both indicated that synaptic gain is reduced by light adaptation.

419

B6la Suld, Qin Zhang, Huichin Yuan, and Kenneth R. Lutchen Department of Biomedical Engineedng, Boston University, Boston, MA 02215. Dudng lung constdddon there is an increase in beth the frequency and tidalvolume (VT) dependence of lung Impedance. This suggests alterations In the mechanisms contdbutiog to beth the linear and nonlinear charactedstica of the lung which were exsrnined utilizing the theory of nonlinear block structured systems. To determine the best lung model we developed 9 2-etep parametric nonlinear system Identitication and structure determination approach. In step 1, using a global optimization procedure, we simultaneously fit input-output data at two input levels with vadous combln~ons of linear and nonlinear blocks. In step 2, we validate the best model by predicting independently measured data. This epproach was applied to time domain pressure-flow pseudorandom data acquired in dogs before and alter histamine Induced conedrlction. The best lung model was a inear airway compedment combined with a W]eser structure for the lung tissues. In control and at late response, a =Ingle airway compartment was sufficient. At peak response, a two-cornparfment ain#ay structure was nscassaw to acoount for the bronchoconstdc~on Induced inhomogenaity. The model excallantty predicted data even at VTs that were outside of the range used In the Identification. We found that histamine caused much larger changes in the linear ~ssue parameters than in the nonlinear coeffidents. Thus, the prirnaqf cause for the increased VT-dependenca dudog constri~on is that existing but unaltered nonlinear mechanisms become amplified through an increase in the magnitude of the linear tissue impedance. Finally, while this approach does not result in direct kernel Identiticetion, It is relatively simple, powerful even when only shod data recordings are available end requires no spedflc properUes of the input signal used. (NSF Bcs-g309426 and NIH HL50515).

422

NONLINEAR

SYSTEM

IDENTIFICATION USING HIGHLY COLOURED INPUTS D.T. Westwick and R.E. Kearney Department of Biomedical Engineering, McGiU University, Montrdal, Canada

FEEDBACK MODELS FOR NONLINEAR TUNING OF AUDITORY NERVE FIBERS H. Carney and Micbele Friedman Department of Biomedieai Engineering, Boston University

Many experimental paradigms preclude the use of the white, or nearly-white, stimuli required by many nonlinear identificationtechniques. In this paper, we present an optimization of the parallelcascade method which allows a wide variety of input signals to be used. The parallelcascade method models an unknown nonlinear system as a sum of Wiener systems, each consistingof a linear dynamic system followed by memoryless nonlinearity. The method proceeds iteratively. At each stage, the difference between the measured output end the output of the current model, the sum of all previously identified Wiener cascades, is computed and a Wiener system is fittedbetween the input and these residuals. The accuracy and robustness of the parallelcascade method depend on the techniques used to estimate the dynamics of the linear elements. The linearelement of the firstpath can be estimated from the first-orderinput-output cross-correlation,leaving residualswhose first-ordercorrelation with the input is zero. Subsequent paths must therefore be derived from higher-order correlations. Here we address the problem of determining the optimal Wiener system given the information in the second-order cross-correl&tion.W e will show that its IRF is eqnal to the principalgeneralized eigenvector of a ms.trix pencil comprising the second-order correlationand a symmetric Toeplitz matrix whose firstrow isequal to the input auto-correlationfunction. This eigenvalue solution accomplishes the deconvolution, commonly used to estimate the impulse responses of linear systems, implicitlyand results in a bank of filterswhose outputs are exactly orthogonal when driven by the test input. W e will also demonstrate that this method can be extended to allow the estimation of IRFs from higher-order correlation functions without computing them explicitly,thus reducing the number of computations required to model high-order nonlinear systems.

Nonlinearities in file andi~P/periphcP/influence the response Im3pertles of all auditory nerve (AN) fthr AN tuning is chmacterized by an increase in bandwidth and, for most mid- to hlgh-frequency fibers, a downward shift in the center ~equeacy as soend level incseases. Changes in bandwidth with level are nccoml~nied by changes in the phase pmpe~ies of the t~spoases. The phase affects the temporal res~nse pmpe=tles, which may play a sole in the encoding of positions of sound sources, as well as in eueodin8 complex low-frequency sounds. This study focussed on the properties of a family of nonlinear feedback models, All of the models were able to simulate nonlinear changes in bandwidth and phase, bet ditIeaexlin their abdity to simetate changes in peak frequency. The f o r w ~ path of each model consisted of a linear 2nd-ordor resonance, repmsantin8 the pes~ve mnchamcaltuning of the inner nsr Each feedback path contains a saturating nonlinearity, plus othe~ linear Ist- anA 2nd-order components. We explored the effects of varying the sequence of the nonlinear and linear elements in the feedback path, and of varying the cutoff f~equancies of lsi-order low-pass filters and the nateral frequencies of 2nd-order resonaut filters in the feedback paths. It was possible with some of the model configurations to simulate not only the noutinem"changes in bandwidth and phase, bet also varying degrees of shift in the pesk heqncncy. A chalieage for phenomenological models of auditory tuning, such as those explored Item, is to 13eable to model significant shifts in peak fi~luency for some fibers, and liUle or no shift for others. Because this peak frequency dominates the periodicity of discharges in responses to complsx sounds, and because loss of the nonlinearity is associated with impairment of hearing, it is important to establish a better understanding of this nonlinear feature of AN responses. (Supported by NIH R29-DO)I641 and a Young lnvesdgater Award from The Whitske~"Foundation.)

420

423

NONLINEAR MODELING OF PHYSIOLOGICAL SYSTEMS FROM STIMULUS-RESPONSE DATA Marmarelis, V.Z. Biomedical & Electrtcel Engineering, University of Southern Califonda

A NONLINEAR SYSTEMS ANALYTIC APPROACH TO MODELING BIOLOGICAL NEURAL N E'I3NORKS T.W. Berger tt, B.J. Sheu=, V.Z. Marmarelist, and R.J. SclabasstI Depts. of rBiomedical Engineering and =Electrical Engineering, and SPmgmm in Neumscience, Univ. of Southern California, Los Angeles, CA 90089; IDepte. of Neumlogicel Surgery and Electrical Eng., Univ. of Pittsburgh, Pittsb. PA 15260

Nonlinear analysis of physiological systmxasis attracting increasing attention with the rapidly cme~ng realizatton that nonlh~earities are esee~ttal in subserving phy'Mological hmclion. A novel approach is presented for practicable nonlinear analysis Of physiological systems by means ol mathematical models obtained from stimutas-msponse data. This approach is based on the use ot a small number of properly chosen wavdorms (termed the "prindpal dynamic modes ~) that yield a parsimonious ~ t a t i o n of the system dynamic nonlinearities. The theoretical foundation of this approach and practical methods f ~ selecting the "prindpal dynamic modes", as well as estimating the requisite parameters, are discussed. The advocated methodology appears to offar a practicable solution to a fundamental problem of biomedical science and engineering that has long been viewed as formidable for low-order nonlinearitiesand as nearly intractable in the case of high-

order nonlinearities. The presence of high-order nonlinearities is welt-established In the case of n ~ e~oding m ~ (action potential generation) and a variant of this approach has been developed to address this particular class of problems in a most efficadous mann~. Likewise, this approach has been extended to the case of systems with tx~nt-process inputs (e.g., action potentials). In addition to neuronal systems, this methodology has been tested with broadband data h'om the renal system (blood pressure and flow). The application ol this approach to actoal physiological systems willbe discussed and preliminary resultswill be presented. It is hoped that the advent of this methodology will facilitatethe broader use of nonlinear analysis in physiological studies.

A combined theoretical and expedmentai strategy for developing biologically constrained models of real neural networks will be outlined. The fundamental pdnciples of the strategy will be illustrated through their application to study of the hippocampus, a brain essential for the cognitive processes of learning end memory. The approach involves developing a 'nonparametdc' model of the hippocampus in which the nonlinear response characteristics of its principal neurons are characterized experimentally using random impulse train stimulation. Nonlinearities in the input/output relation are represented as the kemels of a functional power series. The experimentally determined kernel functions provide the basis for a model of the global system represented as the composite of the input/output functions of its intnnsic elements. Specialized biological preparations allow the principal neurons to be studied in the context of different network cimuitries. Using multi-dimensional z-transforms, a procedure is demonstrated for deriving kernel functions for interneurons that are not directly observable. A scheme is proposed for realizing a model of the hippocampus using both observed and derived kernel functions, and a hardware implementation based on a novel analog VLSI design will be presented. This research is funded by the NIMH, the NIH Biomedical Simulations Resource, and the Office of Naval Research.

Biomedical Engineering for Space Research 424

427

CARDIOVASCULAR MEASUREMENTS IN SPACE--BIOENGINEERING CHALLENGES. Jay C. Buckey, Jr., M.D. Dartmouth-Hitchcock Medical Center, L/ebenon, New Hampshire

TH~ ~ OF PROLONGED WEIGHTLESSNESS ON THE VESTIBULeOCULAR REFLEX OF A~TRONAUTS Charles M. Omen, Direclor, Man Vehicle Laboratory, Det~tment of Aemnantlcs and Asa'onantics, M a s s a c h ~ Insdtote of Teehselogy, Cembfidgn, MA 02139

Significant cardiovascular changes ~c.cut in space=. Unique and u n c x ~ alterations in intravz.~ealar pressures lake place, and the adaptation to microgravity produces ortheststic intoleranes upon re-entering Earth's gravity. Making accurate and reliable meastttements in the operational environment of the Spare Shuttle, however, presents tecfiaieal challenges. Equipment must meet stringent NASA noise, electromagnetic int~dereaee, thermal md vibration standards, Appropriate materials must be chosen that do not offgas undesirable volatiles. Invasive instrumentation must possess two-failure tolerance in any safety critical pathway, i.e. two independent failures can occur without compromising safety. The Spaeslab Life Sciences-I and Spaeslab Life Sciences-2 flights were dadieatad biomedical r~estch missions, and allowed for a comprehensive set of cardiovascular measurements to be made. Measurements of central venous pressure(invasive), blood pressure, hem rate, esrdine output, leg flow, calf compliance, heart size and maximal oxygen uptake were made. Central venous pressure (CVP) was measured continuously from IG to 0G using a specially.designed, ambulatory system, A 4 fr, polyurethane catheter was inserted the night before launch and remained in place as long as the s~coad inflight day, Equipment based on an optical shaft encoder was also built to make the leg flow and compliance measurements using the venous occlusion plethysmography technique. A three-dimensional approach to cardiac volume by echocardingraphy was developed and flown. Grnond-basad measurements of cardiac output parformed ~on-invasivaly using azetylene rebreathing. Reliable measurements of all the key parameters were made successfully in the Shuttle environment. The significant findings were unexpected decrease in central venous pressure upon entering microgravity and inad~oate vasoconstrictor responses in those who had orthostatic intolerance.

INTRODUCTION: R e ~ n t Spacoinb studies ((}man & Balkwill, 1993; Omno and Caikias, 1993) suggested adapdve mtom of central velocity storage in human angular vrstibulo-orul~ re~e~t during prolonged orbital flight. On a more recent Shuttle mission, we at~di~l changes in human VOR in weightlessness in both orbitaland pa~bo~ic flight. and both with and without head tilt.METHODS: Horizontal EOG was recorded in 4 subj~ds dining and afterone minum of 120 dug/see constant rotation in eepear testing conducted before, during, and alterthe SLS-2 Sl~ceinb mission (Oct "93). and also. for the faSt time, in parabolic flight. SFV was calculated via Order Statistic filtering, and an iterative exponential model fitting method was used to detext VOR "dropoots" and to obtain per" and post-rotatory VOR gains and time constants(l}. RESULTS: Two ~aabjecta showed signific~t T reductions in parabolic flight, but when tested after 4-10 days in orbit, T had returned to equal or ~estar than pre-llight, indicating an adaptive incre..a~of vestibules vai~r storage. The T increase carried over to ~.rly postfiight testing. In orbit. T and duration of subjective rotation sensation was apparently unaffecr~l by a 90 dug. head pitch forward alterchair stop. For the other two subjects. head erect T was reduced (60%. 73%) in orbit compared to pre-llight, indicating a persistingloss of vestibolarvelocity storage.Head pitch T and subjective data suggested possible haptio dumping. Early post flight per rotatory T was significantlyshorter than pre-fligfit.For allsubjects testedpost flight,fiendpitch sborennd T relativeto head erect. as in pre-fllght testing. Supportedby NASA Contact NAS9-16523.

425

428

EFFECTS OF SPACEFLIGHT ON STATIC STRENGTH, ENDURANCE, AND EMG INDICES OF FATIGUE S,H. Roy, C.J. De Lute, M.S. Emiey, A. Rodrigues, E.J. Kupa, L.R Young 9 and D.M, Merfeld*; NeuroMuscular Research Center, Boston University, Boston, MA, 02215; 9 Man-Vehicle Laboratory, M.I.T., Cambridge, MA, 02139

EFFECT OF WEIGHTLESSNESS ON LINEAR ACCELERATION RESPONSES. Daniel M , M e r f e l d R,S. D e w Neurological Sciences Institute, L e g a c y Good S a m a ~ t a a Hospital, Portland, OR 97210-

PURPOSE: This study assessed whether muscle strength, endurance, mad surface EMG measurements of fatigue in the lower leg were changed following exposure to microgravity. METHOD: N=4 astronauts assigned to a space shuttle mission (STS-58) were tested during baseline sessions conducted at 130, 90, 60, and 25 days prior to launch and again post-flight at I, 2, 7, end 9 days after return to l-g. Two control subjects (alternate crew members) were also tested pre- and post-flight. Subjects performed sustained isometric contractions at 80*.4 of maximum (MVC) for ankle dorsiflexion and plantar flexion. EMG signals were recorded from the solens, medial gastrecnemius, and tibialis anterior muscles. RESULTS: Force measurements from the load ceils indicated a significant (9<0.05) decrease in MVC (3"/~18%) for dorsiflexion which did not improve during the post-flight test period, MVC for plaatarftexinn was also initially reduced post-flight (4%-12%) but recovered to baseline within 4-7 days. E o d ~ v r or the ability to sustain the 80% MVC contraction, was signifi~=mfly reduced for dorsiflexion but not plantari2exion. No significant changes in the EMG parameters occurred post-ftighl, however there was a trend towards increased signal amplitude mad more rapid decay uf meth~ frequency for the tibiafis anterior muscle. CONCLUSION: Strength and endurance changes post-flight were more pronounced for ankle dorsiflexion than plantadlexion. Changes in EMG signal measurements post-flight were not as consistent as that of force measurements, although the qualitative differences for the EMG parameters from the t/blaHs atnterior were consistent with increased fatigue.

It has been hypothesized that h u m a n s m i g h t b e c o m e less sensitive to gravitational "tilt" cues and m o r e sensitive to l i n e a r acc.eleration "Wanslatlon" cues as part o f an a d a p d v e response to the m i c r o - g r a v i t y e n v i r o n m e n t experienced during spaceflight. T o m e a s u r e any such adaptive changes, w e mea.sured the ability o f f o u r astronauts to perform m a n u a l control tasks ha which the subjects w e r e a s k e d to null their self-motion ha the p r e s e n c e o f a disturbance before and after the fourteen day S p a c e l e b Life S c i e n c e s - 2 (SLS2) mission. In one test, the subjects w e r e asked to null roll flit, w h e r e the experimental g o a l w a s to m e a s u r e decrements in the ability o f the subjects to sense liltrelative to gravity postflighL In the second test, the subjects were ~ked to null translations; hem the goal was to measur~ enhancements of ability to sense lirtear translation. In both test paradigms, the subjects were ~ested in the D a r k a n d w i t h a control condition w h l c h r u l e d out fatigue o r other n a u r u - m u s c u l a r influences. Both subjects tesIed on the l a n d i n g day exhibited sigflifieant d e c r e m e n t s ( p < 0 . 0 5 ) in their ability to control roll flit in the dark. T w o o f three subjects exhibited an e n h a n c e d ability to control their linear translation in the d a r k postflighL N o significant changes w e r e e v e r observed ha the control condition. T h e s e results appear to indicate that there is an a d a p t i v e c h a n g e in the w a y the nervous system haterprets flit a n d translation c u e s f o l I o w i n g e x t e n d e d e x p o s u r e to micro-gravity.

426

429

HUMAN POSTURE MODELING: AN ASSESSMENT OF ALTERED GRAVITY ADAPTATION, Dava J. Newman, C,S. Draper Assistant Professor of Aeronautics and Astronautics, Massachusetts Institute of Technology, Cambridge, MA 02139.

THE LUNG IN SPACE: PULMONARY FUNCTION RESULTS FROM SPACELAB G. Kkn Prlsk, Ann R. Ellintt end John B. West Oepad~neot of Medicine, Universityof Callfomin San Diego, La Julia, CA 92093-0931

INTRODUCTION, Human mechanisms for control of posture and motion ate normally optimized to perform in Earth's one gravity (I G) environment. The task of standing upright involves a complex and not fully understood seesorimotor cnottol system. The basic analogy is that of cnoLmlling an unstable, inverted pendulum by maintaining the center of pressure (COP) within the bounds of stability (the feet) at all times. It is hypothesized that muscular gains are lowered during reduced gravity exposure causing a feeling of heavy legs and then readaptation upon mtum to a 1 G envirrmment. Posturography measurements and an estimator based optimal control model capture the important spatial and temporal characteristics of human posture during quiet standing after exposure to reduced gravity simulation. METHODS. Ten healthy subjects were exposed to 6 minute Martian gravity (3/8 G) running (3 m/s) simulations ",hat proved to be effective in inducing the heavy legs phenomenon. Subjects performed one of t,lv~e recovery-aiding exercises, no exercise (None), a series of five knee bends (Knee), or a series of three broad jumps (Jump) and then completed 6 post expoaure posturography trials. Spatial vasiables include the mot mean square (rms) position of the CoP in the x and y axes. Temporal analysis yields the diffusion coefficiem measuring stochastic activity, and the correlation coefficient measuring trends in the trajectory of the center of pressure. RESULTS. Root mean square values increase significantly (p<0.01) in both axes following simulated reduced gravity exposure. Temporal analysis reveals that human posture can be divided into a short term and a long term period. The short term exhibits higher stochastic activity and persistent trends, while the long term period shows relatively low stochastic activity and anti-persistent trends. The estimator based model succeeds in replicating temporal eharacw.risties of human posture and spatial characteristics by lowering the variable muscle gain in the model, validating the theory that muscular gains ate lowered during low gravity exposure.

Gravity !nd.ucesgradients in alveolar size, venglabon, perfasinn, venlilabonperfuslen ratio (V~G~and gas tochango in the normal updght human lung. We performed the find ever comlXebanslve studies of pulmonary funceon In mlcrogravgy (,~G) during the Spaselab Life Sciences flights SLS-1 and SLS-2. Our subjects were the crew of the missions who performed a battery of pulmonary function tests on themselves several gmesdudng each flight. Data were also collected before and alter spaceflight In both the standing and supine pe~ons. Compared to stand;us In 1G, ~G caused a reduddon in the topographic distdbu~on of venglabon and perfusion measured dudng vital capacity bresths, although significant re~lual inhomogeoeliy remained. Similarly, changes in the pulmonary capillary blood volume and membrane diffusing capacity were cor~s/stentwith much more uniform pulmonary capil~arf filling. There was also e reduc~on In residual volume, probably due to the mere uniform dlstdbo~n In alveolar size at low long volumes in/~G. However dudng gda.Ibreathing, there was no measurable reduddon in the Inhomngeneity of ventilabon or Vt/Q. Data from Inert gas washin tests with gases of widely diffedng diffusivffy suggest that the removal of gravity results in alterations In the conformaUon of the a~nl, although the mecheobm for this is at present unclear. Taken as a whole, these data suggest that while grmdty p~aysa large role in the Inhomngeoeity of lung function seen In 1G. factom Intrinsic to the lung also ~ay 9 signi'hcantrote. (Supported by NASA contract NASg-16037.)

S-89

S-90

Engineering

Applications

in vitro

Diagnostics

430

432

LIFETIME-BASED SENSING IN SCATTERING MEDIA Christina L. Hutchinson, Tamers L. Troy, Dilip Y. Palthankar, Eva Sevick-Muraca School of ChemicalEngineering,purdue University,West Lafayette, IN 47907

ENGINEERINGRESEht~ IN IN-VITROD I d i e S Patricia ~. Conway

The developmentof stable and non-toxic near-infrared (NIP,) probes or compounds together withthe recent advences in laser diode sources end detsotio~technnlngiespmmisa nnw invasive, diagnostic tissue spectroscopy. However, studies show that the determination of biochemical species fi'omre-emitted tissue fluorescence spectra require deeonvolutionof tissue scattering and absorption for quantitativetissue spectroscopy. Since optical properties can be expected to vary from person to person as well as with pathophysiology, nxU'anrdinary difficulties may exist when attempting to extract information from fluorescent or phnsphorescent intensitymessurements. In this presentation, we developa computation end experimental approach to combine non*invasive,frequency-domain mensurements of nptical probe lifetimestogether with a simple algorithmdescribingexcitation end emissionphoton migratinnwithintissues. The goal of this wnrk is to uncoverthe kineticsnf phosphnrescent end fluorescentre-emissionsfrom deep within tissues for determination of local metabolic conditions using the Stem-Voimer relationship (among others) to describe quench. To date, nur results show that optical probes with longlived activatedstates ( > 50 ns) cannot providebiochemicalinformationbeyond the sub-surface tissue layers if systemically administered. However, decahvnlutinn nf probe lifetime for comparable excitation end funresoeot photon "time-nf-flights"is possible from frequencydomain measurements of phase-shift. This is possible even if the fluorescent compound or probe is inhomogeneouslydistributed. These results point to the developmentof lifetimesensitive compounds with NIR excitation and emissionspectra for non-invasivespecUoscopy and for functionaloptical imaging. Supported in part by the WhitakerFoundationand the Natinnal Science Foundation.

In-Vitro Diagnostics (IVD) are the analysesof blood, other body fluids, and tissues to aid in the diagnosis of medical conditions, lhe applications of IVD in health care are broad, with IVD testing perfonmd in locations as diverse as hospitals, blood banks, conmercial laboratories, physician's offices, and at the homeof the patient. Cantral to the applications of IVD are the technologies utilized for the delivery of a diagnostic result. Engineering plays a significant role in the implementationof these tedmologies, especially in device design, autcmation and manufacturing.

Abbott Laboratories, Diagnostic Products Division

The future of IVI) testing is nowdefining e~gineering researchchallenges for today. Trendsin diagnostic testing are movingtoward identification of nucleic acids, making use of infermation from the humangencme, and detection of fewer than ten molecules rather than thousands. Coupledwith these testing trends are economicand market factors ~ich d~rand reduction in the total cost per IVI) test, minimization of hands-ontime required to perform IVD tests, and an increase in the amountof diagnostic information provided by a test result. Tnis presentation will providean overviewof engineering research in In-Vitro Diagnostics with an ~mphasis on future trends.

431

433

ADVANCES IN MINIATURIZINGFLUID CONTROL IN MEDICAL DIAGNOSTIC INSTRUMENTATION A. Alcock, G. Clark, TM Huang, S. Kantor, T. Nemcek, J. Noflie, J. Pan, S. SU'oupe,D. VarLee, F. Waisworth, ST Wong. Abbott LaboratoriesDiagnnsticsDivision

MODEL OF AN IMMOBILIZED ENZYME CONDUCTIMETRIC UREA BIOSENSOE N o r m a n F . S h e p p a r d ~ l r . e , D a v i d J. Mearea, A n t h o n y G n i s e p p i - E l i eb aThe J o h n s H o p k i n s U n i v e r s i t y , B a l t i m o r e , M D 21218, a n d bAAIA B T E C H , 1273 Q u a r r y C o m m o n s D r i v e , Yardiey, P A 19067, U S A

Most clinical diagnostic instrument systems perform some combination of metering, diluting, manipulating,mixing, incubating,end detecting of reaction results on a patient sample and reagent fluids. A tmdifienalapproach has been to utilize sophisticatedrobotics and plastic disposable incubationcuvettes to accomplish these fuid handling processes. Yet, cost pressures, increases in assay complexityand increases in the volume end variety of assays performed has created a need to perform these functions more efficiently and minimizethe solid waste geocmted. The DiagnosticsDivisionnf Abbott Laboratorieshas developeda versatile technology which significantly mlniatu:izes end integrates routine fluid processing functions of a clinicaldiagnosticinsu'urnentinto solidblocks of acrylic - fluid ctrcult technology. A fluid circuit can be thought nfas the fluid analogto a printed circuit. We present the results era project to demonstrate that fluid circuit technology enables an instrument system to conduct multiplesimultaneousfluid handlingoperations with no sophisticatedrobotics and no solid waste generation. Designgoals were established that would enable us to benchmark the performance of the resulting system against an existing instrument. We fi~ther chose to demonstrate the modular buildingblock capabilitiesof the technology and in so doing, significantlyreduce the prototypingtime. The design of this system, critical evaluationsof the fluidcircuit modulesand assay results will be presented.

A b s t r a c t : A m o d e l for p r e d i c t i n g t h e r e s p o n s e o f a c o n d u c t i m e t r i c w:ea biosensor was developed and validated experimentally. The b i o s e n s o r u n d e r c o n s i d e r a t i o n i s f o r m e d by i m m o b i l i z i n g t h e e n z y m e u r e a s e o n t o t h e surface o f a p l a n a r i n t e r d i g i t a t e d electrode array. T h e enzymatic hydrolysis of urea produces ionic products, such as ammonium and bicarbonate ions, which increase the electrical c o n d u c t i v i t y o f t h e s o l u t i o n pro]ritual t o t h e electrode a r r a y . T h e m o d e l combines a n a n a l y s i s of t h e d i f f u s i v e t r a n s p o r t a n d e n z y m a t i c hydrolysis of urea in the vicinity of the biosensor surface with an e l e c t r i c f i e l d s m o d e l for c a l c u l a t i n g i n t e r e l e c t r o d e i m p e d a n c e . To v a l i d a t e t h e model, u r e a b i o s e n s o r s w e r e c o n s t r u c t e d b y i m m o b i l i z i n g urease to the interdigit space of microfabricated interdigitated electrodes. T h e responses of these sensors w e r e investigated in urea solutions prepared in deionized water, at concentrations ranging from 10 ~tM to 5 raM. Using reasonable estimates for the parameters, the predictions of the model were in good agreement with the experimental data over the entire range of concentrations.

433a CO-OXIMETRY MULTI-VARIATE

- QUANTITATIVE ANALYSIS

MEASUREMENT OF HEMOGLOBINS

USING

Ronald S. Seharlack Ciba Cornlng Diagnontics Corp. Measurement of the four principal hemoglobin derivatlven in routinely performed in the clinical laboratory using absorhance spectroscopy coupled with multl-variate analysis. Thln measurement in complicated by the presence of dlnhemoglobins, lipldn and other interfering absorbern an well an pH effects on the spectra. Achieving the accuracy required for clinical determinationn is further complicated by a wide dynamic measurement range and critical nennltivity to wavelength accuracy. Ciba Corning has introduced innovated approaches to improving the accuracy and preclnlon of thin sensitive measurement.

Nonlinear Dynamics in Physiological Systems I 434 DISCRIMINATING 11ARMONIC FROM NON'HARMONIC HEART RATE VARIABILITY WITtl COARSE GRAINING SPECTRAL ANALYSIS. ILL Hughsor: and Y. Yamasnoto:, 'Fg:ulty of Applied Health Sr University of Warerloo. and ~FacaRy of Edueatio~ Onivemty of Tokyo. Human heartbeat intervals are not constant. In 1973, Sayers (E~gonomics 16: 17-32) presented an analysis of heart rate variability {ILP.V) in the frequency domain showing clear spoewal power peaks at specific frequencies. Nine yea~ later. Kobay~hi and Musha (IEEE Trans. Biomed. Eng. 29:456457, 1982) reported that HRV showed no distinct peaks, but rather had a broad band specta'~ with s'~ po',ser inversely proportional to frequency (I/P) when plotted as log power vs. log frequency. "fhe major difference between these studies was the length of data collection, minutes vs. hour. The I/P specmun has been thought to reflect underlying fraetal dynamite. If frae~l dynamics exist In inngterm data, do they also exist in short-term data7 Our early work ha this area convinced us that this was the case. Coarse graining spectral analysis (CGSA) (Physica D 68: 250-26,4,1993) w~s developed to omaet fi-actal and harmonic signals from the muted panama that seem to he common in physiological data. The principle t~oa which CGSA is based is that of self-similarity. The random fr~tal con'tponent satisfies dm ~lalinr, ship xO~t)~h:'x(t), where e~h tide has the same dis'~ribution function, and H is the Hmst exponent (0 < H < I), CGSA computes the fraetal power from the geome~c mean of the crossspectralpowers Sx.~(n) mad Sx.x~~(n).with h = I/2 and h = 2, in which fr~utl powers a~ e~ed while harmonic power is not. The fractal component is assessed by computing the slopo of die 1/1r ~ladoaship, with ~ = 2H =1. The CGSA algorithm has been applied to a number of sirnulated dam sets to show its ability to exeact the correct percentage of fractal ~ora a mixed fiactal and harmonic signal, and its ability to renan the input value of H. For HRV ha resting, healthy humans, it has been cotaslstenflyfotmd that the percentage fi-~ctalcomponent is about 70-80%. and that p - ] for beth shortand long-term data. Under stress.13 changes to values near 2.0 indicating re6uced cardiovascalar stability ~ the mrtdom fr~tal Iin~e selSes ~mrrLs less frequened.. ;o the mest~ Supported by NSERC. Heart and Stroke Fotmdat~on.MRC Canada. ~a~dL~amot~ Fn6n- la~un.

437 INTERBEAT-INTERVAL AND COUNTING STATISTICS OF THE HUMAN HEARTBEAT Malvin C. Teich Cohanhia University., New York, New York 10027 A nmnber of statistical mcosul~ ale ~ 1o identify the form of the point process that desedbes the human hcsflbeat. These include the interevent.iaterval histogram (m-l), rescaled range analysis (R/S), the evant-nomber histogram {ENH), Fano-faeler if'F), Wavelet Fano factor (WF'F), and generalized-rate-based periodogram fRBP). The counting statistics show that the sequence of human heartbeats exhibits Ill-type fluctuations that asian fram longduration power-law correlation, behavior that has been ob~tved in a variety of neural preparations. All of thase measures have been applied to data from beth normal anti heartfailure patients, and various surrogate versions of these data. The results reveal that beth the intezbeat-interval and the lang-lerm counting statistics differ substantially for the two classes, The various measures that we have investigated suggest g,veral conveniently calculated. quantitative indices that indicate to which group (heart failure or normal) a particular data set belongs. We obtain a putative attraetoes capacity dimension from a phase.apace reconstruction based on the generalized heart rate. Though the depeedenco of the capacity dimension on the embedding dimension is consistent with that expected for a law-dimensional dynamical system (with a larger apparent dimension for the normal subjects), surrogate-data analysis sho~vs that this behavior conld just as well reseh from temparal correlation in it purely stochastic process. We present an integrate.and-rise stochastic model with an input rate process that is fractal Gaussian noise, thereby providing the power-law correlation in the heartbeat. The model satisfactorily characterizes the statistical behavior of the heartbeat.

435

438

GENERA'I'ING AND ANALYZING F'KACTALTLME SERIES: METHODS AND LIMITATIONS James B. Bassingthwalghte, Peter J. Tonellato, Donald B. Percival. Gary M. Raymond University of Washington, Seattle., WA 98195-7962

APPROXIMATE COMPLEXITY

Standard textbook methods for generating ffactal time series are the spoc~ral synthesis and Ihe sucees;ive random addition methods (e.g., in Fader, Frnctals, 1988 and Peilgen & Saupe, Science of Fractal Images, 1988). In analyzing artificial signals generated using these methods we found systematic errors in the estimates of the Hurst coeffinienl H ( H - E + I D. where E is the Euclidean dimension. D the fractal dimension); we attributed the errors in H. For "known" time series, to errors in the methods of analysis (Ann Biomed Eng 22: 432, 1994), However. Percival (Comp Sci Star 24: 534, 1992) had noted that Davies and Harte (Biometrika 74: 95, 1987) had developed an "exact" method. Using this method of signal generation, the dispersional analysis method (Bassingthwaighte, News Physiol. Sci 3:5. 1988; Circ Res 65: 578, 1989) gave conect estimates of H for series with 0.1 < H < 0 . 7 (strongly anticorrelated near neighbors for H = 0.1, strongly con-elated for H = 0.9). The method gave modest underestimates of H for higher H series that were short (N.,~IN,< 1000) and even 2 to 4% underestimates for long series (Nt,oi.. = 106). The other stand~l methods f~r fractal signal analysis, Hurst's reseated range, autocorrctation, and spectral analysis, are also highly susceptible to error when Npolnuis small. Fourier spectral analysis gives excellent estimates when Nl~m, is large, as might be expected for a signal generated by a Fourier method. In general all of these methods are subject to systematic and random error and the limilafions are revealed by large-scale numerical testing using tens of thousands of test runs. Of particular interest are the features of composite signals, for example, a sum of two fraclal time series. The addition of random white noise (which is a fractal with H = 0.5) with standard deviation (SD) up to five times the SD of signals of known varied H strongly degrades those with H < 0 . 5 but has much less effect on those with H > 0 . 5 : long range persistence. Supported by NIH grants HL50238 and RRI243.

Steve

436 FRACTIONAL ARIMA MODELS OF PHYSIOLOGICAL SYSTEMS Peter J. Tonellat~, Allen W. Cowley. Jr and James B. Bassingthwaighte Marqnette University, Milwaukee, Wl 53233-Iggl. Medical College of Wisconsin. Milwaukee, WI 53226. and University of Washington, Seattle. WA 98195-7962 Racial analysis of time series has proven effective in assessment ~f physiological signat~. Generally. analysis results in an estimate of a parameter which ret'l~cts the "fraclMness" of the signal Simulations of fractal time series and analysis of the methods used to estimate the parameters have demonstrated the strengths and limitations of this approach. Accurate modeling of the physiological signals would help resolve the limitations, but is problematic, One problem is that some physiological signals appear fracta[ with additional superimposed structure. A class of stochastic models known as fractional ARIMA models arc introduced as one approach to modeling and inteq~reting features of fractal physiological signals. Simulations were performed to demonstrate the types of structures within a fractal signal that cart be modeled. An important application of the resulting models is the characterization of mean arterial pressure (MAP) signals which afford a better understanding of the underlying control mechanisms of MAP. An emerging use of such characterizations is the identification of accurate methods of phenolyping hypertensive populations. Fractional AR[MA models were fined to MAP signals from a population (both parental and F2 interctoss) of hypertensive rats. The features revealed by fitting fractional ARIMA's to the data indicate mechanistic irregularities not identified by frsclal dimension estimation methods. Supported by NIH P01HL29587 and RR01243 and SUN Microsystems, Inc. EDUDUS 940208.

S-91

ENTROPY:

A

MEASURE

OF

TIME-SERIES

Pineus 990 Moose Hill Rd, Guilford, C T 06437

ApEn, approximate e~atropy, is a recently developed farmly of parameters and statistics quantifying regularity (orderliness) or complexity in data. W e discuss the motivation for A p E n d e v e l o p m e n t , f r o m the study o f application o f chaos algorithms to general time-satins data. W e indicate the superiority of A p E n to the K - S entropy and corrdation d i m e n s i o n f o r general statistical application, and for discrimination of both correlated stochastic a n d noisy d e t e n m n i s t i e processes, natural models o f physiologic time series. ApEn m a y b e applied to time-series arising in e.g., d i n i c a l endocrinology, to quantify ptdsatility i n h o r m o n e secretion data, given at least 7 0 data points, and in heart rate data analyses ( w i t h typically 11300 data points) -- representative applieatinns wilI b e shown, as will error bar estimates based o n general probabilistic models.

S-92

Nonlinear Dynamics in Physiological Systems II

439

441

A STATIST]CAL-BIOMECHANICS APPROACH TO POSTURE CONTROL JJ, Collins and CJ. De Luca NemoMuscular ResearchCenterand Dept. of Biomedical Engineering, Boston University, 44 Cummington St., Boston, MA 02215

REDUCTION OF A MODELED BURSTING NEURON TO THE ESSENTIAL DYNAMICS UNDERLYING THE BURST CYCLE

The task of maintaining erect stance involves a complex sensorimotor control system, the output of which can be highly irregular. Even when a healthy individual attempts to stand still, the center of gravity of his or her body and the center of pressur~ (COP) under his or her feet continually move about in an erratic, and possibly chaotic, fashion. In this study. we used surrogate data techniques and algorithms from dynamical systems theory to show that postural sway is indistinguishable from correlated noise. Tbese analyses indicated that balance regulation is better represented as a stochastic process, as opposed to a chaotic one. Accordingly, we then approached the problem of characterizing postu,-al sway from tbe perspoctlve of random-walk theory. Spocifically, we analyzed COP time series as onedimensional and two-dimensional random walks. These analyses revealed that over shortterm intervals of time during undisturbed stance the COP behaves as a positively correlated random wtdk. whereas over long-ten,c, intervals of time it resembles a negatively correlated random walk. We interpreted this novel finding as an indication that during quiet standing the postural control system utilizas open-loop and closed-loop control schemes over shortterm and long-term intervals, respectively. From this perspective, our approach, known as stabilogram-diffasion analysis, has the advantage that it leads to the extraction of COP parameters which can be di~ecdy related to the steady-state behavior and functional interaction of the neuromuscular mechanisms underlying tha maintenance of erect stance.

440 ERACTAL PROPERTIES IN BREATHING DYNAMICS OF PRETERM INFANTS U. Fray ~ M. Sifuerman*, H. Aaron e and B. Suki ~ 1 7 6 ~Oepartmertt of Paediatrics and Neonatology, Royal Postgraduate Medical School, Hammorsmlth HospRal, LondonW120NN (UK); ~ of Biomedical Engineering, Boston University, 44 Cummington Suest, Boston 02215 (USA) The statistical properties of breathing patterns were studied in 10 preterm infants (gestational age 24 - 40 wks) at the age of 0.5 - 22 wks. The interbreath intervals (IBI) were calculated from abdominal movements during a two hour sleep period. Distribution histograms of the IBis showed s log normal distribution for IBis < 1 sac and an inverse power law distribution for IBis > 1 sac that extended over almost two time decades. The slope alpha of the power law distribution varied between -2.9 and -1.7 and decreased with postnatal age. Fast Fourier transforms of the IBI time series revealed 1/f-like power spectra on s log-log scale. The correlations in the time series were further studied by calculating a fluctuation index as a function of time scale according to Peng et eL (Phys, Rev. Let. 70: 1343-46, 1993). This fluctuation function increased almost linearly on a log-log scale. The slope of the fluctuation function was close to zero (0,00B-O.04}. However, reshuffling the time series did not change the slope implying thereby that the correlations were partly due to the power law distribution of iBIs. These three facts together indicate a scale invariant behavior and suggest that fractsl mechanisms might be involved in the regulation of the breathing pattern of preterm infants.

R. J. Butera, J. W. Clark, and J. H. Byrne~ Dept. of Else. & Comp. Eng., Rice Univer~Ry, Houston, TX 77251 tDept, of Neurobiology & Anatomy, U. of Texas Medical School, Houston, TX 77030 A fundamental problem in computational neuroseience is to identify the minimal level of complexity necessary to capture the features of complex neuronal processes. The problem is particularly acute in bursting neurons, where multiple time-scales are evident. To exploce this issue we investigated the adequacy of low-order models to characterize the slow oscillatory waveform o f the R15 bursting neuron. Previously, we developed a Hodgkin-Huxley type model of this cell that cons;sled of 11 state variables. We first used numerical bifurcation analysis and computer simulations to identify the mathematical mechanisms that underly the bursting process, The results were applied to develop a reduced model oftha underlying subthreshold oscillations (slow-wave) in membrane potential, Two different low order models were developed: one model (3 state variables) which mimicked the slow-wave of the fall model in the absenceof action potentials and a secor~d rnodd ~4 state varlables~ which included expressk)nsaccountlng for the averaged perturbatlonal erects of action potentials on the underlying oscillation. Both models predicted the activity mode (bursting, beating, or silence) ~s parameters were modified, but the second model was much more accurate in predicting the location and extent of regions of specific modal act;vity in parameter space. The transient responses of both models were also compared to the original model, and the second model possessed a phase-response curve which was very similar to that of the original model. The results suggest that low-order models of bursting systems which do not consider the perturbational effects of rapid processes may erroneously predict modes of act;v;~ and transient responses of the complex model upon which the reductions are based, These results also show that it is possible to develop low-order models which retain the essential dynamics of the complex model.

Advances in Biomedical Modeling and Signal Processing 442

S-93

445 Adramem ta e i e a m d i ~ E m p u a m ~ at 0 ~ o U m i v m ~ . 3D ld~g.,t:,~ and A m d ~ s of Bones and Bimnelecuinr Mo6etim~ B.V.Me~ ~ p m t m e e d of M e d a m k ~ Engtanm~g Ohle Uaivor~ty, Mh,,m% Ohio 45701

Magnetic Rasonan~ Imaging ( MRI ), Computer Tomagraphy Scanning, (CT-Scan) and Ultra.sound Imaging techniqu~ were used for data scquisifion, in order to comOxtct/ devdop 9 3-D antld model ef tbe Huesaa ~ Fumur, Kane and the SImlL CT-scan wns found to be an 9 tedmique for cadavers, MRPs form 9 better tool in performing this image generation task for living beings, because 9 its high resolution capocity, whcu romparod to images obtained using Ultrasmstd techniques. High resolution poses In be 9 vesyimportant factor, slnro the coeslderation ofvarions material properties of the bones vnm one oftbo emphasis of this res4mreh. MP-J images have the capacity of dlspinying a distlnet boundary between the musdas and the bone in the leg, in addition to the boundary between the eartical and the cancellous region within the bon~ A collaborative and interdisciplinary program termed ~]~mlesll~Ens162 was established at Ohio University in 1991. The program consists of chemical and mechanical engineess from the College of Engineering and Technology, and life sdentists including molecular hiolagists from the College of Arts and Science. This program was founded to carry out rnsrarch on designing and producing novd recombinant proteins. The me~umlcal eagines~ have been and continue to work on the computer molecular simulation of protein structure for the rational design of new hGH analogs. The molecular biologists subsequently carry out corresponding recombinant DNA work including hGH gane muthganesis and development of stable mammalian cell lines that r the newly designed hGH analogs. The chemical angineors are involved in developing preparative- and pilot-scale upb-tramn and downstream procass~ to produce ~ i c i a n t amountsof the hGH analogs for further biniogical testing by the llfe scientists.

WAVELET-BASED FRACTAL ESTIMATOI~ AND MATCHING PURSUIT METHOD IN EARLY HUMAN DEVELOPMENT

Metin Akay and Eduard Mulder Biomedical Engineering Dept., Rutgers University The University Hoapital Utrecht Ftactal methods have shown to be usef~ in characterizing biomedical signals. The use of fractal estimation requires the estimation of parameter H, which is directly related to the fractal dimension D. However, traditional fractal e.ualysls requires that the biomedical signals be stationaxy. Here, we propels a novel approach which is a combination of the wavelet transform and fxactal estimators to cha~acterise the fetal breathing signals before and after the intake of two glasses of wine by a mother. This study was performed on 28 fetuses. The power spectra of the wavelet coefficients were estimated at each scale. The slope of the representation on a lognsithralc plot from the scales 9 to 3 wv.~found to be decreased after alcohol intake. Our results suggested that fetal breathing rates have a rough structure before the alcohol intake and smooth structure after alcohol intake. We also used the matching pursuit method to analyze the same database. The results obtained with the matching pursuit conf~rmed the results obtained with the wavelet based fractal analysis: the horinonta[ structured atoms representing the sinusoldal aetivlty at all frequency ranges di~ppeaxad and the vertical structured atoms representing the discontinuous spike type activity increased. In addition, the circular structured atoms at the high frequency range shifted to the low frequencies. This work was supported in part by AHA NJ Affiliate

443

446

APPLICABILITY OF LEARNING ALGORITHMS TO COMPLEX DATA SETS Donna L. Hudson ~f, Maurico E. Cobent,i'f 1"University of Califoroia, San Francisco, l"'fCalifumia State University, Fresno

TIME-FREQUENCYANALYSISMAY SEPARATERESPIRATORY-RELATEDEVOKED RESPONSES(RREPs)FROMFACIALMUSCLEEMG CONTAMINATION J. Andrew D9 Lisa M. Lim2,and Metin Akay3 L2Physiology Dept., Dartmouth Medical School, Lebanon, NH 03756, a n d Thayer School of Engineering, Dartmouth College, Hanover, NH 03755 and :~Dept. of Biomedical Engineering, Rutgers University, Piscataway, NJ 08855; It has long been known that electrical field9 from facial muscle activity can contaminate the electroencephalogram, even when closely spaced, bipolar electrode configurations ar9 used. We suspect that RREPs stimulated by pressure pulses applied at the mouth may include EMG contamination subsequent to muscle reflexes triggered by the negative pressure. We obtained simultaneous recordings in human subjects of the RREP (from Cz-C4) and masseter muscle EMG (from a bipolar surface electrode pair) during application of 9 (-5 cm H20), 200msec pressure pulse9 In addition, we recorded simuflaneou9 EEG and EMG responses during voluntary tensing (VTen) of the massetor. Preliminary analyses of the VTen responses showed 9 but significant short-latency cros9 for EEG as a function of EMG. Traditional spectral analysis of the VTen EMG 9 EEG signals showed closely similar spectral content. Matching pursuit analysis permitted identification of different time-fraquency(TF) patterns for each signal: the EEG showed periods of low frequency activity unmatched in the EMG TF pattern. These matching pursuit results indicate the EMG and EEG signals are surely nonstationary, and that separation may be possible using TF techniques.(l"his work was supported in part by NIH grants HL19827 and HL29068.)

Since the reemergence of neural network modeling in the last decade, a number of lemning algorithms have been proposed. Probably the most popular has been beck propagation, but recently other approaches, such as rain-max methods, have gained in popularity. When choosing an algorithm, a number of factors need to be considered, including the type of input data, the stability of the data, the number of input nodes, time required for u-aining, and simplicity ofnse for classification. Some learning algorithms are limited to specific types of data, for example binary, while other algorithms may have stability problems under some conditions resulting in chaotic solutions. A new learning algorithm developed by the authors which uses a vector space approach to weight adjustments will be compared and contrasted to other learning algorithms. In particular, the adaptability of cach of these methods to decision making using diverse types of medical data will be addressed. Data types will include binary, ordered categoric, continuous, fuzzy, and temporal. Special t.ttention wilt be given to the inclusion of time series medical data combined with other types of data input. In addition to accepting diverse data types as input parameters, the learning algorithm provides flexibility in the deveinpmsnt of the network atruaturo itself. The basic stroctura is a three-layer network. The second or hidden layer can be restructured by the user to include multiple levels of node interaction. In addition, the number of layers can also be changed through the use of fractional contributions of nodes. These variations in stroctore are made possible through the choice of parameters in the potential function, the Cohen multidimensional orthogousl function, which forms the basis of the vector space. The flexibility of this approach will be shown in several applications in medical decision making.

444

447

ISSUESIN CHAOTIC MODELING OF BIOLOGICAL SYSTEMS Manrice E. Cohent,l"t, Donna L. Hudscul'~f tCalifomia State University, Fresno, t tUniversity of Caliinmia, San Francisco

APIR.,ICATIONOF NEURAL ~ O R . K S (Fuzzy A R T m p and Prohabilistic NN) AND STATISTICALMETHOD IN PREDICTING THE 3-1) STRUCTURE OF PROTEINS B. V. Mehta, & L. C. Rebate, Department of Mechanical Engineering Ohio University, Athims, Ohio 45701

In the last decade, chaotic models of biological processes have become quite widespread, particularly in cardiology. Usually the chaotic approach is used to model nonlinear processes, such as hemodyanmins, where direct mathematical modeling leads to intractable solutions. Analysis of bioingical signals, such as electrocardiograms and clectroencephaingrams, has also been attempted using chaotic models. Such models are prone to many problems. The inherent nature of chaotic models makes them susceptible to computational error, which under some circumstances can propagate and overtake the actual solution. In addition, chaotic models are a discrete representation of the actual situation. Through theoretical development of a conjectured continuous solution of the logistic equation which we have developed, these potentialproblems are addressed. This theoretical approach to continuouschaoticmodeling

will be illustratedin the analysisof the variationin R-Rintervalsfrom 24-hourHoliertapesof individuals with varioustypesof heartdisease. The applicabilityof this approachwill be discussed in terms of the analysis of other biological signals, including electroencephalograms and hemodynamic modeling. A number of methods which have been developed through the continuous chaotic modeling approach will be illustrated in conjunction with these practical applications, including first and second order continuous Poincar~ plots and a measure of central tendency which computes the amount of variability in the data set. This last measure is particularly useful in large data sets, such as Holter tape analysis, in which more than 100,000 points are typically present. This measure is also useful in decision making for categorizing cases as normal or abnormal. It can also be combined with other measures, such as clinical parameters or patient histo~, in a ncural network model for more complex decision making.

ABSTRACT Several techaique~ have becu developed for protein secondary structure prediotice using different type of statisticaland neural nelwo~ techniques. The neural network schemes have emphasized the utilization ofbeckpmpugatlon, whiletbe statistical technique has been based on the Chou-Fassman algo~thm. In this paper, the Fuzzy ARTmap paradigm sad Prohabilistic Neural Natwork(PNN) technique is used for the artificial neural network scheme and a modified r algoritinn is used for the statistical technique. The remdta of this new Fuzzy ARTmap method, cod the prohabUistic neural network method appfied to secondam/stricture prediction is compared with the existing techniques like back propag~ttion method and the modified Cbeu-Fassman statistical method. The artificial neural network based on the supersized adaptive reannance themy provides 9 system which can make 3D structure predictions, comparable in accuracy to the beck-propagation algorithm and the training time required is much shorter. The statistical technique based on Cbou-Fassman rules gives good mcults, but with the Fuzzy ARTmap the network can be improved by adding more proteins to it. In the PNN technique the total time for traininng and testing is significantly less compared to other neural network schem~. The analysis of the results have demonstrated that the developmcut of a multi-expert architecture withdifferent raprescatation schemes could be a better sointion.Thin multi..expert architecatre would utilize roles based on statistical analysis and neural networks to grow w/th examples. In addition, the utilization of edditiousl infornmtion such as energy levels is being cousidem:l.

S-94

Poster

Presentations

451

448

FRACTAL GAIT DYNAMICS: STABILITY OF STRIDE INTERVAL LONG-ILANGE CORRELATIONS Jefrey M. Hausdorff, Patrick L. Purdon, C.K. Pens, Zvi Ladin, Jeanne Y. Wei and Ary L. Goldberger

BIFURCATIONS AND CHAOS IDENq'IHED IN A PRESSURE PULSE DRIVEN COLLA~IBLE VESSEL Shawn Fidd mad Gary ~ e c k i Ringer* Univ~'sity, Biomedical Fag. Dept., Cardiovascular Research Labs. Piscataway. NJ It was hypothesized that a sinusoidal input ~essare with a zero tnsm~ flow can ixodece complex dynamics in a cellapsible vessel. A lumped mathematical model of a collapsible vessel mid an exlmtimmtal smiog ~xa-e devdoped. The c~ttintfity equation was solved for the Imnen re'ca, which, in tar~, is ased to rind the inetUmce and flow tmistance. In addition, the nonlinear prcssme-m~a (P-A) OLU'Vewas used to deterufine the imtantmeons transmural pressure from the lumen are& Flow resistance was assumed to ~omlst of a Sllmm~tti~ of Poiseuille a~l Bemoufli re,stance. The outflow was plugged to prodace zero net flow. Output pressm~ had flows were derived by solving the modelling equatioas via a compmer algorithm. They were me~sared exp~imemtally ~ correspoeding conditiom. downstremn pressmm in the vessel w a s studied for various mmbinatinns of upstream poise amplitude, mean preasure ~md frequency. While p l ~ plot- 6~sUr4 r some c~abinations wea'e found to Ixoduce a simple linear scaling mad phase shift of the input sinusoid, it was also possible to identify frequency blfmcadons and dmos. The phase plot was shown to provide evidence of a strange attractor fee the chaotic pressure waveforms (fig.). It was ceududed that the combination of low average pt'~sure, comparably large pulse amplitude, atad freqttmcy nero the linem resommoe of the vessel c~a prodme getind bifurcafiom md chaos.

Department of Medicine, Beth Israel H~pital, Boston MA, Biomedical Engineering Department, Boston University, and Center for Polymer Studies, Boston University, Boston MA Practal dynamics were recently observed in the apparently "noisy" variations in the stride interval of human walking. Dynamical analysis of the step-to-step fluctuations in the stride interval of healthy adult subjects revealed a self-similar pattern: fluctuations at one time scale are statistically similar to those at. multiple other time scales, at lea.st over hundreds of steps while walking at one's normal rate. Here, we extensively examine the characteristics of this ftactM property. To study the stability of tht*e longrange correlations in the stride interval, 10 healthy subjects wece tested for 1 hour at their normal, fast and slow paces. The stride interval exhibited long-range correlations with power-law decay for up to thousands of steps st all three walking rates, and during treadmill and blindfolded walking. In contrast, during metronomically-paced walking, fluctuations in the stride interval were random ~ n d aon-fraetai. The~e recruits indicate that the fractal dynamics of the stride interval are normMly quite robust and apparently intrinsic to the locomotor system. Furthermore, this fraetal property of neural output appears to be most likely related to the higher nervous system centers responsible for control of walking rhythm.

449

452

CHEMICAL MICROSENSORS FOR REAL TIME MONITORING OF ELECTROLYTE AND MUCUS SECRETION FROM CYSTIC FIBROSIS RELATED Mabintou T r n o r e ] , R o h i t K a s h y a p t , Carol B e r n a r d 2 , U l r i c h H o p f e r 2 and M i k l o s Gratzl: Department o f Biomedical Engineering t and P h y s i o l o g y and Biophysics2, Case Western Reserve University, Cleveland, O H 4 4 1 0 6

AN APPROACH T O REALISM IN COMPUTEK, MODELLING OF MECHANICAL INTERACTIONS: APPLICATIONS OF MOTION DATA AND VIRTUAL REALITY TECHNIQUES G. Jeffzi~, D. IL Wright, N. Rogen*,J. Ward* Brand Univenity, Egha~ Sarrey, UK. *Cardifflmcimte, LLamda/rceatre, Wale~

Little if any is known about the tn~:~sses leadiag to the formation of abnormally viscous mucus in the airways of patients suffenng from Cystic Fibrosis (CF). An entwstanding of mucus polimerisation in a cell line related to CF. HT29-CI.16E, may help to develop more effective ways to control or to alleviate certain symptoms. Besides HT29-CL 16E, another cell line, HT29-Cl.19A, was also used in this work, to evaluate physlco-r changes when electrolyte seerefioe eccared alone. The goal of this project w a s then to evaluate the changes in ionic and redox conditions in the microeavimnnement of the*e two cell lines subjected to stimulated mucus and eleclrelyte secretion. Ion-selective microsensors (o.d. - 200 ttm) for e l ' , Ha*, K + and pH were developed and used to monitor the respective ionic concentration changes directly above the mucus secreting layer grown on millipore filters and placed in a Ussing chamber, Secretion was induced by the apical application of secmtagoguns llke ATP and Carbachol. The results indicate that it is feasible to obtain real time information about the physico-chemical changes due to secre~on at 10-200 pm distance from the studied cell monolayen at high temporal mseIution. With computer simulation of the diffusion precefaes in the extracellular matrix, an agreement was found hetwecn results computed from Isc data and direct micmsenser measurements for the secreted chloride transient

We begin by disctming the devdopmem of a technique for modelling the Fcoduct-user interface. With mechanical modelling software, a linked segmem 'Android' it created, connected by a r,etiet of user defined joints. Tbese joints are driven by motion data, a dynamic analy~ is geffotmed, and from this torques and fotr~t in dae joints are calculated. Our previom work has verified this technique in the analysis of the product-met int~-f~ce in the derdgn of products for the elderly mhritir user. but it has also shown the limitatlom involved. The joints used in the link modal are simplifications, and do not have any mae attatomical constmunts. We discus* how the dcvdopment of a more tophistlcat~ notation of joints such as the hand, tpine, and shoulder contzibutes m more accurate modelling of upper torso movements. We introduce a new methodology, combining our existing tedmique, with motion analytis through virtual reality software. This enables the modelling of more realistic moveanent, and body limitations such as range of motion,and postural far:ton inch as balance. Data collection, especially of the hand. it amcid m the succesfful modelling of the productuser interface. We define a methodology for the collection of joint motion data fxom a variety of sources, including video, goniometer and data#ove based r/~erm. In conclusion we presem a case study u~l;,;ng the new moddl/ng technique and data collection methods, and show how this can be mud to chuacterise motion, with particular reference to the product-user interface.

450

453

BISPECTRALANALYSISOF SIMULATEDECG MEASURES BEAT-TO-BEATVARIABILITY Scott D. Greenwald, Ph.D. Aspect Medical Systems, Inc., Natiek, MA

COMPARISON OF OPTICAL IMAGE RECONSTRUCTION USING DC AND AC MEASUREMENTS OF DIFFUSE LIGHT SIGNALS IN CENTIMETER SCALE PHANTOMS Huabei Jiang, Keith D. Paulsen, and Ulf L. Osterberg Thayer School of Engineering, Dartmouth College, Hanover, NH 03755

Bispoctral analysis is a signal processing technique which quantifies the phase and power relationships among sinusoidal components that make up sigmds (e.g, QRS complexes). The bieoherence quantifies the repeatability of these relationships over multiple observations (e.g.. beats) and yields a measure from 0 to I (i.e., from poorly to perfectly consisten0. The two-dimensional (triangular) domain of the bicoherence is the set of (f,,f~) palr~ that yield ~dl unique frequency triplets (f,,fz,f,4f3) in the digitized signal. Specifically, let X(fl) and P(ft) be the Fourier Transform and power of one ECG epoch at frequency fl, respectively. Then, the bleoberence (BIC) over N epochs is defined byI tl,etl,,l.I = l ~ , x ( l,,x Il. lX'tf, + I D I / ~ - ~ , Compared to isebomie myocardium, healthy myocardium produces highly repeatable QRS complexes2. Thus, the avecagebicoherence (over all (fl,f2)) of boats from healthy myocardium is expected to be higher than that f~ d'seased myoeardium. To demonstrate the reduction in average bieoherence with increasing variability, three levels of variability were simulated by injecting increasing levels of white noise within ]

eI I,IPI ID.(I, . l.,

k,*.~ , ~/~t_ [ | I 'l 9' ~j ~ j ~_ R ]A A [ i ~I .~ -

~,,~ ~

. ' '

~ ,

.

~ -

:

-

.

:

-

-

-

,

. ~ . ".

complexes) while maintaining a constant ~ baseline noise level. One hundred epochs per level Ithe first 2 are shown) were used to generats the average power spect~m a~adbicoborenee pictured above. The change in the bicoberence with increasing beat-to-beat variability is dramatic compared with subtle changes in the time series or the spectrum in these simulations. Future studies will investigamthe use of bispectral analysis to noninvasivelydlagno~L~eorona~/artery disease using the surface ECG. l. Science 172:401-402. 2. Hilth Resolutinn Electrocardiography.EI-Sherif and Turino, eds. pp 299-337

Simultaneously reconstructed images of both absorption and scattering coefficients in optically heterogeneous media based on frequency-domain measurements using DC and AC data are presented. The images are obtained from absolute measured data using a diffusion approximation to light propagation in tissue which is modeled with a finite element method. Images having different contrast levels are demonstrated. Comparison between images using DC and AC data show that those reconstructed from AC data are significantly more quantitative with respect to location, size and shape o f the heterogeneity, contrast level and changes in contrast level than their DC eounterparts. The results presented here highlight the potential importance o f using AC data for biomedical applications involving diffuse light imaging through centimeter-thick tissues.

Poster Presentations

S-95

454

457

TEMPERATURE DEPENDENCY OF OPTICAL PROPERTIES OF PROSTATE TISSUE Wilfiam Nau, R o b ~ Rosefii, sod Dougias Ivfilam Departments of Biomedical Engineering and Urology Vandecbilt University, Nashville, Tennessee

IDENTIFICATION OF CHAOTIC DYNAMICS USING VOLTERRA. WIENER SERIES. Manficio Barahona and Chi-Sang Poon. Department of Physics and Harvard-MIT Division o f Health Sciences and Technology, M.1.T., Cambridge, M A

Over the last several years the~e hns beco so i ~ interest in the nse of insets to treat benign prostatic hype~lasia (BPH). Cfinical investigadons alone have been unable to detea'mice an appropriate desimetty for effective treatment. Therefore, mathematical models of temperatm~ distribufiom in tissue lmXinCed by laser exposure have been used to augment these studies. Since temperatme distributions are dependent upon the fight distribution, it is e~entlal to have accurate measnsemants of the optical properties Of tissue, and an understanding of their dependence on temperatare. Optical absorption and scattering coefficients o~ excised canine Westute I ~ u e have been measured as a function of temperatm~ using a double integrating sphere system. Measurements were made with two differoat lasers; Nd:YAG ('L ffi 1064 van) and IteNe (Z = 633 nm). The f'nst measurements wese made at room temperature, then the samples wore placed in a watorbath and the temperature ~ in increments of 5~ C. Measurements were made at esch tempecature up to 10ft'. Resalts showed that the scattering codrgiant is strongly dependent upon tissue temperature at both wavelengths. For the Nd:YAG laser, the scattering coefficient increased from 2.5 mm 4 to 21.0 mm "t, and for the I-leNe laser an increase from 1.0 mm a to 37.5 mm "1 was measured. The absorption coeffieinnt, however, remained relatively constant over the temperature range considered with values of 0.04 mm "t for the Nd:YAG wavelength, and 0.13 mm "t for HeNe. The greater values for the optical properties measured with the HeNe laser than those measured with the Nd:YAG laser, confirmed the wavelength dependence.

Recently there has been a great dcel of interest in a nonlinear dynamics description of physiological signals. However, conventional techniques o f nonlinear dynamics analysis may not readily detect the presence o f a chaotic component in empirical time series because these are inevitably corrupted by measurement and/or physiological noise. We have developed a new analytical technique to detect chaotic dynamics in noise corrupted time series data. The method is based upon a closed-loop version of the Volterra-Wiener series with an extension to non-white input signals using the Korenberg algorithm. This approach provides an efficient strategy to search for the best polynomial model in the embedding space. Moreover, by comparing the prediction power of linear vs. nonlinear models of the original data and randomly generated surrogate data, a statistical criterion for the presence o f nonlinear dynamics in the data can be established. We have applied this teclmiquc to a wide variety of linear and nonlinear systems, both theoretical and experimental. Results show the technique is highly sensitive and robust in differentiating chaotic signals from random signals even in short time series with a large proportion o f additive and/or dynamic noise - up to a h l signal-to-noise ratio in some instances. The consistently good ~,ii'ormance of this method with such a wide spectrum o f examples suggests the fcasihLlity o f its application to biomedical data. Supported by NIH grants HL50641 and HL45261; ONR contract N00014-95-0414; N S F gram BES-9216419; and a MEC-Fulbright FeLlowship.

455

458

BLOOD, UNIVERSAL PRECAUTIONS, AND BIOMEDICAL ENGINEERING Ott DE School of Engineering, Mercer University, Macon, Georgia

PHOTON MIGRATION IMAGING FOR BREAST CANCER DETECTION Tamara L. Troy, Lena Nelson, Dilip Paithankar, and Eva Seviek-Muraca School of Chemical Engineering, Purdue University, West Lafayette, IN 47907 In the U.S., breast cancer strikes one in eight women in her lifetime. While conventional x-ray mammography is an excellent screening tool for breast cancer in women > 50 years of age, the technique is compromised by the increased scattering capacity of breast tissue in younger women. Consequently, the small risk of x-radiation outweighs the possible benefit in younger women. In this research, we are developing methods which employ nonionizing, nens-infrared light to provide optical imaging of breast cancer in this unserved population. A technique called "Photon Migration Imaging," depends upon measuring the propagation characteristics of a photon density wave as it travels through several centimeters of tissue. Assuming there ls sufficient optical contrast due to the neovascularinafion of thn tumor bed, the solution to the inverse imngbag problem may provide images of diseased tissue volumes in women who are not candidates for x.ray mammngrapby. Using an integrating sphere apparatus, we measured the optical properties of over 50 specimens of normal and diseased breast tissue samples at 37 C to determine the degree of optical contrast possible for photon migration imaging. Our measurements show that while optical property differences exist between normal and diseased tissues, these differences were not significant. Indeed, t'mite element solution of the diffusioo equation which describes the forward imaging problem suggests that the magnitude of contrast necessary for detection is not available endogenonsly. These results poim to the nse of exoganous conttnst ngants in photon migration imaging for the detection of cancer in high-risk populations not well served by mammography. Using a Ti:Sapphire pulsed laser system, we measured changes in the propagation characteristics due to the presence of a volume laden with contrast agent. The substantial change in photon migration kinetics measured in these studies point to the successful use of inverse methods based upon perturbation analysis of photon diffi~inn in order to reconstruct images from multiply scattered light. Supported (EMS-M) in part by the Whituker Foundation, the National Institutes of Health (R01-CA61413-02), sod the National Science Foundation Young Investigator Award.

Combustion and aeroanlized products of human tissue contain blood and blood by-products are not regulated by universal precautions. This material is potentially infectious tad hazardous if inhaled. The blandbourae pathogan standard is a requiremeat of law (Title 29, Code of Federal Regulations, Part 1910). The standards purpose is to limit occupational exposure to blood and other poteatially infeetinus materials (OPIMs). It covers all employees anticipated to have contact with blood or OPIMs. The expo~are control known as universal precautious is for situations where differeatiation betweea body fluids is difficult or impossible. Aerosols created by dormshrasino, bone grinding or smoke from tissue combustion contains blood and bloodbourae pathngees that result in a cirt:umstaace wbera differentiation of body fluids is impossible. In these situations all body fluids, including bioaerosols, should be considered poteatially infectious materials and need to he guvemed by universal precautions.

459 SHORT-TERM IN VIVO EVALUATION OF G L U C O S E SENSOR Ebtisam Wllkins and Flamen Atonasov ~ t ul"C~ainal and Nuclear F~inesfing University of New Mexico. Albuquerque, N'M 87131 A new binaansor suitable for long-term continuous glucose monitoring has been developed. Extension of the sensor lifetime is aclaieved by refilling the sensor bioreactor with fresh enzyme without sensor disassembly. The sensors perfornxmces has been evaluated in vitro in model solutions and serum. An integrated implantuble device was constructed consisting of the sensor, a miniature potentiostat, FM signal tmasmitter and power supply. The device was implanted under the abdominal skin of large dogs for'in vivo evaluation of sensor performance. The short term implantations (up to 3 h every test) were conducted trader atm~the~a. The blood ginecee levd was increased by intravenous injection of 0.6 g glucose per kg body weight. Blood samples wore collected periodicaLly to obtain the blood glucose level changes during the test. Sensor signals were remotely recorded via telemetry. The sensor signal closely followed the blood glucose change with approximately a 10 min. delay. Signal to noise ratio was in the acceptable 5-10 range. The reproducibility of the umsor performance obtained during these tests make it possible to begin long-term in vivo characterizatioo of the device. Successful long-teem canine implantation of the integrated device will demonstrate the potential of the sensor and will allow incorporation it in an insulin delivery system. The insulin pump coupled with the implanted sensor as a feedback controller will provide a complete artificial pancreas. When it is used to provide continuous control of glucose levels within the normal physiological range, it will markedly reduce the incidence of the diabetes complications, hence, reducing health care costs.

S-96

Poster Presentations

460 EFFECTS OF ORDER, MEMORY LENGTH AND INPUT SIGNAL IN HIGHER ORDER KERNEL ESTIMATION: APPLICATIONS TO LUNG MECHANICS Oin Zbeng, Bdla Suki and Kenneth I~, Lutchen Department of Biomedical Engineering, Boston University, MA 02215 A nonlinear system can be characterized by e functional series. Estimating the kernels in the series can lead to the identification of system structure. In practice, due to experimental and computational limitations, one must truncate the memory end order of the sedes, Moreover, the mathematically preferred white Gsusstan noise Input may not be appropriate for certain physiological Systems such as lungs, In this study, we examine in detail the influence of these truncations on kernel estimates. We use Korenbetg'a algorithm to estimate the Volterrs kernels of block structured nonlinear lung models in which memory length may be s critical issue due to the low-frequency characteristics of lung tissues. Simulations demonstrate that improper choice of order Or memory will significantly distort the kernel estimates and subsequent Inference on system structure. We present a new method that incorporates the Akaike criterion into Korenberg's algorithm to identify the kernels simultaneously with the memory length. This method correctly recovers the memory length of a finite-memory system for noise-free case and data with small amount of additive noise. We also applied various random and pssudorendom inputs to the models and found that sufficiently wide input bandwidth is required for Korenberg~ method. One of our pseudotandom signals contains frequencies st avery integer multiple of the fundsmentel (which allows kernel identification) but the magnitudes and phases ere optimized to enable physiological ventilation. Therefore OUrspproech makes the estimation of kernels more practical by using physiologically relevant inpute and avoiding erbltreW choice of memory length. (NSF BCS-9309426 and NIH HLS0515)

Articular

Cartilage

Mechanics

S-97

464

461 FE FORMULATION OF' A POROVISCOELASTIC MODEL FOR ARTICULAR CARTILAGE J-K. Suh and S. Bai Masculoskele~I Re,~m~h Center, Dept. of Onhopoedic Surge.'y University of Pittsburgh, PiUsborgh, PA 15213 Introduction: At~caler cartilage, a composite mixture of extracellular matrix and interstitial fluid, manifests viscuelasfic~ t e d s t i c a , perdy due to the frictional interaction between the porous ~ substrato and the imerstitlal fluid and pardy due to the intnesic viscoelastic bshavi~ of pmteoglycan gel. The poroviseor.lasdcmodel has been used to pl~ict the I-D deformaticeal behavior ~ anlcular c~usilage ~ t h a simple geomct~'k: configoration [1,2]. In this study, we constructed a finite element fonnuin6on of the poroviseoeinstir model to investigate the mechanical defcenatloa of anicular cartilage under cyclic medmnical compt~sion. Materials and Methods: The intrinsic viscoelastic characteristics of extracclluler matrix were represented by the quasi-llnear ~ a s t i c (QLV) theory [3] and incospmated into the biphaslc thonq, [4]. The governing equation was formulated into a weighted residual form using the Galed~ method. The incompre~ible coesteai~ts for exWaCelhilarm a r x and interstitial fluid were incorporated u.dag the pensiq, method, and the time integration of the reduced relaxation fonctlon was dJactefized esmg the incremental trapezoidal role. A cyclic compressive force, F = -Fo cos(0)t), was applied to a cylindrical plug of articular cartilage, and the tissue deformation and interstitial fluid velocitywereu a l c u ~ Results and Discussion: The intrinsic viscoelastic characteristics of extraoellular matrix tend to slow down ~ defurmafimal response, causing a decrease in the interstitial fluid velocity within the tissue mamx under a cyclic compressive forec. The FE formulation of the porovincorJasdcmodel is useful fur studying the physiologicallyrelevantproblem of ca~lage mechank~ under dynamic loading. References 1) Mak. A.F.. J. BiomeclL Eng. 108:123-t30. 1986. 2) Setton. L.A. et al. J. Biomech., 26:581-592,199I. 3) lung, Y.C., Biomechanics. Mechanical properties of living tissues. Springer-Veriag. New YoA. 1993. 4) Mow V.C. et al. J. Biomech. Eng, 102:73. 1980

THE EFFECTS OF CRACK FACE FRICTION AND ANGLE ON SURFACE C R A C K S IN JOINT C O N T A C T

A.W. Eberhatdt,I-LM. Stobblefr and B.S, Kim ~ p = t m ~ o~M~hanlc~ ~ University of Alal~ona at Binniogham, Birmingham, Alabama The advanced stages of osteoanhtitis are charanterized by severe fibrillation of the articulating joint surfaces. While there is evidence that post-traumatic ~ may begin with surface damage induced by hlgh-energy iml~ct, this damage need not igogress to osteoanl~tis. Our recent studyI inve,~Jgated the streSSintensity lasSoS, K I and KIL m the tip of a vertical surface breaking crack in a two-diment'ional, layered hstf-space. subjected to Hcrtzian loading. Using f'mite elements and the "stress method" for extrapolating KI and Kit values, the effects of load location, friction, crack length and layer thickness on the slress intonsities were examined for the case of a layer/anbstreto ccmbinatice simniafing cartilage on bone. The msnits e~"the study mppmtod sbear~g as the dominant mechanism of surface crack pmpagetlce in articularcartilage. MasJmum KII vst ues incseased with increased crack length and de.erea.wxlwith thioning or stiffening of the cartilage layer. The results indicated, however, that compre~ve crack tip stresses coior with maxumum KII values, indicating the need for an analysis which includes entek face friction. As impact induced surface cracksmay eminate flora the surface at an anglo, it is of further interest to examine the effect of crack angle on KI and KII. In the ~ t study, crack face friction is inuoducod to the finite element model through the inclusion of frictional gap elements along the modeled crack face. K I and KE values are examined as a function of crack face friction and crack angle, as well as contact load location, crack length, cartilage thickness and stiffne.ts. IEberhardt, AW and Pert, S, J. Biomech. Eng., 117, 153-155, 1995.

464a

462 ANALYSIS OF CHONDROCyTE DEFORMATION AND MEMBRANE STRETCH AND SHEAR DURING CARTILAGE COMPRESSION Nathaniel M. Bachrach, Gall S. Chorney and Van C. Mow Department of Mechanical Engineering, Columbia University, New York Many recent studies have demonstrated the ability of chondrocytas to sense and respond to their mechanical environment. An analysis of the mechanical stimuli applied to the cell during tissue loading is essential to an understanding this mechanical transduction process. A mathematical model has been developed to analyze the deformation of a cell and its membrane within a tissue explant during confined compression. The model is based on Guilak et aL (Adv Biueng, BED17:395-398, 1990), treating the cell as a mechanical inclusion embedded the extraccllular matrix (ECM). Both cell and ECM are modeled as linear biphasic materials (Mow et aL, J Biomech Eng, 102:73-84,1980)with distinct material properties: aggregatemodulus (H,) Polsson'$ratio (v) and permeability (k). The cell is consideredto be spherical and small compasodwith the extent of the ECM. Biphasic potential functions are solved inside and outside a spherical rogion, superposod on the confined compression solution and the interface conditions for two biphasic materials are then imposed. This method provides a close form solution in the Laplace domain, Numerical calculations show that upon loading of the tissue membrane (celI-ECM interface) areal strain is positive near the poles of the cell when vcell>yECM. In time as tissue strains increase this region of positive surface strain spreads and their magnitudes increase. The ceil volume change can range from 0 to 1.5 times the applied tissue strain, for H~en/HxEc:M= = to 0, with v E~ = 0.1, Volume change and membrane strain are dependent on both time and position of the cell in the explant, This analysis provides precise information about the deformation of sells when the tissue is loaded. The model will be useful in evaluating the stress, strain, pressure and fluid flow fields in and aroundchondrocytesduring tissue loading.

463 BIPHASIC CONTACT CREEP BETWEEN TWO DISSIMILAR ARTICULAR CARTILAGE LAYERS UNDER A STEP LOAD RajeevKelkar and Gerard A. Ateshian Departmentof MechanicalEngineering,Columbia University, New York The analysis of joint contact is of interest to researchers investigating the function and longevity of articular cartilage. "Fnis abstract de.scrlbes a theoretical solution of the time dependent contact creep response of two dissimilar cartilage layers under a step load. Plane strain potential functions are used m solve the biphasic equations developed by Mow et al (J Biomech Eng 102:73-84, 1980). The configuration for this problem is similar to that in the asymptotic analysis of Ateshian 9 al (ASME Adv Biueng 22:191-194,1992). Beth cartiinge-bonc interfaces are assumed to be rigid and impermeable, while the contact interface is assumed to be frictionless. Nondimensionalizatlon is performed using an appropriate combination of material and geometric parameters used to define both layers. Using the interface boundary conditions defined by Hoe et al (J Biomech Eng I l 1:78-87, 1989), continuity of the fluid pressure, effective stress, normal displacement, and normal fluid flux are satisfied. Double Fourier-Laplace transforms of the governing equations are obtained, and their solution results in a set of four integral equations for each cartilage layer. Once the surface traction and pressure are determined, the dependent variables (stresses,strains,displacements)at the surfaceand through the depth of the layers are caleotated.The solution demonstratesthe dependenceof the responseon the relative modull and permesbilitles of the two layers, and describes the preferential direction of fluid flow in the contact region. For example, when one layer is made more permeable (as in osteoarthritic csrtitage), the solution predicts fluid flow into that layer within the region of contact, The solution also demonstrates that interstitial fluid pressurization supports up to 90% of the applied load immediately after load application. However, in layers with higher permeability this load support mechanism is defeated at a much higher rate, resulting in greater loading of the solid matrix, This analysis provides the first exact solution for the contact creep response between two dissimilar cartilage layers, and can be used for predictions of normal and abnormal joint contact.

EFFECTS OF MEDIAL/LATERAL DISPLACEMENT CALCANEAL OSTEOTOMY ON CONTACT STRESSES IN THE ANKLE S.J. Steffensmeier, C.L. Saltzman, T.D. Brown Depts. of Orthopaedic Surgery and Blomed. Eng., Univ. of Iowa, Iowa City IA 52242 Medial or lateral displacement transverse catcaneal osteotomy has been aneedotatly reported to afford relief of symptoms for vedous disorders ostensib/y involving untoward articular contact stress distributions in the ankle. However, objective data are lacking to document the relationship(s) between distal segment displacement end contact stress alteration. We here report results from a sedes of eight (8) fresh-frezas cadaver specimens, in which 1-cm distal segment offsets were imposed in both the medial and lateral directions. Mid-shank transection specimens were subjected to physiologically representative axial loads of l x and 2x body weignt, using special fL,dudng that ashieved a 90%:10% load allocation between the tibia and the fibula. Contact stresses were mapped using Fuji PrasSensor film, quantitated by digitst Image enelysis. Small metal beadlets (two per spedmen) imbedded in the articular carriage allowed trecklag of osteotomy-induced shifts in the center of pressure. The data showed consistent shifts of the center of pressure In each compartment, antero-medistly for the 1..cm medial displacement osteofomy (mean medial shift = 1.50 ram, s.d. = 0.93), and posterelaterelly for latera=displacement nsteofomy (mean lateral shift = 1.06 ram, s.d. = 0.55 mm). Histogram representation of the contact stress distdbutioes pre- and post-asteotomy showed no noteworthy changes in the amount of severely-loaded cartilage for either the medial or lateral displacements. Within the limitations of the model, we therefore conclude that catcanest osteotomy does achieve its intended objective of shifting the center or pressure, end that it does so wftbout causing spprer contect stress elevations, However, the ~ effect Is mild, and therefore appears most suitable for stress-miiof of relatively small focal lesions. E f f e c t s of Medial~Lateral D i s p l a c e m e n t Calcaneal Osteotomy on Contact Stresses in the Ankle S.J. S t e f f e n s m e i e r , C.L. S a l t z m a n , T.D. Brown

465

EFFECTS OF SUBCHONDRAL PLATE STIFFNESS ON CARTILAGE STRESSES Mente, P.L. and Lewis, J.L. University of Minnesota Department of Orthopaedic Surgery Much debate has centered on the contribution of subchondrel plate stiffness to the stresses that develop In the overlying cartilage layer. In order to examine the effect on the cartilage of lnoreaslng the slyness of the subchondral hone layer, a large strain finite element model for joint contact was developed using ABAQUS(verslon 5.2, HIbbitt,' Kadssco, & Somnsen, Inc., Pautuckst, RI). A canine joint was modeled as an axisymmetdc layered hemisphere with four tlssoel layers, cartilage, calcified cartilage, subchondral bone, and trabocular bone. A flat rigldl plate was used to load the hemisphere to simulate contact between two similar bodies. Linear-elastlc material Wopartias were used to model the three hard tissues trabecular bone (E=0.30 GPa, v-O.3, Layer thickness (t) - 30.85 ram), subchondral bot,,e (E-2.30! GPa, v=0.3, t - 0.15 mrs), and calcfflad cartilage (E-0.32 GPa. v-0.3, t - 0.12 ram}. Anl Ogden hyperelastic constitutive model, was used to model the non-linear stress-stralnl behavior of the cartilage layer (t=0.50 ram). In this model the strain energy Is expressed I In terms of arbitreW powers of the stretch. Values for the power terms were dstermined~ by fitting a 6-term strain energy function to expadmental strass-stretch data collected: overs stretch range of 0.3 to 1.5. SUffening of the s~bchondrel plate was accomplished by adding additional element layers to the Initial subchondrat bone layer. The subchondral bone layar thickness was increased by 50% and 100% and the resultant changes in the cartilage stresses calculated. Despite doubling the thickness of the subehondrel bone layer cartilage stresses increased only onthe order of 2% to 4%. These small increases in stress indicate that cartilage stresses are relatively Insensitive to stiffening of the subohondral plate, and that, therefore, initiation of a degenerative process in the subohondrel bone will not cause elevated cartilage stresses and subsequent cartilage breakdown.

S-98 466

Bone 469

COMPARISON OF CEMENT STRESSgS IN ALL-POLYETHYLENE AND METAL,-BACKED TIBIAL TOTAL KNEE REPLACEMENTS Frances B. Biegler, M.S., Christopher Wnag, M.D., and Jeffrey D. Reuben, M.D., Ph.D. University of Texas- Houston Health Science Center, Dept. Orthopaedics

CALCULATION O F B O N E ' S FADING MEMORY T I M E USING IN V I V a STRAIN MEASUREMENTS S.P. Fritton, C.T. Robin, and K.L McLcod Museulo-gkeletul Research Laboratory, State University of New York at Stony Brook

Tibial total knee replacement components manufactured entirely nf polyethylene are a less expensive alternative to cemented metal-backed components. Clinical failure of total knee replacement procedures is commonly a result of component loosening due to bone cement failure or failure of the bone cement - implant interface. This study examines the stress state of the bone cement following total knee replacement as a fimctinn of component type (all-polyethylene or metal-hacked) and cement mantle thickness using an anatomically detailed finite element model of a tibia with a tibial component. The finite element model's geometry was generated from CT data ofa cadaveric tibia implanted with a radiopaque plastic replica of an all-pely component. The tibia was scanned at 3ram increments. Element material properties were altered so that the tibial component was modelled as identically shal:~l all-poly, Ti-baeked, and CoCr-backed implants. For each implant type, the cement mantle thickness was set at 1.5ram, 3ram and 6ram. Models were loaded with a total of 2000 N distributed across the tibial plateau. All.polyethylene tibial components generate lower cement stresses than their metalbacked counterparts, especially near the janctioo of the bone cement and the implant post. This may contribute m longer implant life due to decreased failure at the bone cement - implant interface. Cement stresses do ant vary significantly with cement mantle thickness in either the allpoly or the metal-backad implants, indicating that there is no need to achieve a specific tibial cement mantle thickness in total knee replacement.

Although it is well aeeeptod that bone adapts to its loading environment, the specific algorithm bone utilizes to adapt has not been ascertained. Typically, models o f bone adaptation emphasize peak strain events during activities such as walking, running, or flighL However, for a large portion of the day, many animals stand and sit, activities that produce smaller strain magnitudes. What eontriburion these low magnitude strains make to bone morphology has not been determined. Recently, bone strain during pastural activity was recorded from the turkey ulna using/n viva strain gages at 16-bit resolution. Fourier techniques were used to analyze the spectral content of the strains. The strains were found to be inversely related to the frequaney, exhihidng a 11~ (0.8 < ct < 1.2) characteristic that has been observed in other complex, dynamic systems. We propose that these strain characteristics can be used to nalculate bone's "fading memory" time (i.e., the length of time that bone integrates strain information). Frequency spectra were analyzed over various time periods (e.g., 20 see, 1 min, etc., up to I hr) to determine which time period best predicted the peak strain magnitude events that occur during a day. These calculations suggest that bone has a fading memory time of approximately 5 rain. Other uses of spectral analysis to further understand bone adaptation are currently being explored.

467

470

BONE REMODELING AROUND FEMORAL TOTAL HIP COMPONENTS: 3D RESULTS USING A STABLE FINITE ELEMENT METHOD Timothy P. Hanigan, ScD, Frances B. Biegler, MS., and Jeffrey D. Reuben, MD, PhD, Department of Orthopedic Surgery, University of Texas Health Sciences Center at Houston

REMODELING SIGNAL INITIATION AND TRANSMISSION IN BONE: AN ELECTRICAL CABLE MODEL Dajun Zhang, Stephen C. Cowin and Sheldon Weinbeum Department of Mechanical Engineering, City College of City University of New York, New York, NY 10031

Bone formation and remodeling is affected by many mechanical and biological factors. By comparing purely mechanical remodeling predictions to clinical re*uits, we hope to lay the foundation for more detailed study of the biological factors and their interplay with mechanical factors. A 3D finite element mesh was developed describing a proximal femur with a total hip femoral component inserted without the use of bone cement. The femoral implant was designed based on CT scans of a patient's upper femur, and the CT data used in the femoral implant design was used to develop the geometry for the t'mite element model used. The femoral component was custom made to fit the patient and implanted into the patient, and clinical followup studies are now in progress. Two loading conditions were used in the finite element model: one simulating the mid-stance phase of gait, and another simulating stair climbing. The bone remodeling response was calculated for each load case by itself, and for the combination of the two loading cases. The bone elastic modulus was taken as 17 GPa, the prosthesis modulus was taken as 200 GPa, and all materials were taken as isotropic with a Poisson ratio of 0.3. Set points in the remodeling rules were varied from I00 I/m s to 1000 JtmJ.The results showed substantial bone density decreases as set points were increased, as expected from prior results. Several features common to remodeling responses seen around total joint femoral components were seen. Stair climbing loads resulted in more bone adjacent to the prosthesis, and combination loading (each weighted by one half) showed still more bone. We concluded that several features of the remodeling response could be simulated using a mechanical model, and that combined loading requires more bone than a single load case in our remodeling simulations.

It is proposed that the connected network of osteocytes, csteoblasts and bone lining cells in bone tissue is capable of remodeling signal initiation and transmission. A cable model is formulated to estimate the spatial distxlbution of intrac~|inlar electric potential and ettrrsnt, from the cemtmt line to the lumen of an osteon, as the frequency of the loading and conductance of the gap junction is altered. Here the intraselinlar potential and oarrrnt are driven by the mechanically induced s~a'alngenerated streaming potentials (SGP's) produced by the cyclic loading. The model differs from earlier studies in that it pursues a more physiological approach in which the micro-anatomical dimensions of the coanexoo pores, asteecytic processes and the distribution of cellular membrane area and capacitance am used to quantitatively estimate the leakage of current through the osteoblast membrane and the relative resistances of the nsteecytic processes and the connexons in their open and closed state. The model predicts that the cable demonstrates a strong resonant response when the cable coupling length approaches the osteenal radius. It also predicts that the pore pressure relaxation time for the draining of the bone fluid into the osteooaI canal is of the same of order of the characteristic diffusion time for the spread of current along the membrane of the esteocytic processes. This coincidence of characteristin times produced a spectral resonance in the cable at 30Hz. These two resonances lead to a large amplification of the intracellular potential and current in the surface ostanblasts which could serve as the initiating signal for ostcobtssts to conduct remodeling. This resonance might also explain why live bone appears to be selectively responsive to the mechanical loading in a specific frequency tango (15-30Hz), as demonstrated in the experiments of McLcod and Robin (1995) for several species.

468

47i

INTERCEPT MEASUREMENTS ON CANCELLOUS BONE SHOW A NON-RANDOM

EXPERIMENTAL DETERMINATION OF RESONANCE IN THE HUMAN I.,4-L5 VERTEBRAL SEGMENT J.C. Fritton, C.T. Robin, M- Pope*, T. Hanssen* and K.J. M c ~ Museulo-Skeletal Research Laboratory, State University of New York at Stony Brook *Department of Biomedical Engineering, University of Iowa, Iowa City, IA qnstitute for Orthopaedics, Saldgrenska Hospital, Gotbenberg, Sweden

MICROSTRUCTURE Timothy P. Harrigan I, Anders Odgaasd 2 and Jeffrey D. Reuben I (1) Departmcot of Orthopedic Surgery, University of Texas Medical School at Houston (2) Department of Orthopedics University of Aashus, Denmark The processes which control the microstracmre in cancellous bone are not well defined, and the degree of order within the tissue is not well understood. In this study, we.tested whether a two-state Markov process could model the linear intercept measurements in cancellous bone. A two-state M~kov process is fully defined by the density of the tissue and the mean intercep~ length in the scanning direction. We measured the difference between scanning results for actual bone tissue and those predicted by a two-state Markov process. The tissue sample used was taken from a proximal tibia and digitally quantified using the methods of Odgaard et al. For line lengths spanning 2 voaels to 400 voxeis, the number of intercepts along lines starting from 10,000 randomly selected points within the microsmtcture were measured, and histograms of intercept numbers were made. Two-point correlation measurements were made similarly. The clearest differences between the predictions of the Markov model and the experimental data was in small numbers of intercepts. Also, The likelihood of 2 and 3 intercepts was overestimated at small line lengths and underestimated at larger line lengths by the Markov process. These observations were independent of whether the scanning lines started on bone or marrow space. The two-point correlation me~rements showed a significant deviation fi'om ~at predicted by the Marknv r~ode~ at se~ lengths below apprexima~ety 2 MIL. The experimentally measured probability for both ends of a scan line to be on bone showed an initial decrease which was fmter than that predicted by the Merkov process, and oscillatory behavior which was not predicted by the Markov model. By comparing actual geometry meesarements to a minimally controlled pr0babilistic model, we can estimate the range of mechanical or biological coupling between trabecuine at 4 to 6 MIL.

Definition of resonances in the wnight-beasing portions of the skdeton is a necessity for understanding the influence of dynamic external loading. While local resonances may result in excessive morion, they may also also have value in certain clinical applieatinas. Recently, Goel et al. [1994] found axial resonant frequencies of 17.5 and 27.7 H2 in motion segments of the human spine utilizing modal analysis of a finite dement model of the L4-S 1 vertebral segments supporting a 40 kg mass. As part of a larger study, accelerations resulting from vertical whole body vibration were directly measured at a spring-supported platform surface and at the 1.4 and I_5 vertebrae of a single standing human female (63 kg) volunteer. The platform was driven at 13 N ~ and data collected at 500 Hz. The transfer function of the signal was calculated from L5 to L4 to explore the mechanical behavior of the motion segment between 10 and 35 Hz. At 19 Hz a peak in the transfer function was obtained indicating a resonance Ixtween the two vertebrae. These data largely confirm the results of Goni et aL with differences that may refioct some of their nck'aowledged limitations, such as d~e lack o f damping. The effect of a lower mass and viseoolastic damping should be a resonance shift to higher frequencies and lower magnitudes. Geel, V., Park, H., and Kong, W. (1994) J. Biomech. Eng., 116: 377-383.

Bioactive Materials 472

S-99

474 CALCIUMPHOSPHATE MODELS OF BONE A.L. Boskey, N.P. Camacho, S. Gadaletu, E. Paschalis and F. Botts The Hospital for Special Surgery 535 East 70th Street New York, NY 10021

Bone is a composite of caldum phosphate mineral, oriented on an organic (collagen) matrix. Physico-chemical characterization using x-ray diffraction, high resolution electron microscopy, solid state N1VIRand Fourier Transform Infrared (NTIR) spectroscopy have characterized the calcium phosphate phase as a poorly crystalline, carbonate-enriched, apntitu containing HP0, groups. Non-apatitic phases such as OCP brushite, TCP and whitlokite have not been detected. Recently, coupling a light microscope with an ~ spectrometer has enabled mapping of both spatial and agedepended variations in bone mineral content, acid phosphate and carbonate content at 20ft resolution. Such studies show that as the tissue matures (in the absence of discose or dietary alterations) the content of acid phosphate and the non-stuichiometry of the apatite decrease, while the carbonate to phosphate ratio increases. The synthetic analogues of bone mineral made by direct precipitation, conversion of amorphous precursors, liposome mediated mineralization or precipitation on biometric templates, and other low and high temperature techniques each differ in one or more parameters (size, solubility,composition,orientation, architecture, mechanical properties) from bone mineral. Mineral formed in osteoblast cultures where remodeling does not occur also differs in size and composition from physiologic mineral. It is suggested that the different analogues may be appropriate materials for biocompatibility and mechanical studies. Supported by NIH grants DE04141, AR41325 and AR37661

473 HYDROXYLAP~/IIT/~ C O A T I N G S : A C E R A M I S T ' S V I E W G a r y Fischman I n s t i t u t e for B i o e e r a m i c s NSF I n d u s t r y U n i v e r s i t y Center for B i o s u r f a c e s NYS C o l l e g e o f C e r a m i c s a t A l f r e d U n i v e r s i t y H y d r o x y l a p a t i t e - Calo(PO4)6(OH) 2 Is a m a t e r i a l t h a t h a s b e e n used in devices monolithically and as a coating. Coatings are u s e d , I n h e r e n t l y . a s s t r u c t u r e s t h a t will a l t e r t h e p r o p e r t i e s o f a material and its surface. The most common, though not the only. coating technique for this material is plasma s p r a y i n g . I n t h i s t a l k . c o a t i n g s i n g e n e r a l , h o w t h e y effect a material, and the plasma spray coating technique in particular will be discussed leading to a ceramlst's perspective on the plasma spray process.

PROPERTIES OF HVOF SPRAYED CALCIUM PHOSPHATE COATINGS H*man J.D., Chlthlr K.K.*, Lucas L.C. Univ. of AI. at Birm., Dept. of Biomed. Engr.; *Univ.of AI. at Hunts., Dept. of Chem. Engr. Ilatreductlon High velocity oxy-fuel (HVOF) thermal spraying has been used for the past ten yeags to provide coatings within the s u p e r - c o ~ and wear-reslstanca coating industries. This Studywas conducted to determlns the chemicaland structural diffegcocasbetween HVOF and plasma sprayed hydmxyapatlte coatings. Materials and Methods Hydmxyapatite (HA): powder was applied to titanium ('ri) coupons by either HVOF thermal spraying or plasma' spraying using typical application parameters. The cwstallinity and phase content of the coatings were then analyzed using diffuse reflectance fourier transform infrared spectn~py (DRIFTS) and x-ray diffraelion(XRD) teclmiques. The phosphate region(900-1200 ema) of the FTIR sl~ctmand the 20-400 20 range ontheXRDdilEtactogramsweredeconvolvedand curvefitted to determinethe percent ~stallinltyand phase content of the coatings. Results g~ults of the FTIR sad XP.D ~Frm H A C'~allmi~

| 61.71 :i:7.02. n=l$

~ 45.12:~2..97.a - 15

HACr~llinit~ $7.49.k~.0S. n*9 44.~'2~'~.40.a-9 Lat~ep~ a=b=9.41~-0,05.r a~) o=be.9.~.ll. ~6.90:~0.10.j=9 DiscussiOn DRIFrs analyses indicatedthat both coatings contained the same phases, while XRD indicated the presence of calcium oxide in the plasma sprayed coatings. The HA ctystallinitieswere significantlydifferentwith the HVOF process providingthe more cwstaUine coating. Future Studies The difference in ctystallinitysuggests a difference in bond strength. Therefore, current and future research includes four point bend mechanical studies to determine the main at whichthese coatingscrack.

S-100

Tissue Response to Implants

475 TISSUE INTERFACE ANALYSIS J.E Lemons The University of Alabama at Birmingham Devices, tissues, and clinical records have been collected and analyzed from musculoskeletal surgical procedures since 1970, with totals now exceeding 3,000. Research emphasis has been on interactions along the device-to-tissue interfaces, primarily synthetic metals, ceramics, and polymers in contact with hone. Investigations Include extended studies on the roles of elements and forces that are transferred during load bearing function. The observations from retrieved devices have often resulted in hypotheses about cause-effect relationships which then lead to laboratory-based MS, PhD, and MD or DMD Resident projects. The presentation will provid e examples of implant-to-tissue interactions that have been analyzed in terms of intended and unintended clinical outcomes.

476

PARTICLE-MEDIATED EFFECTS ON CELL FUNCTION S.R. G O L D R I N G , N E W E N G L A N D B O N E A N D J O I N T I N S T I T U T E H A R V A R D M E D I C A L SCHOOL, BOSTON, H A 02215

Multiple different factors determine the biological responses to foreign organic and inorganic particulate materials. These factors can be broadly considered under two general categories, one of which relates to the properties of the particles themselves, including particle size, shape, surface topography, relative hydrophobicity and composition; and the other to the unique individual host responses. In vivo, the tissue reaction associated with particulates exhibits features of a granuloma. Granuloma represent a host attempt to isolate and/or sequester foreign materials that are resistant to degradation or removal. Reactions to particles are classified into non-iramune granuloma and immune granuloma depending on the nature of the cellular response. Immune granuloma exhibit features of delayed type hypersensitivity and arc characteristically seen with organic particulate stimuli. However, inorganic materials may a/so elicit hypersensitivity reactions and this may markedly alter the nature and outcome of the inflammatory response. Recent evidence demonstrates that much of the pathogenicity of granulomatous reactions is related to the release of potent inflammatory products from cells within the granuloma as a consequence of direct interaction with the particles or after particle endocytosis. Insights into the nature of these products and the mechanisms associated with modulation and activation of cells by particulate stimuli are of critical importance in reducing the adverse reaction to foreign particulate materials.

477 PROTEOGLYCANS SYNTHESIZED BY OSTEOBLAST-LIKE MC3T3-E1 CELLS ON CP-TITANIUM. M.M. Klinqert, F.RahemtullaZ, C.W. PrinceS, and LC. Lucasl. Departments of 1Biomedical Engineering, ZOral Biology, and 3Nutrition Sciences, University of Alabama at Birmingham, Birmingham, AL 35294. Ultrastructural studies of retrieved implants and of bone cells on implant matedals have suggested the presence of a thin layer of extracellular matrix at the bone-implant interface. Histochemical staining of this layer has indicated the presence of proteoglycans, a class of complex, anionic molecules thought to play a role in controlling mineralization. These findings had not been confirmed by biochemical analysis. In this report, osteoblast-like routine MC3T3-E1 cells were cultured on tissue culture plastic and on cpTL 3SS. labeled macromolecules extracted from the cell layers were analyzed by ionexchange chromatography and by enzymatic degradation followed by gel filtration. Greater than 90% of the radioactivity from both substrates required 0.5-0.7 M NaCI for elution from DEAE-Sephacel. Gel filtration of the extracts on Sepharose CL-4B yielded simi[ar broad size distributions for both samples. Digestion with chondroitinase ABC degraded 45% and SS% of the label from cells on plastic and cpTi, respectively. Digestion with both chondroitinase and heparitinase degraded essentially all of the label from both samples. The results indicate that these bone cells on cpTi produce a mixture o f chondroitin/dermatan sulfate and heparan sulfate proteoglycans. We also observed that free 3SS-sulfate binds far more tightly to cpTi than to tissue culture plastic. Supported in part by the National Science Foundation.

478 EFFECTS OF METAL IONS ON OSTEOCLAST FUNCTION I N VITRO K.C. Gillonwater and D A. Pulco Center for BiomedicalEngineering, Unlvetrdty of Kontucky, Lexington, KY 40506 Introduction: If metal ions present in the peripmsthetic environment affect the formation and functions of bone-forming estooblasts orthe bone~resorbing estsoclasts, thoy could play a role in loosening of orthopedic implants. The objective of this study was to investigate sublethal effects of metal ions found in commonly used ogthopedie implants on the formation and activity of nstceclastio cells. Materials and Methods: A bone marrow culture system was used to investigate the effects of ions on the formation and function of o~tooclastio cells derived from hematopoietin precursors in ~itm. Bone marrow cells were harvested from juvenile rats and cultured on devitalized wafers of bovine bone. Individual atomic absorption standard solutions of Ti4*, AI3+, V5+, Co2*, Cr~, and Mo6* and combinations for Co-Cr-Mo and Ti-AI-V alloys were used to represent biomaterlals in ionic form. Ion concentrations were chosen as 0, 1~, and 10% of TCS0 (50% toxic concentration) for each ion. Total protein content of cnll lyesto samples was used as an indicator of cell growth. After removal of cells, bone slices were stained with toinidine blue, and re~rption features present in bright field light microscopic images were quantified using semiautomated image analysis. Remits and Discussion: According to total protein levels, the cells proliferated in the presence of all ions examined, with the exception of Co2*, indicating that the ions were sublethal over the four week period of culture, as desired. Cells cultured with 1% TC50 ions did not show significant changes in levels of resorption activity as compared to control onllures. However,inlzibited resorption activity was observed in somecultures ex33nsedto ions at 10%TC50. Theseresults demonstratethat ionsat sublethal concentrationscan affect the differentiation and function of ostsoclastic cells.

479 DETECTION OF BIOTRANSFORMED SILICONES WITH NMR Leoncio Ganido, Bettina Pfleiderer, Leo L. Cheng and Jerome L. Ackerman Biomaterials Laboratory, NMR Center, Department of Radiology, Massachusetts General Hospital and Harvard Medical School, Charlestown, MA Silicone, generic name of synthetic polymers with a backbone of s i n bonds (polysiloxanes) and organic side groups (aliphatic and/or aromatic), is one of the most widely used materials for biomedical purposes, primarily because it was considered to be nonbiodcgradable. However, silicone has often been found in adjacent tissue and regional lymph nodes in patients with implants, suggesting that the prostheses might degrade over time. Nuclear magnetic resonance (NMR) is able to provide detailed information about a substance's chemical composition and molecular architecture, both reflected in the chemical shift spectrum, as well as about the molecular and polymer chain dynamics related to NMR parameters, such as the spin-lattice (T1) and spin-spin ('1"2) relaxation times. Thus, we have used 29Si N M R spectroscopy to investigate the migration and chemical modification of silicones in animal models and humans. This work demonstrates that silicone migrates from the implant site to adjacent tissues and distant organs, such as liver and spleen. In addition, it shows that silicone is not stable but undergoes chemical transformation in a biological environment. 29Si N M R spectroscopy of animal tissues exposed to polysiloxanes and blood samples from women with silicone gel-filled breast implants show the presence of chemical unchanged silicone, partially hydrolyzed silicone, silica and other silicon compounds.

Cancellous Bone Structure: Mechanical Properties 480

482

Marrow Contribution to Dynamic Stiffening of Cancellous Bone over a Wide Range of Porosity. Yves P. ~'ramon and Stephen C. Cowin Department of Mechanical Engineedng, The City College of the City University of New York, New York N.Y. 10031

IS T R A B E ~ BONE TISSUE DIFFERENT FROM CORTICAL BONE.,TISSUE? X. EdwaM Guo mul $tevett .~ Goldgein Orthopa~c R.e~meh ~ The U n i v e c ~ of ~fichigan, Ann Arbor, } , ~ g ~

A review of experimental measurements of pem~eability of cancellous bone is reported. A poroelastic model is used to establish the extent to which the marrow shares the loads with the trabecuiee in cancellous bone. The results indicate that the diffusional constant associated with the poroelastic equation is too small to produce a significant transient effect except for very high load rates. This result is applicable over a wide range of porosity and implies that the hydraulic stiffening in cancellous bone may be insignificant within normal physiological loading and loading rates.

481 WHAT STEREOLOGICAL METHODS OFFER THE GREATEST HELP IN QUANTIFYING TRABECULAR STRUCTURE FROM BIOLOGICAL AND MECHANICAL pERSPECTIVES? M.B. Sclmffler, D.P. Fyhrin, A.M. Parfitt Bone and Joint Center. Henry Ford Health Sciences Center, Detroit. Michigm~ 48202 Tmbeculm bone is a complicated interconnected network of rods. plates and sheets of lm'nellar bone tissue. Understanding its structure in terms of the elements that comprise the network and their a~hitectural relationship to each other is key to understanding both the biological and mechanical attributes of Irshecular bone in health and disease. Osteoblastic and osteoclastic activities in trabecutar bone occur on bone surfaces. Surface activities in tuna regulate tmbecutar bone mass and architetaure, parameters which have great importenee to mechanical funedon. Bone histomorpbometry, based on classical stcreoingy, has been used effectively and extensively to characterize bone surface activities. These techniques were "designed" to measure the surface.based cellular parameters which are known to be important in bone physiology. Recent significant advances in stereological theory eliminate the bias in measuring anison'opic structures, making sterenlogy an even more powerful tool for studying the biology of caneelinus bone. Beyond measuring bone mass, however, the architectural parameten which need to be measured, and accordingly, the morphomet~ic tools which are needed to describe the mechanical properties of trabecular bone, are less clear. Histomorphometric parameters designed to quantitate activity do not yield direct information on the shapes of trshecular elements or architectural interrelationships, which are key factors influencing the mechanical properties of trabecular bone. Those stereological parameters which encompass pertinent information about mechanical properties of trabocular bone are largely correlative and were not designed to describe bone structure as related to mechanics. Recent applications of v.CT and MR microscopy to trabecutsr bone provides 3-D data sets; to date morphometric descriptors from these data have been extrapolated from standard stereologieal applications. Availability of 3-D data presents an opportunity to extract new information about trabecu[ar bone architecture, using morphometric tools designed specifically to measure mechanically relevant structural properties of tzabecular bone.

S-IOI

Advauee8 in anden~ding sge mlau~d bone fragility, ccmpotatlmud medeling taxi bone dsme adaptalion emphad-- the ueed m ~ bone ~ at a fl~ue iews Mot~hdiogk.ally. Irabecul~ and cortical fi.~ues appear to have significantly diff(a~tt ixopeati~ Cot'tical bone ~ comlx)~d l~qua'ily of nsteons while aldxgul~ boue is cJunctet/z~ by pure htme41~ organization. ~ tisme may be less ~ than cxmical lisme and data suggests that trsbecular bone is mtr.h mote ataive in remodeling and tin'novel', gece~ data c,oncetning mechanical w o p e n ~ of r or trabecular ~ have presanted camlktin$ msul~ Experimantal or omnputattoanl medmds have been developed to quendfy the d~ue mothdea ~ individual wabe~tlan including buckling, untsxtsl tension, tl~ee or four point bemltag, ultrasonnd and finite elemant calmladm~ The modales ~ Irabem~ lisme has been relx~ed to range from 1- 14 GPa. "I]fis largn vm'intlon in medulm values may be the re~dt of tinting _me. _~ho,42,or ~.~t~-~ w'~.~,,,,u-t,._~_t ~ .m_;.,v~,_ _t ~e'~n.,__~.td i f f - ~ "-.-.s. ~ .'a...~. ~ :'~...ent howeve*, presents an emerging comens~ that trabecular tissue b 20-30% le~ taiff than oart~cat bone t~as~e. The data also indicates that tis.me miueral de,.~ity alone cannot exptsin the m ~ variance in m ~ . d m values, m g g e . ~ g a signir'r.am ~ n c e on mkammgmral featmm (lacunae, cement linen, etc.). The t~elts from fatigue ext~'imen~ of both Uabeculm and cortical bone tiasne* demonstrale that cortical boue tissue may have similm"fatigue ttslganee m~ trabocuinr bone ttsme despi~ the diffea'ev,~* in tissue morphology mul modulns. This ~ g may emphasize the imlx3nam~ of tissue u l ~ in damage a~umulation. Pteliminmy data on trabecular and cordcel tissu~ from vertebral bodies have d ~ a signifiumt age-related vadatloo in tissue modulus and higher values in ccetical l~_me when compared to trabocutsr tissue. Continuing studies ate focused on quantifying the tissue m ~ t m e aad correlating these ~ to physical propeni~ mcastwed utilizing nanoindaatation arm ~tignr expe~imems ia an effort to more cle~ly delineatest~ucon~function reladonshil~ in bone tissueand potential impli~ans in agn-mlatodbone f~'a#Lityaad adaptetice.

S-102

Cancellous Bone Structure: Trabecular Architecture

482a THE RELATIONSHIPBETWEEN CANCELLOUSB(~IE ARCHITECTURE AND MEC2,'IANICALPROPERTIESAT A CONTINUUMlEVEL C. H. Turner Bk)medlardcsand Biomal~lals ResearchCer~r, IUPUI,;ndklnapols,IN 41~0~ t~tny naU'a m ~ s ha~ a slgr~r~nt amountof porosity. Examl~s Incline grardar materials(soil sand], 9eolo9~ saids (~od~), and ~lul,~" m a t e ~ (*~xl and b(me). Forlhese matwi,~s,g'~ m e d ~ strength and rinds'./are ,wersely related to the porosity. 1hiS rela~nship generallytakes the form of an expomn~ or powedaw. When porousma~eriaJshave9eomeldcaladso~opy,a seco~l measureof Io~1 structure,called ~ fat~ ellipsoid, is necessa~ to characterizethe ~ee-dimensional med~mical pmpertes. Fatx;r measurenumthas beenemp~o~l ~ chafactedzationof granu~ mate~Js, soil and. more r e o ~ , camelkx~ bone. C a r o m brae is a ~ ~ t y , ce,u~' m~e~ ,~at is gemra~ considered to be or~olxop~c,Le. it has different elas~ prope~es in each of Ihree orthogona~directions. Sb'uts,called ~ , Ihat make up cancellousbone tend to align wilh pdndpal stress~ ~~ lo~e. In a lyplca] pieceof canceilousbonefrom the kneeJoint,~le Young'smodulusIn I ~ drec~on~ ~ struts is o~r five ~mesgreater than the bate;versemodulus. ~ l c , J s sb~r is characterizedby measuresof apparent der~ty - g]e v,sightof a bo~ed~videdby Its bul~volume;and fabric - gle ~ variation in a stereologicalparametercalled mean in~ercepll e r ~ . Bonestrengthand dgi~ty generagyare corr~ated wflh apparent dendty raised Io some power ranging from one to Ihme. In most studies, mechanical proper~esare best ~ by apparentdensity raised to the second power. Fabric is best representedas a secondrank tensor. Usingtensoralgehia,1henine independentelasticcoeffidents of an orlhotropic material have been presentedas lunc~ns o! labric and apparentdensity, "l~ese functiors contain emplrical r ~at have beendeterminedby experiment. However,Ihe unk'ersalityof expedment~ly-denvedlunclions is limited because o! the vadabilityof Ihe basic tissue properties of bone w~th~ e and or,~aase. "1]~ y~eldsbeng~ of cancellousbone is ~lso , ~ l y correlatedwi~ f l a , , c ~ of labdc and apparent dens~y,yet ~ yiek:lstrain is relaSvelyconstantregardless of loading ,'nrection. Car~e~eusbonecan sm~n rnom s~as'~cs'~a~nin compres~on~an In tendon,~ ~ s ~ ~ m s ~ m ~ be Independer~of f~bric and ap~amnl~nsily. Theseobservaf~onssuggestthat faflu:e of canceflousbone can be characterizedby a maximumsl~n criterion.

483 WHAT IS WOLFFS LAW OF TRABECULARA R C H - - ? Stephen C. Cowin The City College of CUNY Wolffs trajectodaJ theo~ of Irahaculararchitecture, and its hlstodcal development, are reviewed from the perspective of the mechanical concepts involved, namely the continuum model and stress trajectorias. Over a century after Wolffs book on bone, students of bone structure ask if Wolffwas contct, do ~ l a e align themselves with pn, cipal stress directions? In the last half century, the scholars reinvestigatingthe foundation of Wolffs law have been less convinced that Wolff was correct. Though none say he was totally wrong, most add caveats to Wolffs statement of his law. It is easy to show that one set of principal stress directions must lie within Irabeculae in any trab~olar architecture, and therefose these principal stress directions will lie within ~ u l a r as trahaculae reorient. Wolffs law offers little guidance in answering the question of the relationship between the enviroumentalstresses and the process of trabecular reoriantalion. The fail~y of the rigid constructionist view of Wolffs trajectorial theory, which requires that the trabeculan of cancenous bone intersect at right angles, is explained.

484 A NEW EXPERIMENTALAPPROACHTO THE STLrDy OF CONTINUUM LEVEL STRUCTURE/FUNCTIONRELATIONSHIPSIN TRABECULARBONE Brian K. Bay Orthopaedic Research Laboratories, University of California Davis The relationship between trabecular bone, structure and function has been studied for over a hundred years. Progress has been paced by the availability of techniques for the determination of both quantities. Initial work employed 'by eye' determination of trabecular arrangements and graphical stress analysis techniques. The most recent work involves quantitative stereology for the determination of structure, and finite element analysis for estimation of functional response. These methods offer considerable precision, but assumptions required for the determination of material properties bring into question the accuracy of finite element analysis. A new experimental method for the measurement of strains at high spatial resolution offers an alternative approach to the problem. Samples of trabeculat bone approximately 5 m m thick are radiographcd in unloaded and loaded states. The radiographs are digitized and a computer correlation procedure run for the measurement of a detailed displacement vector field. Displacements are filtered to reduce high frequency noise, then the sVain tensor field is calculated by finite element postprocessing techniques. Principal strain vectors derived from the strain tensor field can be compared directly with the localized trabeeular structure of the sample at a high level of spatial resolution. Principal stresses cannot be derived without constitutive data, but the level of detail is suitable for direct comparison with finite element results. There is no upper limit on sample size: a section from a whole bone, or even two bones in contact, can be analyzed. Extension of the technique into three-dimensions, given tomographie data sets from unloaded and loaded samples, appears feasible.

Cancellous Bone Structure: Mathematical Theories 485 INTRODUCTION TO FINITE ELEMENT BASED SIMULATION OF FUNCH'IONAL ADAPTATION OF CANCELLOUS BONE R.T. Hart and S.P. Fritton* Department of Biomedical Engineering Tulane University, New Orleans, Louisiana *Presently at the Musculo-Skeletal Research Lab State University of New York at Stony Brook, Stony Brook, NY This introductory presentation will begin to explore the use of finite element methods as a tool to understand, describe and ultimately predict the adaptation of cancellous bone to mechanical usage. The specific questions to be addressed during the session are: 1. What have we learned about mechanical control of cancellous bone architecture from finite element models of bone remodeling? 2. What are the limits of continuum mechanics and finite element methods insofar as trabecular geometry and principal stress calculations are concerned? A survey of the progress that has been reported in the literature for developing finite element based methods will be presented. Previous numerical implementations for architecture simulation are a result of choices regarding issues such as material scale size (continuum and/or trabecular) and time scales. Highlighted issues will include the various hypothesized mechanical measures thought to be largely responsible for changing cancellous bone density and orientation, and limitations and contributions of work to date.

S-103

S-104

Mechanics and Electrokinetics of Cartilaginous Models

486

489

TENSILE TESTING O F PORCINE PATELLAR TENDONS: EFFECTS O F STRAIN RATE GW Ringer, IS Wayne, and WA Zuelzer Orthopaedic Research Laboratory, MCV Box 980694 Medical College of Virginia/Virginia Commonwealth University. Richmond. VA 23298

VARIATION IN CARTILAGE STREAMING POTENTIALS WITH

DEPTH FROM THE ARTICULAR SURFACE Albert C. Chen and Robert L. gah Department of Bioengineering and Institute for Biomedical Engineering University of California, San Diego, La Iolla, CA 92093-0412

Strain rate is one factor that may affect the mechanical properties of patellar tendon (vr) determined during uniaalal testing (1.2). The purpose of this study was to detewalne the mechanical properties of porcine PT during tmiaxial testing at two strain rates. Seventeen feesh, skelctally immattwe porcine (Yorkshire, 14-16 woeks) knees were isolated and dissected to leave only the patelia-PT-tibia complex intact. The patella and tibia were clamped in a closed loop hydraulic testing machine (lnstron, Model 1321) so that the PT was parallel to the direction of loading. The medial and lateral portions of the tendon were removed to isointa the centrni section. A premed of approximately 1 N was applied to the specimen. A rectangular cross-section was assumed, and tissue dimensions were measured with digital cnlipors. A differential variable reluctance transducer (MicroSwain, Burlington, VT) with a gauge length of 10 mm was inserted to measure tissue elongation. The specimen was conditioned for 20 cycles of a 1.5 mm amplitude triangular wave at an elongation rate of 20 mm/min. The 1 N preload was again applied to the specimen before loading to failure at either 20 (n=6) or 200 ram/rain (n=l 1). Tangent modulus, ultimate stress and strain, and failure mode were noted. Cnlculated strain rates were 0.23 :t: 0.06%/s and 1.95 • 0.41%/s. Swain rate significantly affected failure mode of the tendons. Seven of the eleven specimens tested at the faster rate failed midsubstance. All other specimens failed by avulsion. Tangent modulus was also significantly greater at the faster rate (1246 • 77.9 MPa) versus the slower rate 0 0 7 4 • 108.4 MPa) (p<.001). Ultimate stress and strain were 81,4 • 3.00 MPa and 15s • 6.06% for specimens tested at the faster rate that failed midsubtance. Therefore, strain rate is an important consideration in the tensile testing of porcine IT, 1. JOR, 11: 5g~67, 1993. 2. AmISnMed. 22: 328-333, 1994.

The compoaldon and stmcmm of cartilage varies with depth from the articular surface. Extension of homogeneous theoretical models that describe the mechanical and electromeshzeical behavior of cartilage requires experimental assessment of the depth-dependence of tissue properties. In the uniaxialconfined compressioncreep test,the linearporoclastic(biphasic)model combined with electrokinetic theory predicts that during the onset of an applied stress, an electrical streaming potential, AV, will be developed and related to the applied compressive stress, AV=ke.AO, where I% is the electrokinedc coefficient.To determine the depthdependence of ke, successivecartilage disks, 4.8 m m diameter x 0.5 rum thick, were harvested from just under the articular sur fac.e of the femoropatellar groove of four adult bovines. Each disk was tested in the uniaxial confined compression configuration by applying a tare stress of 0.01 MPa and then measuring the streaming potential response to a series of 10 s pulses of 0.05 MPa compressive stress. The ke of cartilagn just under the articular surface was significantly less than that of cartilagn from the despe* region (-0.69i-0.18 mV/MPa vs. -0.32s mV/MPa, mean:i:SD, n=8, p=0.001) while the preteoglycan density, measured as dimethyimethylene blue dye-reactive material, was lower in the tissue layers near the articular surface than the deeper layers (39.4+4.4 mg/g vs. 55.8:1:19.4 rag/g, p<0.05). The validation of depth-dependent electromechanical models of articular cartilage may be important in assessing the proper'des of normal, arthritic, or repair cartilage.

487

490 EFFECTS OF DEGENERATION AND AGING ON THE MATERIAL BEHAVIOR OF T H E NUCLEUS PULPOSUS Iatrtdis JC, Sutton LA ~ Wetdenhaum M, Mow VC Orthopaedic Research Laboratory, Columbia University, New York, NY "Department of Biomedicnl Engineering, Duke University, Durham, NC

We hypothesize that the marked morphological and compositional changes seen in the nucleus pulposus (NP) of the intervertchral disc with degeneration are associated with concomitant mechanical changes. The objectives of this study were to determine the shear behavior of the NP, and to investigate changes associated with degeneration and aging. A new protocol was developed to determine the viscoelastic behavior of normal and degenerate NP specimens in torsional shear under dynamic and transient conditions. N'P specimens were harvested from 21 human lumbac L2-5 discs ranging in age from 16 to 87 years and graded for degeneration (normal, n=13; mild, n=l 1; severe, n=I 1}. Cylindrical (dia.=g ram, thick=1.44 ram) NP specimens were prepared and tested in torsional shear. Shear tests consisted of a dynamic frequency sweep (Fo = 0.01 red. I s ~ s 100 red/s) and 3 ramp stress relaxation experiments (Fo=0.05, 0.10, and (3.15 red; to=25 ms). The magnitude of the dynamic shear modulus IG*I and tangent of the phase angle (tang) increased with frequency (p<0.05) and were significantly affected by degenermion with values for the normal and severely degenerate specimens ranging from 8 to 53 kPa, and 0.45 to 0.32 for IG*I and tang, respectively. The stress response of the NP to the ramp shear strain was characterized by a large initial peak stress followed by a rapid relaxation to stress values near zero at tbe termination of the test. The instantaneous shear modulus (6-22 kPa, normal to severely degenerate) increased with degeneration while the water content decreased (82-76%). In conclusion, the relatively large shear modulus of the NP and value for tan~t
Supported by the Anhriris Foundation, National Science Foundarlon, and Whitaker Foundation.

EFFECTS OF MECHANICAL STRESS AND DEFORMATION ON PASSIVE ION TRANSPORT THROUGH HYDRATED PERMEABLE TISSUES W. Gu. W.M. Lal and V.C. Mow Orthopaedic Research Laboratory, Columbia University, New York, NY Passive ion transport through a hydrated permeable (charged or non-charged) tissue is governed by the gradients of electrochemical potentials of ions and solvent. Using the theory in [ 1,2], a formula minting the ionic current of the a-species (In) to the fluid pressure (p), osmotic pressure Ur), tissue dilatation (e), Nernst potentials (E#, defined as -[RT/(z#Fc)]Vln(F~c#) [2]), and electrical potential (~) is derived: I a = -ba(Vp * VTr + BwVe) + Xhufl(E ~ - V~) (summation over ~ for all ion species),

(1)

where ba, Bw, and ha# (with ha~=haa) are the material properties of the tissue. The coefficiem b a represents the cffzet of the gradients of fluid pressure (Vp), osmotic pressure (V~r), and dilatation (Ve) on the ionic current. The coefficient ha# represents the effect of the Nernst potential of the ~th species on the ath ionic current. Thesd material coefficients (ba and ba/~) arc functions of tissue dilatation as well as ionic concentrations. For excitable membranes, these coefficients are potential-depeedcnt as well [3]. For cases where no gradient of solvent chemical potential exists and the interactions among different ionic species arc angligiblc. Eq (1) can be reduced to I u = ga(Ea - VV), where ga is the ionic conductivity. Note that this equation has been widely employed to study ionic channels of excitable membranes [3]. Equation (1), derived in this study, provides a basis for the quantitative study of the effect of mechanical stress and deformation on passive ion transport through chargnd or non-charged, hydrated permeable tissues. It could also be used to study the effect of mechanical stress and deformation on signal conduction in nervous tissues. REFERENCES: [ I I Lai WM. Hen JS, Mow VC: ./Biomech Engn#113:245, 1991. [2] Gu W, Lai WM. Mow VC: ASMEAdv Bioengng28:217, 1994. [3] Hille, B (1992) Ionic Channels of ExcitableMcmbrane~,Si~uet Associales,Sunderland,M~s.

488

491

A NEW APPROACH FOR THE REPAIR OF LARGE ARTICULAR SURFACES JS Wayne, CL McDowell, R Tuten, KJ Dailey, HH McGuire Jr. McGuire Veterans Affairs Medical Center, Richmond VA 23249 & Orthopaedic Research Laboratory, Virginia Commonwealth Univ., Richmond VA 23298

VARIABLE FREQUENCY-WAVELENGTH S U R F A C E E L E C T R O K L N E T I C SPECTROSCOPY FOR DETECTION OF CARTILAGE DEGRADATION Scott Berkenblit,David Bombard, Steven Treppo, EliotFrank, and Alan Grodzinsky Continuum Electromechanics Group, Center for Biomedical Engineering, Department of Electrical Engineering and Computer Science: Massachusetts Institute of Technology

The limited ability of articular cartilage for intrinsic repair has wompted numerous studies into biological approaches to promote repair of damaged articular surtaces (1,2,3). Currently, no method has Woven satisfactory and durable for the repair of large portions of joint surfaces. We have deveicped a new approach for cartilage repair by hypothesizing that the new cartilage growth must initially be protected from large stresses until it attains sufficient qualities to promote maturation into hyaline cartilage. The in-vivo model consisted of denuding patellar cartilage down to bleeding subchondral bone. On the experimental knee, joint spacers were inserted into the patella to ntechantcally separate the patella from the patellotemoral groove while still maintaining motion of the joint. The patella of the control knee was also denuded of cartilage without insertion of spacers. Preliminary results at six weeks showed a marked difference between the neocarfilage growth on control and experimental patellae. Gross and histological evaluation revealed minimal growth on the control skle, while cartilaginous growth on the experimental patellae covered the entire patellar surface up to the height of the spacers. Biomechanical analyses of this neocartilage on the experimental patellae demonstrated the tissue to be more permeable and less stiff than normal cartilage. Fudher investigation is underwayt6 determine the potential for a durable repair. This work was supported by the Veterans Affairs Rehab R&D, Grant #A799-RC. 1) CORR, 275:263, 1992. 2) Trans ORS, 19:483, 1994. 3) JBJS, 70A:595, 1988.

We have observed that cartilage exhibits electromechanical transduction properties that change in a very sensitive manner following less of proteoglycans and the resulting alteration of the tissue's extracellalar matrix. Previous experiments in a uniaxial confined compres~on geometry showed that cyclic compression of cartilage disks produced electrical streaming potentials, and application of cycfic electric current density across the tissue generated mechanical deformations and stress. These electrokinetic effects were found to be sensitive indicators of molecular-level changes in cartilage matrix. We recently developed a surface sensor based on electrokinetic transduction (current-generated stress) which can nondestractively detect molecular level changes that occur during cartilage degeneration in v/ere. Application of a very small electric current density s t the cartilage surface via an interdigitated surface electrode array produces a mechanical stress within the tissue that can be detected at the surface by a piezo-sensor array. The electrical and mechanical signals are periodic in both space and time, and the imposed temporal frequency and spatial wavelength cam be independently varied. Thus, an imposed standing wave of current density produces a corresponding standing wave of mechanical stress, measurable at the articular surface. We observed that the magnitude and phase of the resulting surface stress changed in a sensitive manner when the tissue matrix is modified by enzymatic or pliysicochemical techniques (e.g., titration of bath pH to the tissue's is~eisctric point, and trypsin digestion of proteogiycans). Surface loss of proteoglycans led to a more pronounced loss of short wavelength compared to long wave stimulation, consistent with the known confinement o f short wave currents to the surface region of the cartilage. A long term goal of this study is to develop a diagnostic system for nondestructi~e, in rive quantification of early degenerative changes in articular cartilage associated with diseases such as osteoarthritis. Supported by Pfizer, x.\~ Grant V525P-1743, and N S F Grant BCS-9111401.

Simple but Effective Orthopaedic Biomechanics Models

S-105

492

493

Spine Models

CHRONIC STRESS TOLERANCE LEVELS FOR HUMAN ARTICULAR CARTILAGE: TWO NONUNIFORM CONTACT MODELS OF LONG-TERM CDH FOLLOW-UP TA. Max=an,T.D. Brown, S.L. Walnstein Depts. of Odhopas~c Surgery and Biomed. Eng., Univ. of Iowa, Iowa City IA 52242

to Estimate the Risk of Vertebral Fracture SE Wilson and ER Myers, Orthopedic Binmechanics Laboratory, Beth Israel Hospital and Hm'vard Medical School, Boston, MA Age-related vertebral fractures afflict 500,000 Americans each year. In terms of years of morbidity and transition from independent to dependent living, vertebral fractures represent art important public health problem. Assessment of fracture risk is based on bone mineral properties of the lumbar spine, yet the compressive loads applied to the spine should also be considered in assessing risk so that hazardous activities can be identified. Based on preliminary data from a sequential survey of eases with vertebral fracture, we have found that approximately 15% of subjects recalled lifting and/or bending and over 50% recalled a fall. The objective of this study was to develop two models (one for controlled actions and one for falls) to obtain estimates of the compressive forces on the spine as a result of these actions. The model of controlled actions such as liftingand bending was developed from previous models of lumbar spine loads. Adaptations were made to include the rib cage to estimate loads in the thoracic region. The model of failing included two parts: a consideration of a backward fall to fred the impact configuration and a model of impact and resulting forces on the spine. We found that the compressive force on the spine for simple actions such as lifting 8 kg can exceed the ultimate failure load of the bone in an elderly individual.

Two computer models of nonuniform contact stress on the erlicolar surface of the human hip ware used to study the relationship between chronically excessive articular cartilage contact streSS and Iong-ten'n clinical outcome, in a series of patients with congenital dislocation of the hip (CDH). The analyzed dalabass consisted of 409 stylus digitized radiographs from 83 patients with unilataral CDH, who had been treated by closed reduction, and whose average fonow-up lime was 29.2 yearn. The first model (nonuniton'n Legal) involved a three-dimensional contact stress distribution function whose pole was coincident with the resultant force acting through the hip, and which acted over a contact area whose borders ware determined solely by bony landmarks. In the second model (Brinckmann), the direction of the pule of the contact stress distdbuiton function was initially unknown; one border of the contact region was determined by radiographic landmarks, while the other border depended upon the pole of the itemtively determined contact stress distdbuiton function. In beth models, the contact stress distdbulJons were converted to area engagement histograms, corresponding to the fractional areas of cartilage experiencing specific ranges of stress (0.5 MPa increments). These histograms were integrated over time to calculate a cumulative contact stress overdose, which was then compared to clinical outcome. Reasonable corralaitons (Spearman p = 0.63 to 0.66) with patient outcomes were obtained for optimally chosen damage thresholds, although these thresholds were appreciably different (2.0 versus 4.5 MPa) due to the respective modelling assumptions.

S-106

Imaging and Modeling for the Study of Joint Mechanics

494

497

MEASUREMENT OF ARTICULAR CARTILAGE THICKNESS IN THE INTACT KNEE CW Bmdrick, N Mukherjee, P Krishnan, JS Wayne Orthopaedic Research Laboratory, 325 McGuire Annex, 1112 E. Clay Street Virginia Commonwealth University, Richmond VA 23298-0694

ADVANCED COMPUTATIONAL METHODS FOR THE ANALYSIS OF B1PHAS1C TISSUES IN A JOINT RL Spilker, MS Shephard. IE Fisherty, MH Holmes. BK Szymanski. Rensselaer Polytechnic Institute, GA Ateshian, VC Mow, Columbia University

The thickness of articular cartilage is an important parameter in analyses of the tissue's function within a diarthrodial joint. Measurement of this thickness has been achieved by such techniques as radiographic, MRI, ultrasonic, and optical (1,2,3). It Is often desirable to measure cartilage thickness in the intact joint, without disarticulation or sectioning. We have developed a radiographic/imaging approach to measure cartilage thickness on the lemerel condyles of the knee. The len~Jr and t~oiaof intact porcine knees were secured in extension onto a materials testing machine. A smell capsular incision was made to visualize the most distal point ol the condyle. The joint was flexed, and a thin line of barium sultate/cyaneactylate mixture placed on the condyle. Thin metallic needles indicated the ends of the line. The knee was again placed in maximal extension and anterior-posterior radiographs taken of the undeformed thickness. After disartioulation, condyles were sectioned in the trental plane at the barium line to measure the cartilage thickness via an optical technique. Image analysis ot the radiographs demonstrated repeatability to be 35_+2@(2.5_+2=/=) Of the cartilage thickness at several locations on the condyles. Thicknesses obtained radiographically and optically compared well, with less than 71+23p. (6.5+2~ differences. This approach was found appropriate to measure cartilage thickness in the frontal plane and will be used in further studies of cartilage function iwvivo.

Solution of the fundamental mechanics problem of sliding tissuo-tiasuc contact in a diarthrodial joint requires sophisticated numerical methods including state of the art enmputational formulations, algorithms and hardware. We approach this as a continuum problem of contact between biphasic tissue layers, and are developing methods and computational R)ols which will provide accurate and efficient solutions to these problems over realistic joint geometry and at physiological load levels. This is an interdisciplinary projcnt, supported by the NSF High Performance Computing and Communication Grand Challenge Grant, which involves: the measurement of joint and tissue layer geometry; construction of solid models based on this geometry; application of attributes, such as material properties, boundary constraints and loads to the model based on measured data; development and evaluation of nonlinear tissue constitutive models; formulation of finite element solution strategies for the 3D nonline~ analysis of biphesic tissues in contact; automatic and mesh generation and adaptation; parallel methods and implementations for solution of the nonlinear first order time-dependent equations; and effective visualization of the 3D response predictions. In this presentation, we summarize the status of our efforts in various of these areas, As an example of our methods, we consider the glenohumeral joint of the shoulder and show: solid models of the glenoid and humeral head tissue layer, constructed from measured stereophotogrammetric data. oriented in anatomical positions; overlap areas identified by the solid modeler (neglecting in this calculation the deformation of the tissue); 3D meshes of each layer; 3D time-dependent biphasic finite element solutions of each tissue layer with load defined according to the tissue-tissue overlap pattern. This example w211demonstrate the numerical complexity inherent in these problems and the importance of effective parallel methods throughout the solution process.

This work was supported by the Whitaker Foundation. 1) JBJS, 61A:744, 1979, 2) J Biomech, 28:231, 1995.3) Trans ORS, 20:194, 1995.

498

495 3-D PATIENT-SPECIFIC FINITE ELEMENT KNEE MODELS: CONSTRUCTION AND VALIDATION A.W. Ebcdmrdt I,P2. Czawala 2, LM. Cacklat3 l ~ n t of Meehanlcal Engineering, 2Dcparenent of Biomedical Engineering 3Divlsinn of Orthopaedics University of Alabama at Birmingham, Birmingham, Alabama As many as 85% of Americans over age 65 show signs of o s ~ t l s (OA), and the knee is one of the most commonly affected joints. Abnormal stresses have been correlated with Da:e OA in 2-D models, however, these models were unable te distinguish cartilage and hone stresses of the true 3-D joint sm~cture. The primary aim of this project is to develop the means to accurately quantify and predict knee stresses in individual patients through the combination of gait analysis, imaging and finite element (FE) modeling. To date, we have developed 2-D PE models which ~ strain-rate depem:leot linear and nonlinear elastic cartilage properties in nonlinear contect analyses. We have also established e methodology for examining the tendency for cartilage surface cracks to propagate under normal and frictional jointloading. Pilotwork in our lab has established the methods for conatmcfing 3-D solid models from 2-D magnetic resonance (MR) images. To validate the FE models, we have begun to use quantitative thin fdm pressure sensors to measure contact fomes, and strain transduecxs to measure ligament forces in cadaver knees during dynamic cycles of flexion/extension. Initial steps have been made toward the inclusion of a biphasic cartilage layer using poreetastic elements provided in A B A Q U S . Kineticanalyses willbe performed on subjectsunder normal and abnormal (shoc-lifr)conditlons to establishthe FE models capacity to predictabnormal knee s~v.ssesduring gait. Ultimately,these methods will be applied to correlatekneo stresseste observed walking characteristics,reportedpain measures and clinicalevidence of OA.

496 STRESS M O D U L A T E D

GROWTH

OF C A R T I L A G E D U R I N G CIIONDROGENESIS J.H. Eeegaard, Stanford University,Stanford, C A

During early joint formation, both contacting cartilage surfaces are convex. However, as development proceeds one of the contacting surfaces progressively loses its convexity and eventually becomes concave, while the other surface remains convex. Understanding how such changes occur may provide new insights into the relativeimportance played by biological (e.g.,genetic) and mechanical (e.g.,epigenetic) factors in skeletal development and adaptation. In the present study, a computer model was used to simulate cartilagegrowth and to study the influence of mechanical stimuli on developing diarthrodisljoints. The model included two opposing cartilageanlagen, whose shapes at the interface were initiallyconvex. These segments were linked through a network of elastic strings representing ligaments. Alternate flexion/extensioD forces were applied, causing the joint to move. These forces also generated stresses in the cartilage and at the joint surfaces. Growth of the joint was ~ssurned to be locally shape preserving. Hence a spherical volume element would grow into a larger one and not into an ellipsoid. Changes occurring in the joint'sshape during growth resulted from spatial gradients of the growth rate. Growth of the cartilageanlagen was assumed to result from a preset biologicalsignal. This signal defined a scalar fielddepending linearly on the chondrocyte density. Furthermore this signal could be modulated by mechanical stimuli resultingfrom the flexion-extensionprocess. In the present model modulation was considered as a simple linear combination between the biologicalsignal and hydrostatic pressure. Using the~e hypotheses, the cartilage anlagen grew forming concave-convex joints in good agreement with patterns typically found in adult diarthrodiM joints. The relationship between joint motion and shape could be explored with the pre~nt model, and provided insights in the role played by biomeehanical factors in joint morphogenesis.

DEVELOPMENT OF MDLTIBODY MODEL FOR DIARTHRODIAL JOINTS USING ACCURATE 3-D CARTILAGE AND BONE SURFACES. SD Kwak, GA Ateshian. L Blankevoort. CS Abroad, TR Gardner, RP Grelsamer, VC Mow Departments of Mechanical Engineering and Orthopaedic Surgery Columbia University, New York A general multibody mathematical model has been developed for predicting the kinematics, contact areas and contact forces in diarthrodlal joints; this model can accommodate multiple joints described by any number of accurate 3-D geometric models of hones and articular surfaces (e.g. thumb, knee). Other features of the model include ligaments and tendons, both of which may wrap around these surfaces; tendons may also wrap around frictionless pulleys attached to any number of hones. External forces and moments may be applied to any of the bodies; internal forces resulting from stretching of ligaments, contact of articular surfaces, or wrapping of ligaments and tendons against surfaces or pulleys are automatically calculated. The unknown degrees-of-freedom are determined by solving equations of static equilibrium for all bodies, using a Nev.~on-Raphson iterative solver. Optionally, any translational or rotational degree-of-freedom may be constrained in the analysis. The mathematical model has been validated by solving problems of known analytical solutions. Furthermore, the model was employed to precisely reproduce the configuration of an actual knec joint experiment. The knee was tested at 30 ~ 45", 60% and 90" of flexion by loading the components of the quadriceps muscle (rectus femoris+vastus intermedius=267 N, vastus lateralis=178 N, and vastus medialis oblique=g9 N). The kinematics of the pate[la, the muscle and ligament insertions, and the bone surface geometries were digitized using a coordinate measuring machine (CMM. accuracy=50/Jm). The articular surface topographies were obtained by stercophotogramme:ry (accuracy=90#m). Model analyses converged for all flexion angles; results demonstrated that differences in patellar translations between experimental data and model predictions never exceeded 0.7 ram, showing excellent agreement between model and experiment.

Poster Presentations

S-107

499

502

AUTOMATIC MEASUREMENT SYSTEM FOR ORTHOPAEDIC APPLICATIONS TOTAL HIP ARTHROPLASTY AND PATELLO-FEMOKAL KINEMATICS CatherineG. Ambrose,Ph.D.. Jeffr=y D. Reuben,M.D., Ph.D., and Mieheal A. Androo,M.D. University of Texas Medical School Houston, Department of Orthopaedic Surgery

USE OF CARTILAGE ELECTROKINETIC PHENOMENA T O STUDY MACROMOLECULAR TRANSPORT THROUGH THE EXTRACELLULAR MATRIX Miner~a Garcia, Eliot Frank, Paul Grimshaw, Miehelle Jen, and Alan Grodzlnaky Continuum Electromechanics Group, Center for Biomedical Engineering, Department of Electrical Engineering and Computer Science, Massachusetts Institute of Technology

A computer program has been designed to automatethe measurementsusually performed by clinicians on either x-rays, CT images or MR images. The program consistsof a file convener to translatefiles from x-ray digitizers, CT imagers,or MR imagers into a standardized format, ~Lnda CAl)-typr packagefor measurementacquisition. Currently, the fiins from the following machine types can be converted: Rayven Xray Digitizer, Lumisys Lumiscse Digitizers, GE 9800 CT, GE Hilight Advantage CT, GE Signa MR, and Resonex MR- The program has basic CAD capabilities to aid in making measurements and has the capability of taking the user through the steps of making standard measurements for total hip anhroplasty and measurements for patella-femoral kinematics. In both cases the amount of user interaction has been reduced to make measurement more automatic and less time consuming. In th: case of total hip arthroplasty, measurement time was reduced by more than a factor of 5 when us~_g the software package. Results am printed in ASCII files capable of belng configured for input into many standard database and spreadsheet programs.

Physiologic compression of cartilage produces matrix compaction, fluid flow and streaming potentials. We have developed an approach to quantify the individual Contributions of these mechanisms to the transport of neutral aud charged proteins and low MW solutes in the extracellular matrix (ECM). To simulate physiologic levels of fluid flow through the ECM, an eleetroesmotic flow was induced by applying an electric field at a coestant current density across thin cartilage disks. The disks were damped at the periphery in a chamber such that the surface area for transport was free-swelling (no applied stress) and there was no pressure drop across the tissue. The combined mechanical and electrical boundary conditions result in an electroosmotic fluid flow without inducing matrix compaction. The magnitude of the induced fluid flow lies within the range estimated to exist in peak flows during physiologic loading. In the presence of the induced convection~ the flux of neutral solutes through the cartilage reached levels much gre~ter than the flux dae to diffusion alone. Our results confirmed the increasing importance of fluid flow in the transport of solutes of increasing size (and Peclet number). In studies of charged solutes, our results suggest that electrical migration which would occur in the presence of streaming potentials can also enhance the flux of charged macromolecules through the matrix. Thus in the absence of matrix consolidation, the fluid flows and electric fields of magnitudes relevant to physiologic cartilage loading can have a significant effect on the movement of macromolecutes through the ECM. By altering both the applied field and the charge properties of the macromolecules our measurements allowed us to assess parameters of matrix hindered transport including the intratissue diffusivity D,, a convective hindrance factor W,, and the intratissue electrical mobility/2,. Separating the contributions of the different transport mechanisms provides insight into the physiologic loading configuration in which macromolecular concentration gradients, unsteady fluid flow, electric fields and m;ttrix consolidation occur simultaneously.

50O

503

DESIGN ANALYSIS OF COMPRESSIVE TESTING METHOD FOR SMALL-SIZED SAMPLES S. O. Capps and C. F. Abrams, Jr. BMES Members

PLASTER OF PARIS AS A VEHICLE FOR TOBRAMYCIN E.F. Berkman, D. P. Mukheriee, J.A. Albright Dept. of Odhopaedio Surgery, LSU Medical Center, Shrevepod, LA 71130

Traditionally, compressive strength testing of materials such as steel are performed on small cylinders and ideally, compressive strength tests on biologlcal materials should be performed in the same manner. However, such an approach is not always possible due to geometric and material limitations (Capps, 1992). In biomaterlal studies, the slices of tibial bone from which cancellous bone cylinders could be cut may be small in cross-sectional area and rather thin. In two studies on poultry cancellous bone, the cross-sectional area of proximal tibiotarsal cancellous bone ranged from 1.4 cra 2 to 23 o n 2, and thickness of bone samples was from one to two millimeters. In preliminary experimentation many difficulties were encountered In efforts to create cylinders of cancellous bone from the slices ot tibial bone. Such difficulties included obtaining a cutting device to make cylinders of a small diameter (3 to 5 millimeters was generally used), removing cylinders from such a device without damaging the bone samples, and the creation of true right cylinders, Therefore an alternative method of testing the compressive strength of cancellous bone samples was sought. Through experimental and finite element analysis, a punch was developed for use in materials testing machines. Data collected from this punch were "found to be a reasonable estimate of compressive load to failure, deformation to failure, and elastic modulus when compared to right cylindrical samples of cancellous bone.

Plaster of Pads (POP), being an absorbable delivery matedal for antibiotics, may be used for treating esteomyelitis. We studied the elution of tobremycin from POP pellets and determined the Minimum Inhibitory Concentrations (MIC) for two strains of bacteria: Staphylococcus aureus and Group A. Beta Hemolytic Streptococcos. The dry molded pellets of POP contained tobmmyr concentration of 25mg/g immersed in tubes, each containing 20as of phosphate buffered saline (PBS), pH 7.4. in a 37 degree Centigrade bath. A 2cc aliquot was drawn on days 1, 2, 3, 7, 14, 21, 28, adding 2cc of fresh PBS each time to keep the volume constant. The samples were then assayed for tobramycin. Staphylococcus aureus and Group A Beta Hemolytic Streptococcus were used to determine MIC by a Scepter System. Moreover, the MIC was also determined by a standard serial dilution (2X) method with proper controls. The breakdown of the POP pellets was followed by visual absarvations. The dilution method gave an MIC for Steph as 5.5 mg/ml while the corresponding value for Strap was 11 mg/mI. For the automated tachn~lue, the MIC was I mgtml for Staph and t6 mg/rnl for Strep. The POP pellets did not show signs of degradation after 60 days. Ths ratio of tobremycin concentration in the elcate (after 1 day) to the MIC values was greater than 100x for Staph and greater than 40x for Strap. It was concluded that: the eluste concentration of tobramycin was markedly higher than the MIC values for that of Staph or Strap, tobramycin eTutionsfrom POP reached a maximum around 7 days.

501

504

AX.IAL A N D R A D I A L C O N F I N I N G S T R E S S E S I N C A L F E P I P H Y S E A L CARTILAGE DURING UNIAXIAL CONFINED COMPRESSION Partap S. Kh~lsa t , Kengo Baba, James L. Prandi, Solomon R. Eisenberg IDepaxtment of Physiology, Univetdty of Massachusetts Medical School Deportment of Biomedical Engineering, Boston University

THORACOLUMBAR PEDICLE SCREW FIXATION WITH ZERO, ONE, OR TWO CROSSLINKS G. Lynn, D.P. Mukherjse, R.N. KnJse, K.K. Sadasivan, J.A. Albdght Dept. of Orthopaedic Surgery, La. State University Medical School, Shreveport, LA

It is has been proposed that cartilage in uniaJdal confined compression behaves like a solution of proteoglyca~ (PC), and that the axial and radial coafintag atrc~ses shou]d therefore equal the PG swelling p r ~ u r e . To test this hypothesis, the axial and radial confining stresses were mesanred during uniaxlal confined compression in discs of r physeal cartilage (800pm, 6.35mm diameter) harvested from the distal ulna of newborn calves. Because the glycosamJnoglycan (GAG) content of calfepiphyseal cartilage increases monotonically with depth f:om the growth plate, specimens harvested st several different depths were examined. After equJJJbrating at a given [NaCI], a series of e=xialstratus was applied (4-20%), and the equilibrium a~dal stress measured. Two orthogonully positioned transducers ha the wall of the confining chamber measured the orthogonal radial confining stresses. Measurements were repeated at different [NaCI] (0.01-0.5 M) for each caxtilaSr disc. Axial and radial stresses varied linearly with axial strain, suggesting that the samptes were always well confined. The axial (HA) and orthogonal rad~.l ()q, ~2) moduli were determined [tom the slopes of the regression lines. In all cases, HA > ;h = ,~ =-- ,~. These results indicate that the equ~btinm stress-strain behavior in u~Lxlal confi~ed compression is not adequately modeled as a pressure, although an isotropic m e c h ~ c a l mode~ is approp~ate, polsson's ratio (v = ~/[Ha + ~]) was calculated groin the respective modulL For every sample tested, HA, ~ and v decreased with increasing [NaCI], consistent with previous observations. The values ~f HA, ~ and v at any given [NaCI1 were also found to increase linearly with distance fiom the growth plate, indicating that these material propetti~ are directly correlated w~th tissue GAG content.

Purpose: The relationship between dgidity and the number of crossiinks to fix thoracalumbar fractures is not dear. The objective of this study was to compare torsional and lateral bending rigidities of the thoracolumbar fracture fixed with AO "F'txatuer Interne" using either zero, one, or two cmsalinks. Methods: Eight embalmed three segment thorecolumbar spine segments were estestomized to simulate fracture. Each sample was F~ed by AO "Fhretuer Inteme" either using zero or one er two creaslinks. For each construct the rotational stiffness was measured st 2.5", 3.5". and 5" and the lateral bending stiffness was measured at .25, .4 and .5 inches of displacement. Results: The rotational stiffness of the two crossiinks construct was higher than that of the zero cressiink construct at 2.5~ (F= 3.85) and 3.5" (F = 3.98). it did not show any difference at 5.0~ because clamps holding the crosslinks slid up and down reducing the stiffness which may be addressed by adding serratioes. In lateral bending, the tv~ r constmut had significantly higher stiffness than that of the zero erosalink erosalink construct (F> 7.58). Only at a dispt=cement of 0.5 inches did the one crosalink construct have higher stiffness than that of the zero croeslink constt~t (P=SA8}. Conclusions: Rotational stiffness of the two cmsslink construct was higher than that nt the zero crosslink only at 2.5~ and 3.5" of rotation. The lateral banding stiffness of the two crosslink construct was significantly higher at all displacements.

S-108

Functional Electrical Stimulation

5O5 R E H A B I L I T A T I O N E N G I N E E R I N G FOR PHYSICALLY HANDICAPPED H. Thoma, M. Bijak, W. Girsoh, J. Holle, K. Howorka, N. LanmUller, G. Prulamp, E. Unger Dept. of Biomedical Engineering and Physics and Reconstructive Surgery, Univ. of Vienna

The application of "High Tech" instruments for physically handicapped persons depends on: (i) the knowledge of the market, (2) the education of the rehab team, (3) an individual adjustment of the device to special needs of the handicapped, (4) the maintainance and service of the devices, (5) the ressources for financing and (6) an integration in a rehabilitation strategy. Activities of our group concern educational approaches for patient rehabilitation and research, development and clinical application of new systems in Functional Electrostimulation (FES). 1982/83 two 8channel FES-implants (Medimplant Inc., Vienna) with external radiofrequency control and power transmission were implanted in 4 complete paraplegic handicapped persons for standing up and walking (first clinical trial). In the years 1983 till 1994 the same implant was used in 26 patients suffering from complete ventilatory insufficiency with different indications for fatiguefree stimulation of both phrenic nerves with best results.

5O8 PRIMARY AND POST-RECOVERY MUSCLE FATIGUE IN FES J. Mizrahi, O. Levin, A. Aviram, E. lsakov and 7. Susak Tecimion-IIT, Haifa, and Loewenstein Rehabilitation Center, Raansua, Israel The relation between muscle force, myoelectric activity and metabniic parameters dmlng FES were studied in primary and post-reeovery fatigues. Primary fatigue (PF) occurred when the muscle was supramaximally stimulated for a period of 3 min from an unfatlgued state. Postmcovetv fatigue (PRF) occm"red when the muscle was restimulated with the same stimulation parameters as in PM, after rest periods (RP) of varyiog time dumtians. A developed analyticci exprer of the fatigue curves enabled us to consistently define pcrfnrmsuec paramete~ such as maximal force (Fmax), maximal EMG, force impulse, fatigue index (FI) and recevery index (RI). Values of FI in PRF were found siguificaatly higher than those in IF. Values of the rise time of the M wave at t=30s from the onset of stimulations increased with increasing RP. The descending part of the force curve correlated well to the corr~pooding part of the peak-to-peak of the EMG M--wave, using the expooentiaJ, fit y=a exp(bx), hi PRF the coefficient b was found to decrease and coefficient a was found to increase with increasing RP. RI, expressing the ratio of Fmax at PRF to that of PF was found to increase with PP. A deveinped mtr,cuin-tendon mode[, incorporating fatigue and recovery functions, was used to estimate the metabetic profiles and predict the latigue/recovery dynamics of the muscle under FES. Acknowledgements: This study was supported by the Segal Foundation and by the Walter and Sandra Kaye Funds,

506

509

FES-AIDED GAIT USING THE CONTROLLED-BRAKE ORTHOSIS

STABILITY AND PLASTICITY OF FORCE-FIELD PRIMITIVES AND THEIR LOCATION MAPS IN THE SPINAL FROG. S.F. Giszte% W, Kargo, and M.R. Davies. Department of Anatomy and Neurobioiogy, MCPHU, Philadelphia, PA 19129.

W. Duffec l , M. Goldfarb2, N. Walsh 3 and A, Wieguer3 IUniversity of Minnesota, Minneapolis, MN 2Vaederbilt University, Nashville, TN 3VA Medical Center West Roxbury, MA FES-alded gait is hampered by poor control of limb trajectories resulting in steps with little or no repeatability, and by rapid muscle fatigue which limits walking distance, The controlledbrake orthosis (CBO) system was designed to address these limitations by utilizing FES in combination with a long-leg brace that contains controllable friction brakes at the knees and hips. The system achieves desirable limb trajectories by using the stimulated muscles as a source of coarsely regulated power and regulating die power at each joint by computer control of the friction brakes. The system reduces the effects of mnsele fatigue by locking the controllable brakes to provide the isumetric joint torques necessary during stance, A research version of the CBO has been designed, constructed and tested on a single, T-6 complete paraplegic subject and preliminary results demonstrated that the system can lead to more repeatable gait trajectories and significantly longer gait distances, In ongoing experiments, the CBO system is being tested in more subjects and plans are in place for redesign to achieve a version more acceptable as a product useful to the user. This work was supported by the Department of Veterans Affairs, Rehabilitation Research and Development Service.

507 WRIST FLEXION/EXTENSION PASSIVE PROPERTIES: IMPLICATIONS FOR WRIST CONTROL VIA FUNCTIONAL NEUROMUSCULAR STIMULATION Michel A. Lemay, and Patrick E. Crago Case Western Reserve University, Cleveland OH We measured wrist passive momenta in four able-bodied subjects, and in four tetraplegic subjects (two (24, two C5, and one C6 level). The wrist passive torque-angle properties were measured with a position servo motor. The servo motor cycled the wrist ainasoidally from flexion to extension at a frequency of 0.05 Hz while the subjects were asked to relax and not react to the imposed motion. Trials at 0.02 and 0.1 Hz were run on the able-bodied subjects to study the influence of velocity, To facilitate the implementation of the experimental results into a biomechanioal model of the wrist, the passive moment data were fitted to a model of the passive moment developed for the MP joint of the index finger [Esteki and Mansuur, 1995], The model decomposes the total passive moment M into an elastic component F and a viscous component which is the product of two functions: a wrist angle function G, and a rotational velocity function D, i.e. M ( 0 , 0 ) = F(O)+G(O)D(O). Explicit forms for F(0), G(O), and D(0) have been given by EsW.ki and Mansour [1995]. In most cases the passive fomes were significant, especially in light of the limited active Functional Neuromuscular Stimulation (FNS) momenta. Extending the wrist to 30~ could require up to 50 N,cm. The viscous component of the moment was also significant, up to 20 N,om in some cases. We did not see a rotational velocity effect on the hysteresis in the range of velocities tested (0,13-0.62 rad/s), and therefore the normalized thickness of the hysteresis loop was fitted to a power function of the form D(0) = 0", where n was much less than one, i.e. ,-0.1, When used in a biomechsuical model of wrist motion via FNS, this formulation significantly improved the stability of the wrist joint, as compared to a linear model of viscosity. Esteki and Maasour also showed that viscosity behaves like friction, appearing as soon as you move and not being very dependent on the modon velocity. Our experimental results conrttmed theirs and our simulation study shows the importance of the mechanism for joint stability.

Static force-field 'primitives' have been identified in the spinal cord of frogs by micre,stimniation. These force-fields represented stable limb postures and had a characteristic structure which was simply scaled through time as a result of stimulus current variations or stimulus duration variations. Only a few types of force-field patterns were found, and these were located in specific regions of the spinal cord grey matter. Maps of these primitives throughout the lumbar spinal cord suggests they are located quite repeatably in specific regions of spinal cord in the majority of frogs. Further, the primitives could often be combined by vector summation when elicited simultaneously. Finally, it h ~ recently shown convergent foree-0eld primitives also exist in the rat, These data suggest rehabilitation using direct intra-cord functional electrical stimulation (FES) with electrode arcays may be feasible. However, several open questions remain that are critical issues for such an FES approach. One unaddressed question is understanding what dynamic interactions occur across the midline between the two halves of the spinal cord and between force-fields elicited in the two limbs. To date, detailed mapping and force-field analysis has examined only a single limb in preparations after acute transection. Interactions among the pattern generators and primitives in the two halves of the spinal cord and the two hind-limb's reflex systems are likely. We are now examining how the primitives elicited in two hind-limbs interact. We are testing how force production and force-field summation are related when stimulation is applied in sites in each hemi-cord and examining whether some regions of spinal maps in one hemi-cord might be dynamically reorganized as a result of microstimulation in the contralateral hemi-eord. Supported by ASRI and NIH R29'NS34640.

510 PARAPLEGIA: HYBRID STANDING WITH THE IMPLANTED COCHLEAR FES-22 STIMULATOR AND ANDREWS FRO BRACES. Ross Davis, Thiervy Houdayer, tBrian Andrews, SJim Patrick, :l:Aedrew Mortlock Neural Engineering Clinic, 26 Eastern Ave, Augusta, ME, 04330, USA. tDept. Applied Sciences in Medicine, U. of Alberta, Edmonton, AB, Canada, ICochlear Ply. Ltd., Lane Cove, NSW, Australia. The study is to develop the Cochlear FES-22 implanted system (Cochlear Pry. Ltd,) and to use the Andrews Floor Reaction Orthosis (FRO) to restore standing in thoracic level paraplegies, The FES-22 stimulator was implanted in a 23 year-uld T-10 paraplegic male (CS) in November 1991, A computer controls and powers the stimulator through a radio-freqnency link, The FES-22 system delivers a balanced biphasic, constant current pulse adjustable up to 4.3 mA amplitude and 400 ~ pulse width, through 2.5 nun platinum epineural disk electrodes sutured over the branches of the femoral, gioteal and sciatic nerves, Conditioning of the knee and hip extensors is done by CS lying supine with hips in 30* of fiexion. Each knee extends alternately with a duty cycle of 4 see ON, 4 see OFF. CS muscles can contract for 20 minutes before onset of fatigue, During the 40 months of implantation, there has been no medical complication. 16 of the 22 electrodes am still functional including the knee extensors and half of the hip extensors. Subject CS is able to stand safely for up to 5 minutes at a time using the Cochlear FES-22 stimulator and the Andrews FRO. Standing is achieved by nerve stimulation of the knee and hip extensors muscles. When the knee extensors are manually turned on/off, the FRO (legankle brace) is able to maintain the knees in extension, so allow continued standing, The standing is done by open loop stimulation of the knee and hip extensors. Future plans are to increase the standing time by closing the loop with knee sensors and allow safe reaching at home and work.

Functional Electrical Stimulation 511

513

A NEW NON-LINEARHETEROGENOUSMODEL FOR ARTIFICIALLYACTIVATED HUMANNEUROMUSCULARSYSTEMS. S.J. Dorgan, M.J. O'Mallcy, Department of Electronic & Electrical Engineering. UniversityCollegeDublin, Ireland

DETECTION OF FES INDUCED FATIGUE AND RECOVERY OF WRIST FLEXORS IN TETRAPLEGICS Dejua TepavaeI, Dejua Popovi~t and Lasr.lo Sehw/rtliehta 1) The Miami Project to Cure Paralysis, University of Miami 2) Institute for Rehabilitation "Dr. Miruslav Zotovi6", Belgrade

Human Skeletal Muscle is a myriad of small, non-linuat, inter-dependant systcras s'witcningin and out of various contsactile stalusat different thnes. It is typically modelled using simple maerusoupic model ~o.n:,ctur~ capable of relatively easy param~r identificatio~ However, in using such simple macroscopic models, valuable insight into the underlying bshaviour of this complex biological system is lost_ Modelling the underlying physiologicalbehaviunr of muscle activated by Functional Electrical Stimulation, FES, is possible as beth the system inputs and ouq3uts are well defined. In doing so one can easily gather data for modelvalidation, incorporate neural feedback from propriocepturs, and gain significant insight into the biological system behaviour of FES muscle. A new model for artificiallyactivated human skeletal muscle is presented. The musoulotendonmodel st~tcture is highly non-linear, and heterogeneous, so acknowledgingmuscle's st~cture, and behaviom. as a highly non-iiuear system that behaves diffareutly in various cour~'actilo states. FES stimulation frequency and intensity ate taken as inputs to the model, while the underlying physiologicaldynamics, from in-,~rso recruRment and Ca2. ion release dynamics, through to non-linuar force and length dynamics, have all been reddened. The system equations are hJgidy coupled, a fact recently shoran by others to significantlyincrease model accuracy and reflects the physiological reality, A non-linesz spindle model has also been added to transform the new mus4:'niotendon model to a neuromuscular model, thus enabling investigationinto the role proprioecptor feed~ck plays in a neuromuscular system.

Abstract: The estimationof the externallyelicitedmuscle forces is essential fur the design of better functional electrical stimulation (FES) based assistive system.~. We am comparing results obtained when employingdifferent teclmiques of determination of the decresse of muscle force after continuous stimulation. All the techniques includeduse of surface recordingsof the evoked potential (sEMG), since the system that is in use at this time for restoration of motor funetious in most case still relies on transcutuncous FES systems. The wrist flexors muscles were stimulated under isometric conditions, and simultaneously the sEMG hxs been recorded in able-boded and s p ~ cord injured volunteerhumans, while thejoint torqua was measured. The joint torque was detennined from recordingsof the force generated by the wrist flexors, with the forearm immobilized. The sEMGwas recorded utilizinga preamplifierwith the stimulationartefact suppression circuitry. The signal obtained was processcA in time and frequency domain. The force-sEMG curves wen~used to establish the relationshipthat can be used for detection of the decrease of the force associated with the FES induced muscle fatigue. The best correlation was found between the median frequency and the force changes imposedUy continuous stimulation. We found that it is possible to use the processed sEMG as a trigger signal to change the pattern of stimulation, and allow (if possible) for the muscle to recover.

512 BIPHASIC PROGRAMMABLE CURRENT-SOURCES FOR IMPLANTABLE MINIATURIZED STIMULATORS Sawan M., St-Amted R., Savatia Y. l~..ole Polytechntque de Montreal lmplantable functional electrical stimulators are used in numereotts applications, as respiratory pacing, cochlear stimulator, visual implant, they assist In bladder continence and evacuation, etc. These systems enclosed numerous output stages containing cunent-sourees which derive electrodes connected to nerve tissues. Each current-source has to be highly flexible, energy efficient, compact and offering an appropriate output llnearlty. The commercially available devices ure bulky and lack the desired programmability. They often uses capacitor for D e decoupllog purpose In order to eounterbaianee the Injected charge during the stimulation period. These capacitors do not allow the desired ro.lntaturlzation. We propose an optimization procedure of a blphasic current-source which Is based on a miniaturized digital to analog converter (DAC). This programmable current-source (PCS) is intended to occupy the output stage of a wide range of neuromuscular tmplantable mihiaturized stimulators. The programmability of the global stimulator Is necessary to update the algorithm and the parameters of stimulation which permit a COntinuOus change with patients. However, the current-source should be able to generate a highly precise stimulus in term of amplitude, frequency and linearity. The proposed design method allowed to optimize area, linearity and power consumption and to minimize the output DC offset and then the charge accumulation that are noxious for the stimulated neuromuscular tissues. The described PCS has been designed using the 1.2 ttm CMOS technology, that allow to reduce value or to eliminate the coupling capacitors, while keeping the DC offset at a very low level. The global stimulation system as well as the PCS will be presented. Then, a De offset analysis and the optimization method to design the modular PCS, as well as the preliminary resultS will be provided.

S-109

S-110

Control of Movement

514

516

MECHANISMS CONTROLLING COMPLIANT BEHAVIOR OF THE ELBOW. H. B. Ahashanab, G. L. Gottlleb NeuroMuscular Research Center, Bestou University, Boston, MA

DIRECT EXPERIMENTAL TESTS OF "EQU'rLIBRIUM POINT" HYPOTHESES Justin WOU ([email protected]) and Neville Hogan Department of Mechanical Engineering, Massachusetts Institute of Technology

Different compliant loads differentially alter the trajectory, net muscle torque and EMG patterns of voluntary elbow flexiuns. These effects also depend upon. whether the load is correctly known to the subject in advance of the movement. With acearate l~ior knowledge, subjects reduce the influence of load upon trajectory by adjusting the central command, Lacking such knowledge, vtsco-elastlc mttscle properties and load dependent reflexes provide a degree of compensation. We propose that central planning of voluntary movement has three components. F'n~t (alpha) is a "program" for generating the excitation pattern to motconnson pools that can produce force,s expected to produce a desired trajectory using an internal model of dynamics comtmeted from prior experience. Second (lamda) is a moving, instantaneous eqaltibrium command, a *plan" of the desired trajectory. It provides a reference fc:~"reflex action during the ctwrant movement and coi'ruction and updating of the model used to generate ~ Third (gamma) is a reflex gain command that deteminns the "volitional set" by adjusting the degree and manner in which multiple reflex mechanisms may contribute to the muscle activation patterns. This triad control model incorporates aspects both of "equilibrium point" kinematic models and kinetic "tooter progra~" models but stresses different features as well as adding features those models lack. Among these is the insufficiency of elastic muscle models; the nonstereotypical, load dependent action of reflex mechanisms in shaping muscle activation patterns; and direct rather than feedback accommodation to changes in load compliance. Although a unitary plan for both the dynamic movement and the final posture might also account for the data, the parcellation of the plan into positional (postural) and force (movemen0 components provides a simpler and more complete model than current alternatives.

In recent years, reasonable doubt has been raised whether "equilibrium point hypotheses" for motor control could ever be disproved. These doubts result mainly from the fact that equilibrium points will always exist as a natural description of muscle activation. However the key issue to be tested is whether the CNS uses equilibrium points to generate movements and control coutant. In fact, several studies by Bennett etal. and Kawato etal. have recently shown that although an equilibrium point may exist, its overall iofineece over movement is relatively small compared to the dynamics of movement. Through a novel experimental study of the stability properties of the human neuromuscular system while it performs simple point-to-point arm movements, we evaluate the concepts of equilibrium and virtual trajectories as a means of executing movement. Human subjects grasped the instrumented handle of a planar two-link robot while performing specified point-to-point two-joint arm trajectories. Computer controlled clutebes warn used to suhtly change the movements by constraining the trajectory to lie apon a circular are perturbing the arm away from its natural straight movement. Three situations were tested: (1) unconstrained throughout the movement, (2) constrained through the entire movement, and (3) initially constrained and then released during movement. Experimental results showed that the constraint evoked significant forces strongly oriented so as to restore the hand to the unconstrained hand path. In addition when released from the constraint, these forces caused a strong tendency to return the hand to the unconstrained path before the end of the movement was reached, These results directly contradict the work of Bennett and Kawato showing that strong positional stability properties of die ann do exist reinforcing the notion that a moving equilibrium point dominates the dynamics of die ann during movement. Furthermore, the shape of the virtual trajectory was estimated indicating that the equilibrium point remains close to the actual movement produced, This result was further supported by a set of forward simulations using simple virtual trajectories to reproduce the experimental results, This results shows that a controlled equilibrium point may be used for planning and coordinating multi-joint movements,

515

517

THE ROLE OF INTERNEURONS IN THE CONTROL OF SPINALLY ORGANIZED MOTOR PRIMITIVES. Schodand, J., Dept. of Brain and Cognitive Science, M.I.T.. Cambridge, MA 02139.

MOTOR CONTROL EVALUATION OF NEUROSURGICAL INTERVENTIONS IN PARKINSON'S DISEASE (PD). IKD Pfann, 2RD Penn, I D M Corcos. IKincsiology, Univ. of IL at Chicago, 2Nenrosorgecy, Rush Medical Center.

Spinal interneureunl netwra'ks have been implicated in the coordination of redlex behaviors and equilibrium postures (Bizzi ctal. 1991). The following experiments are part of an investigation of the anatomical organization of spinal inteme~ons (INs) end their contribution to the set of reflex behaviors that corresFoads to mlerostimulatiou-induced equilibrium postures (Glszter etal. 1993). Anatomy: In one series of experiments, HRP was placed on the cut surface of the heimsected spinal cord at different locations. INs were found in highest nambe~ throughout the dorsal and intermediate gray matter of lumbar segments 8 and 9, In another set of experiments in progress, Fluorogold and WGA-HRP are injected into the ventral horn at vanons positions along the rostrocandal axis. INs projecting into both injection sites transport the label hack to their cell bodies and so are double*labeled. In this way, INs projecting to multiple motor pools that may play a role in the co~xdiention of muscle synergies can be localized. Elcctrophysiology: The frog spinal cord end hiadlimb (with electrode wires implanted in four muscles that discriminate among the reflexes) were placed in separate chambers. connected only by the spinal nerves. INs were recorded extracellularly, and identified by their response to stimulation of descending pathways, and by their lack of antidremic response to stimulation of the maid nerve trunk to the hindlimb. Reflex behaviors (wipe, flexion, and extension) were elicited by touching or pinching the skin of the hiodlimb. Iotemearous, recorded at depths ranging from 5-730 am, could be active d m ~ g muce then one different reflex behavior, they could be selectively activated thnSng pa~leular movement sequences, or they could be active during muscle contraction, even though no movement occurred. The period of activity could closely mateh the period of muscle contraction,could be considerably shorter, or could outlast the period of muscle activity. The activity of some INs could not be related to any obvious features of the experimental paradigm and SOme did not participate in the generation of any reflex behaviors.

We are evaluating the effects of two neuresmgleal intet~nntions in PD on the control of rapid single-joint elbow movements. First, we tested patients undergoing i~llidotomles, which were performed on L-dopa ~ i v e PD patients who have Ironble with fluctantinns. The clinical effects of pallidotumles can include reduced fluctuation, freezing, beadylrncsia, rigidity and/or tremor. Patients performed elbow movements over distances "as fast as possible'. We recorded angle, aeeeleratlon, torque, and surface EMGs ~ h i . i r a "~li~dle.biceps, and lateral and long heads of triceps. Oar I~liminnry results suggest that the movements and the associated EMO patterns are more consistent after the p a l l ~ y . Second, we tested one PD patient who had a stimulator chronically implanted in the VIM nucleus of the thalamus to control tremor. The subject performed elbow flexions over three distances "as fast as possible" and maximal isometric contractious under three conditions: 1) without stimulation, 2) with high stimulation (which minimized tremor), and 3) moderate stimulation. Tremor was dramatically reduced with stimulation. The patient could generate the mum force with moderate stimulation but could best generate fast movements with high stimulation. This suggests that modifying the slimuintinn level may be a useful tool in managing PD and that further motor control smdles would be beneficial.

Control

of Posture

S-111

518

521

PINNED POLYMER MODELOF POSTURECONTROL Carson C. Chow NeuroMuscular Research Center, BostonUniversity

THE DEVELOPMENT OF A COMPREHENSIVE MODEL FOR POSTURE AND

Recent analyses of quiet-standingposture data suggestthat stochasticallydriven dynamics may be present in the human posture control system. Motivated by this fact, a stochasticallydriven phenomenologicalmodel of posture control is presented. The model captures all the underlying dynamics observed in the measured two-point correlation functions of the time-varyingdisplacementsof the center of pressure (COP) under the feet of quietly standing subjects. The model is analogous to an elastically pinned flexible string or polymer under the influence of stochastic forcing. The simplicity of the model allows analyticalcalculationsof the correlation function using methods of nonequilibrium statistical mechanics. The model produces three scaling regions with scaling exponents that match those found in the data. The locationsof the break points between the scaling regions are given by parameters in the model. These parameters relate directly back to physiological origins. This allows meaningful quantitative information to be extracted from posturography. In addition the dynamical response to perturbations can also be calculated from the model. The model can then be used to test whether or not dynamic and static posture control utilize similarmechanisms.

519 INDEPENDENCE OF STATICVS. DYNAMIC POSTURAL STABILITY" M.D. Gmbiner Department of Biomedical Engineering, The Cleveland Clinic Foundation Age-related changes in the neuromuscular system, some irreversible and others somewhat reversible are determinants, to some extent, of the increased incidence of falls by elderly people, a major cause of morbidity and mortality. For decades, measurement of static postural stability has been widely used as an index of the predisposition to falling in the elderly. Our studies using healthy elderly people have partially challenged this contention by demonstrating that the age-related increase in postural sway is not obligatory. Further, in characterizing the biomechanical requisites for successful stepping responses following large postural perturbations (dynamic stability), we have shown that in healthy elderly people the amplitude of static postural sway normally represents only a small proportion of the total distance through which the COP may be voluntarily moved (stability limits) and unrelated to performance of mobility tasks. In addition, static postural sway amplitude, unlike stability limits, seems to be independent of parameters to which successful stepping responses appear to be subservient such as voluntary and automatic reaction latancies, and skeletal muscle strength and power. In contrast, stability limits, appear to demonstrate stronger relationsbips to these aforementioned parameters end a greater sensitivity to decreased neuromuscular function associated with aging.

M O V E M E N T CONTROL Jiping He Arizona State University, Tempe The control posture and movement system is a complex nonlinear system. It has a multi-layer structure with a diverse time scale subsystems: neural modulation of motoneuron membrane properties in microseconds, muscle recruitment and proprioeeptive feedback in milliseconds, and limb movement in seconds. Each of these subsystem has its own complex dynamics and intrinsic properties. Progress has been made in understanding these subsystems in separation, providing sufficient data for developing models of simulation study. Many detailed models have been developed to simulate the behavior of each subsystem. Some good modeling packages have become available to develop models of different systems: GENESIS for simulating neuron properties and membrane dynamics, SDFAST for developing models of complex mechanical system such as limb and whole body dynamics, SIMM for 3D animation of limb or body movements, as a few good examples. Yet a model combining different layers of the movement control system with adequate detail has not been reported. The progress in neural rehabilitation and functional electrical stimulation has raised the demand for a comprehensive model of posture and movement control system. This system should include motoneuron properties, muscle dynamics, proprioeeptive circuitry and limb mechanics. A progress report in an effort to develop such a model will be presented.

522 TIME-VARYING CHARACTERISTICSOF POSTURALSWAYIN THE ELDERLY Patrick Loughlint and Mark Redfern2'3 University of Pittsburgh Depts. of I Electrical Engineering, 2 Industrial Engineering and 3Otolaryngology Pittsburgh, PA Abstract-- The ability to maintainbalance generallydegrades with age, a factor that contrib-

utes toinjarles due to falls in die elderly. Balance is achievedthrough feedback conUvlutilizing visual, vestibular and somatosensory inputs. How these sensory systems degrade with aging, and the consequent influenceson balance, is a dynamic process that is not completely understood. Recendy, it has been recognized that posturai sway, which is an indirectmeasure of balance control, is a nonstationaryprocess. As such, conventionalmethods of LTI process* ing. e.g. spectral analysis,provide an incompletedescriptionof the system. In this study, we apply time-frequency analysisto investigatethe time-dependent nature of postural sway in the elderly. Comparisonsto control populationsare provided.

520

522a

MODELING RESPONSES DURING COMPLEX THREE DIMENSIONAL VESTIBULAR STIMULATION WITH APPLICATIONS TO MOTOR CONTROL Daniel M. Mcffeld R.S. Dow Neurological Sciences Institute, Legacy Good Samaritan Hospital, Portland, OR 97210

MULTITAPERSPF.,CrRALESTIMATESOF THE CENTER-OF-PRESSUREVELOCITY Thomas E. Prietu, Eric G. Lovett,Joel B. Mykiebust,and Barbara M. Mykiebust Dept. of Neurology, Medical Collegeof Wiscottsin;Laboratory of Seasory-Motor Performance, VA MedicalCenter;, Dept. of BiomedicalEngineering,Marquette University,Milwaukee,WI

A mathematical model, based upon conceptsfrom observer theory,has been developed to help understand the complex sensory interactionsinduced by three-dimensionalstimulationof the otolithorgans and semi-circularcanals. Thc primary hypothesis is that the central nervous system incorporates information about body dynamins, sensory dynamics, and the external world into an internal model of these systems. This internal model allows the nervous system to combine the sensory information from many different sensory modalitles into a coherent and comprehensive estimate of self-motion. More specifically, the internal model is used to predict "expected sensory

signals"which, when compared to the actualsensory signals,yielda differenc~defined as "sensoryconflict."This errorisused to drive the central estimatesof angular velocity,gravity,and linearaccelerationtoward more accuratevalues. The model successfullypredictsthe eye movement responses of humans and squirrelmonkeys during severalcomplex motion paradigms. Extensionsof thismodel to includeaspectsof motor controlwillbe discussed along with other issuesrelatedto sensory-motorintegration.

Posturai sway during tests of standinghalanee is often characterized with measures based on the displacementof the nenter-of-pmssm'c(COP). Previousstudieshave found tirolthe mean COP velocityduringquiet standingchanges with age and neorologicdisease. The frequency content of COP velocityincludes informationassociated with activityof die posturalcontrol system m maintainstandingbalance, and possibly other physiologicalprocesses, such as uemors and heart rote. The COP is definedby an~rior-postevior(AP) and medial-lateral(ML) ccordinates; die resultant velocitymagnitudeis independent of foot orientationrelativeto die force platformaxes. The power spectral density(PSI)) of COP velocityis estimated using the sinasoidal muliitaper(SM) method over a data st~cific frequency range. SM PSD estimatesof the AP and ML COP displacementtimeseries are used to estimatebandwidihs(f~ and t'/~) for these time series. Bandwidthis definedas the frequency above which the PSD is less than an arbitrary factor abovedie noise floor, which is definedas die averagepower in die upl~r half of the frequency range of the PSD estimate. A linearphase discrete time differeatiatoris used to estimate the AP and ML COP velocity time series. The PSD of COP velocityis computed by summingthe AP and ML velocitySM PSD estimates over the frequency range from 0 to fBw, where f~w is the greater of lAPand t~Ml. The PSD of COP velocitymay be useful to bend ondesstundthe processes affecting postural sway, and die changes in these processes related to age and neurologicdisease. Supported by research funds from VA RehabilitationR&D and The Whither Foundation.

S- 112

Assessment of Muscle Function

523

526

FROM MUSCLE PROPERTIES TO HUMAN PERFORMANCE USING MAGNETIC RESONANCE. Kevin E. Conley Department of Radiology and Center for Bioengineering, University of Washington Medical Center, Seattle, WA 98195

EFFECTSOF MUSCLEFIBERMORPHOLOGYON THE EMG SIGNAL E.L Kupa, S.ILRoy, S.C. Kandarian*, and (2.1. De Luot NeoroM~ ~ h Center, BostonUniversity;,and *Sargent Collegeof Allied Health Professions, ~ n t of Health Sr Boston. MA 02215 This abstranl describes an in vitro method for comparing surfac~tncted EIdG median

Our goal is to show how muscle properties can be used to understand the exercise performance limitations in humans. We show that magnetic resonance (MR) imaging and spectroscopy are useful for non-invasively characterizing the structural and energetic properties of muscle in rive. Determination of muscle volume and cross-sectional area is easily and rapidly accomplished by applying quantitative morphometric methods to MR images. New MR spectroscopic techniques provide a non-invasive "biopsy" of the oxidative, glycolytic and contractile capacities of muscle fibers. We show how the structural and energetic properties measured by MR can be used In define the functional capacity of muscle and the contribution o f this capacity to the performance of the whole body ( e . g . , ~/O2max). Finally, we relate these laboratory measures of muscle properties and performance to activities meaningful to the functioning in everyday life, such as sustained walking and stair climbing.

frequency (MF) and conduction velocity (CV) pmameters with measm~menisof muscle fiber composition and cams-sectional atea (CSA). EMG signals wetc recorded during electrically elicited tclanic contractions from rat solc~, cxtcnanr digitomm longus and diaphragm muscles plased in an oxygenated Kn:be beth. EMG signnis from control animals and a ~ l s which were h/edl/mb unwelghtedfor periods of 7 and 21 days have been collncted. Muacle fibers were typed as slow oxidative (SO), fast oxidativeglycolytic (FOG), and fast glyeniytie(FG) based on hlstochcmicnienzyme stains. Control mmcles with a greater pen:eniageof FG and FOG fibera exhibited greater in/tlal values of MF and CV as well as a greater reduction in these variables Over the course of a contraclio~ Reg~+siouanalysisof control data indicatedthat fiber-type compositioncouldbe pnedicted based on two MF parame~rs. A weighted measure of muscle fiber CSA wss found to be linearly related to beth initial MF and CV for control animals. Preliminary results for hindiimb unwcightedanimals indicate that initial MF of the EMG signal is decreased in the soletts. There is also a decrease in the peak to peak amplitude of the signals. These changes ill the EMG signal are meg likely related to the reduetlon in mmalv fiber CSA that results from hindlimb unweighting.Accompanyingthe change in fiber CSA area is a shift in fiber-type percentsgos toward fast fibers. However it appears that detection of this change is precluded by the effect of differences in muscle fiber size. These results suggest that MF and CV garamelcrs recorded during a muscular nonunion ate related to muscle fiber-type compositionand muscle fiber CSA.

524

527

Organization and Principles of Muscle Bloenergeflcs for Energy Balance Martin J. Kushmcrick Departments of Radiology and Physiology and Center for Bioengineering, University o f Washington Medical Center, Seattle, W A 98125

SURFACEEMG SPECTRALIMBALANCI~ ASSOCIATEDWITH LOW BACK PAIN Oddsson, L., Roy, S.H. & De Luca C.J. NeuroMuscular Research Center, Boston University, Boston,/viA, USA Over the past decade, a technique to assess muscle h~rtction during l o w back pain has been developed at the NeuroMuscular Research Cenber, Boston University. The technique is based on the fact that a sustained muscular contTaction causes a compression of the power density spectrum of the electromyographic (EMG) signal towards lower frequencies. This neuromuscular sign of fatigue can be monitored as a decrease in the median frequency (MF) of the EMG spectrum. The subject to be tested performs a constant isometric trunk extension in a device called the Back Analysis System 03AS).In our current test protocol the subject maintains hhis contraction for 30 s at 80% of a maximal voluntary contraction while surface EMG signals are assessed bilaterally from three paraspinal lumbar muscle sites 0-1, L2 and I.,5). Previous studies have shown that parameters extracted from the MF signals can be ~ to discriminate between LBP patients and healthy control suh~cts. In this study a posbhoc analysis was performed on a selected group of chronic back pain patients with pain at a position corre~pending to one of the EMG niecerode sites. A set of parameters was derived to describe the right/left segmental balance of EMG amplitude (RMS) and MF of the EMG signals. Patients showed marked deviations in these parameters at the site of pain when compared to a group of control snbiocts. In addition, a cumulative score of right/left RMS and MF imbalances indicated that control sub~ects, tmlike the patients, appeared to compensate for segmental imbalancm. Thus, ff a control subject displayed a segmental imbalance at L1 in one direction there was likely to be an imbniartce in the other direction at L2 and/or 1.5 which appeared to counteract the L1 imbalance. These results Indicate that an important aspect of healthy back function is a non-rigid activation behavior, not present in LBF patients, which appears to compensate for minor segmental imbalances present even under normal conditions.

With all the intricacy of metabolic pathways, there is only a small set o f common metabolites and biochemical reactions involved in energy transducing mechanisms with ATP and P e r (phosphorylcreatine). Them is also a short list of principles which governs energy transformations in achieving energy balance. These are: 1. ATP provides the energy for all forms of muscle work. 2. Chemical energy is stored in cells as a bioebemieal capacitor, PCr. 3. The sum of the coupled ATPases sets the demand side o f the balance and defines energetic states. 4. The products of ATPases provide control signals for energy balance. Feedback regulation by products o f ATPase is mechartistieally necessary, and appears sufficient in skeletal muscle to account for energy balance. Muscles, which may differ quantitatively in the magnitude of their ATPase and synthesis capacities, achieve energy balance but with different metabolite concentrations and biochemical steady states. Recent work in human muscle demonstrates that both components of energy balance, ATP utilization and ATP synthesis, can be separately and quantitatively measured. We find that individual differences are identifiable and meaningful mechanistically. We translate these principles into a simple model of muscle bioenergetics. Simulations reproduce the features observed in human llmb muscle activity.

525 SIMULATIONOF EVOKED ELECTROMYOGRAPHIC(EMG) SIGNALS E. Avignone, P. Gugllsiminotti, E. Kupa*, L. Lo Conic, R. Merletti*,S. H. Roy* Dip. di Ftsiea Spcxlm, Univ. di Torino, Politecnicodi Torian, Torino 10129, Italy *NeumMaacular Rescatch Center. BostonUniversity,Boston. MA. USA A mathematicalmodel has been developed that represents the stu'face EMG signal as a summation of contributions from the single fibers of individual motor units (MU). Each MU has parallel fibers scattered within a cylindrical volume of specified radius in an anisotropicmedium.Two depolarizedzones move from the ocuremescalarj anntion (NMJ) to the tendon endings. Each zone is modeled as a current t-/pole generatedat the N M I and extinguished at the tendon endings. NMIs and termination zones are scattered within regions of specified width. Average fiber length and conduction velocity (CV) are also specified. The signal produced by MUs with different geometries and CVs are superimposed. Moonpolarand bipolar signalsare computed in equallyspaced locationson the surface of the muscle. Neuromuscularpreparations of the seines, extensor digitomm longes and diaphragm musclesof rats were placed in a bath over an array of 12 silver pins 2.3 mm apart; 0.5 mm diameter. The array spanned the entire muscle length. Eloctriual stimulationof the nerve elicitedM-waves detected by the 12 pins (monopolar)or by the l 1 pairs (differential).The model was used for the simulationand inte*pretationof M-waves along the muscle length and to study EMG parameters. Interpretation of evoked EMO signalsdetected from the human biceps brachll (12 contact array with 5 mm spacing) was also attempted with the model. Results confirm that the model provides a close approximation to actual signalsand that there is a very narrow set of parameters that fit the experimentsl sequence of M-waves.It is concluded that the combinationof this model with the signal detection and processing system is expected to providenon-invasiveestimation of geometricand functionalpropertiesof superficialmuscles.

Assistive Devices 528

531

DISTINCTIONS BETWEEN PRODUCT DESIGN AND CUSTOM DESIGN IN ASSISTIVE "rE-O-N:E73Y Michael J. Rosen, Ph.D. Rehabilitation Engineering Program, University of Tennessee, Memphis

DEVELOPING ASSISTIVETEO'-INOLOGYFOR MICRO-MARKErS T. Walley Williams, III Liberty Technology, Hopkinton, Massachusetts

The student engineer attracted to designing assistive devices can pursue this interest through two distinct modes of practice: product design for commercial production, and custom design for individual users through a rehabilitation engineering service delivery program. The practice of design is sufficiently different in these two settings that somewhat different training is required. Some salient features of custom design of assistive technology which distinguish it from product design are as follows: 9 The fundamental dogma of QFD that all design and manufacturing decisions should be traceable to the "voice of the customer' is trivially easy to implement in custom design in that s/he can be a direct partner in the design process. Developing e device for an individual can be very iterative and interactive, involving several test-and-refine cycles on the way to delivery. 9 Assistive technology service delivery has - - by regulation, clinical politics and common sense - - come to be the responsibility of a multi-disciplinary team which includes the engineer. In other words, concurrent engineering happens as a matter of course. A corollary of this and the engineer's direct involvement with the user is that the engineer is involved in point-of-use ergonomics and integration of the new device with functions beyond its intended use. 9 Custom design as practiced in typical assistive technology service delivery programs entrains a direct follow-up obligation for the designer. This essentially imposes an indefinitely growing case load on him/her but also provides a rich and intimate channel for feedback and design improvement.

Too often developers of assistive technology launch a new product because of a strong awareness of the need it will meet and its advantages over existing devices.i While the designers' understanding of its potential users and its functionality may be accurate, the new product can easily fail in the marketplace. One reason for the economic lailure of a new assistive device can be the absence of 9 marketing strategy, especially when the size of the market is extremely small (compared with mass-market products or even larger-market assistive devil;as such as wheelchairs). The Boston Arm offers a case in point. Twenty five years ago, Liberty Mutual first developed this self-contained powered myoelectricallycontrolled prosthesis, and it was indeed an economic failure until it was made part of e marketing strategy. The plan involved offering ~ in addition to the Boston Arm - - several related products which would all share the same support personnel and all oenefit from the same sales and marketing effort. The original add-on products were powered hand prostheses for children, imported from Canada. They are sold to the same limb shops as the elbow and use the same kinds of technology so that no new fitting and service expertise is needed. More recently, Liberty has developed a full line of myoelectric controls, pressuresensitive controllers, positional serves and a line of batteries and chargers all of which provide mix-and-match compatibility. While each of these new products sells in small quantities, Liberty is more successful because the original market channels and support personnel can be utilized to attract a somewhat larger and more diverse market without increase in cost.

529

532

COMMUNICATING ABOUT COMMUNICATION: THE ROLE OF HUMAN FACTORS

CRITICAL APPRAISAL OF UNIVERSAL DESIGN Michael J. Rosen, Ph.D. Department of Biomedical Engineering, University of Tennessee, Memphis

Cheryl Trepagnier University of Tennessee, Memphis One of the challenges in marketing augmentative COmmunicationdevices is to communicate effectively with consumers and support persOns without raising unrealistic expectations. At the same time manufacturers need to make competitive claims about effectiveness. The problem is complicated by several factors: (1) Discrepancy between popular conceptions of language and current linguistic knowledge: Most people have the sense that they understand quite well what language is and that children learn it by imitation and correction. Linguistic studies have pointed to innate structures and constraints which act to limit possible grammatical hypotheses. It is difficult otherwise to account for how children acquire language without direct teaching. Imitation is inadequate since most possible sentences of the language are not heard. Negative feedback is not adequate since most language errors are ~ot corrected, When correction is given children have little way of telling what is being corrected, unless they have already formed grammatical hypotheses. (2) Attribution of complexity and p~wer to augmentative communication devices: Augmentative communication devices are expected to be complex and difficult to understand; and complexity is expected to be related to power. (3) Lack of evidence relating device features (o communicative effectiveness for consumers with different characteristics. Most studies comparing features of devices have invo[ved computer modeling, use of non-disabled participants, short-term in-laboratory tests, or case studies with no controls. As a result, real knowledge about what approaches can be expected to he most beneficial to individuals is lacking.

530 MASS-MARKETASSISTIVETECHNOLOGY:C'E'VlCESFOR ELDERLY CONSUMERS Geoff Ferule, Ph.D., P. Eng. Centre for Studies in Aging, Sunnybreok Health Science Centre, Toronto Assistive devices for people with disabilities must often be marketed successfully while selling only hundreds or even fewer units annually. A dramatic exception is technology aimed at meeting the daily living needs of the millions of eldedy people in North America, i.e. "special" products such as handrails, toilets and bathtubs. The size of this population segment is, in fact, growing at a higher rate percentagewise than any other (and its average age is increasing as well). Besides sharing limitations in physical and cognitive ability with the populations typically identified as disabled, the two markets are also both characterized by dramatically reduced financial resources. Another feature elderly North Americans share with people with disabilities is that the younger able-bodied population views both of them as distinct and different, almost always in negative ways. It is just as important, then, when designing for eldedy consumers, to develop assistive products which are functionally and aesthetically consistent with the mass market built environment. They will be perceived by their users as less stigmatizing and providing greater value. Moreover, if "elderly market" products successfully cross over and appeal to younger users, additional economy of scale is gained. Meeting everyday needs under these "soft" design constraints turns out to be hard, requiring as much or more engineering effort and application of science as solutions to uncommon problems under looser restrictions on cost and user appeal. Rehabilitation engineers engaged in designing products for eldedy users are in the unusual position of meeting the special needs of a large market, and doing so with products which don't look special or carry a special price tag.

S- 113

As an ideal, univemal design eliminates the need for assistive technology altogether; ff the built environment can be used by anyone on the human distribution of abilities, then no one is placed at a functional disadvantage. Assistive technology is obviated since everyone is effectively "cured" of his or her handicap. This concept is an asymptote for one dimension of design quality - or a desirable mindset for the product designer ~ rather than an attainable goal. Schematically, universal design is a point in a two-dimensional space of solutions to functional needs. On one axis are "orthotic" devices which augment the person with a disability with wearable technology. On the other is activitycoupled special technology. In general, matching a person with a disabUity to a task Involves both, e.g. a wheelchair and a curb cut. On this plane, universal design is at the origin, i.e. the class of solutions which requires neither. If treated as one practical product design goal among many, universality will generally need to be traded off against other goals. In particular, mediocre functional utility may result from one-size-fits-all design. This may be particularly true if a product Is designed to accommodate people who are categorically different from each other (e.g. blind and sighted computer users) rather than quantitative/), (users with typical and reduced finger mobility). Certainly, universal design can be defined on an even more tactical scale as the exploitation of the economy of scale of mass-market components and systems by making maximal use of them in small-market technology. Digital electronics, composite materials and speech input systems are examples which make it clear that this kind of universality is now the norm in assistive technology.

S- 114

Ergonomics/Sports

533

535

S H O U L D E R M U S C L E ACTIVITY AS A R I S K F A C T O R F O R T H E D E V E L O P M E N T OF S H O U L D E R PAIN Westgaard, R H University o f Trondheim, Trondheim, Norway

THE VISCOELASTICITY OF SPINE AND ITS EFFECTS ON LOAD SHARING AMONGST THE ACTIVE AND PASSIVE ELEMENTS Wang JL*, Shirazi-Adl A+, Parnianponr M**, Engin AE* *Department of Engineering Mechanics, **Industrial Welding and System Engineering, Ohio State University, Columbus Ohio 43210, +Mechanical Engineering Ecole Polytechniqne, Montreal Canada

Muscle pain localized to the shoulder and neck region is an increasingly common syndrom in many occupations with light, repetitive work. The problem is not easily handled through regular biomechanieal guidelines, meant to protect workers who perform work tasks at high exertion levels from injury. In fact, vocational studies with E M G recordings from the trapeaius muscles, performed by our group, have shown that there is no relationship between load level and shoulder pain symptoms in light repetitive work and also in regular office work. In contrast, the pattern o f muscle activity and also the activity level at rest appear to distinguish those that have complaints from those that are symptom-free, at least for some occupational groups. The pattern o f muscle activity as a risk factor for muscle pain can be quantified in terms o f E M G gaps, ie. short (~0.2 s) pauses in the E M G interference pattern. Other prominent risk factors include psyehosocial problems, which are often related to low-level E M G activity in laboratory studies. The possibility o f developing supplementary guidelines based on E M G recordings, or introducing measures that will counteract the development o f muscle pain in occupations with low physical work load will be discussed in light o f these findings.

Biomechanical studies of the spine are cencerned with the quantitative analysis of internal and ~ loading of the spine during physical exertions. It has been documented that due to the dynantics of muscle acsion, the internal forees arn affected by the tlme dependent charsctensties of external loads. The changes in the temporal and spatial recruitment pattea'm lead to significantly different net reaction forces in the spine. Due to kinetic redandanr pre~nt in terms of multiple muscles spanning the joint, the mapping of external moments to rausr fort~ is indeterminate. Models either do not r the load sharing role of passive elements or only have assumed the response of an elaslir spine. In this study the viscoelasticity has been considered by the development of finite element model ofL2-L3 cousidering: geometric and material noulineasity, facet contact, and time dependent material properties of annius and nucleus pniposus. Parameter estimation was achieved using experimental data of eight lumbar spenimeus and modeling in addition to relevant data present in literature. A series of loading conditions has been simulated to illustrate the change on load sharing of the passive elements as a function of time: creep, stress relaxation, cyclic loading at different rates, simulation of lifting and lowering at different rates. The detailed mess-strain analysus indicated that higher stresses were developed in the annius fibers at lower strain levels (smaller joint angles) and at higher loading rates. The results indicate that evaluation of risk due to physical activity must fully account for the dynamics of loading conditions.

534

536

MECHANICAL STABII.XIT OF THE IN VIVO LUMBAR SPINI] CholewicH', J. and McGill, $.M. Dept. of Kinesiology, University of Waterloo and "Dept. of Oahot~uxlics, Yale Unlverdty.

RF3~TIOI ~SHIP BETWEEN MECHANICAL AND ELECTROMYOGRAPHIC MANIFESTATIONS OF FATIGUE Manninn AF, Dolan P. Dept Anatomy, University of Bristol, Sonthwen St, Bristol BS2 8El, UK.

One important medagnie9 function of ~ lumbar spine is to support the uppea" body by tntusmitting compressive and ~ fore~ to the lower body dusing the perfotman~ of eve,-~/day ~etivilies. To enable the tau:ee~ful hnmsmisdoa of these fosees, mechanical stability of the spinal system must be Msmed. The propose of this stndy wsa to davelop a mathod and to quantify the ~-h~.icsl ~ability of the lumbar spine in rive during various 3 ~ o a n l dynamic tasks. An EMG Assisted O ~ r (EMGAO) model w ~ used to estimate the lumbar spine stability index three times per second throughout the duration of each trial. Anatomically, this 18 degrees of freedom model included 9 rigid pelvis, ribeage, 5 veaobrae, 90 mu~lo f~eiel~ md lumped ptrametef dlsca, l i g ~ t a and f'~O~. A erou-bridgu bond Di~ailmti(m-Mom~t muscle model was used for the inldal eatimatce of muscle force mad miffmessfrom the EMG. The relative stability index was cslcolatad using 9 principle of minlnmm poteatlal eaetgy for 3 tpabjects. It sppeata that there is an ample stability safety margin during that demand 9 high m ~ , l & effort. However, fighter tasks pre~eet 9 potential hazard of 9plne't~eklmg, upeci~iy if deficiency in passive joint ~ffile~ is trident. We c~a now expl~a the cc~a-venco of spine injury during activities requiring low muscular fore~. P e r ~ the momentary loss of splno aability that would lead to unexpected dispiscemeata, irritate nociceptors in connective and/or moR ~ or nerve roots. Conversely, the sudden need to regain spine stability m y result in an oveslc~d of 9 single muscle. If the motor control syslem faces the danger of lumbar spine buckling, 9 plausible response would be to aetivale s small intrinsic mm~le(s) that c ~ a Im~ieuiarly unstable joint to counteract the large displacemcota. Larger muscle4 qmmlng t~vertl joints are not suitable for such action, as this would the load borne by the infea'inr intefvertebnd joints, magnifying the effects of buckling.

In the assessment of human physical performance, muscle fatigue is commonly defined as *the failure to maintain the requireA of expected force output". As such, it is sometimes argued that the use of EMG power spectral statistics to 'monitor' muscle fatigue is Inappmpslate, because, during the maintenance of a submaximal force of contraction, EMG changes are readily observable in the absence of any decline in the muscle's mechanical outtmt. However, it is possible that these changes in the EMG power spoctnan reflect the changing metabolic status of the muscle during contraction and hence its maxima/force generating capacity. If this were so, a sfight rel'mement of the definition of fatigue wOuld establish electromyogmpby as a valid tool in the es.~ssmcot of fatigue in a way which would complement, rather than be at variance with, more conventional methods of measmmnenL The maximum isometric voluntary contraction (MVC) of the knen extensors was determined in 11 yotmg, healthy subjects. On 5 separate occasions, randomly assigned forces equivalent to 20-60% MVC were held to tile limit of endurance. At intervals throughout the maintained contractions, subjects were r e q ~ to rapidly generate an MVC for 1-2 s, then retaro to the fixed submaximal target force. Throughout each endurance test surface EMG signals were recorded from the rectus femoris (RF) and vastes lateralis (VL) muscles, from which the power spsctrnm median frequency (MF) was calculated. Regression analysis revealed highly significant relationships between the rate of decline in MF (percent psr second) and the rate of decline in MVC (percant per sex:end) dining the tests. It is concluded that the decline in MF can be used to monitor fatigue, where fatigue is defined as the inability to generate the maximum force ll~'. can be produced by the muscle in its fresh state.

Gait Analysis 537

540

WHAT REHABILITATION ENGINEERS SHOULD KNOW ABOUT THE E N E R G Y C O S T OF G A I T C o r c o r a n , P a u l J, M D Harvard Medical School, Spaulding Rehabilitation H o s p i t a l , D e p a r t m e n t of P h y s l c a l M e d i c i n e and R a h a b

BIOMECHANICAL ADAPTATION VERSUS NEUROLOGICAL DEFICIT IN DISORDERED GAIT K G. Hoit Boston University, Department of Physical Therapy, Motion Analysis Laboratoqf

T h e a n a t o m y and k i n e s l o l o g y of the l o w e r e x t r e m i t i e s d e t e r m i n e s the e f f i c i e n c y of h u m a n l o c o m o t i o n ; energy c o s t of n o r m a l g a i t is m i n i m i z e d as the c e n t e r of g r a v i t y (COG) f o l l o w s a p e r f e c t sine curve. Its v e r t i c a l t r a v e l of two inches w i t h e a c h s t e p a c c o u n t s for 9 0 % of the e n e r g y cost of a m b u l a t i o n . D i s a b i l l t i e s i n c r e a s e tbe v e r t l c a l t r a v e l of the COG, w i t h c o r r e s p o n d i n g i n c r e a s e s in e n e r g y cost. P e d e s t r i a n s w i t h disabilities stay w i t h i n t h e i r a e r o b i c c o m f o r t z o n e b y c h o o s i n g l o w e r w a l k i n g speeds. P r o s t h e t i c a n d o r t h o t i c d e s i g n s h o u l d a i m to m i n i m i z e the v e r t i c a l t r a v e l of t h e COG. T h i s v e r t i c a l t r a v e l c a n r e d u c e d to zero b y the u l t i m a t e m e b i l i t y aid: the w h e e l c h a i r .

Data relating the gait psttems of children with spastic cerebral palsy (CP) to metabolic cost and stability are presented. Results Indicate that while metabolic costs and vadabitity of heed traJectoP/are higher than in non-disabled peers, the gait pattern Is nevertheless optimal for that individual. In Wing to understand this phenomenon, the notion is pursued that pattern development and functional edaplatlons In locomotion ate driven by the underlying dynamlos of the task and the dynamic resources available to the individual. A theoretical model of the gatt is proposed that employs the Newton.Eu[er Equation of Motion tot a forced hybdd pandulum-sprlng model of gait to identify the needed resources. The model can potentially be used to parse out the energy forms that are needed for continued oscillations, and relate the energy forms to the dynamics that are available to the CP. The avatiable dynamics a r e then related to the observed gait patterns. The gait patterns that amarge are argued to be a consequence of a system that takes edvantaga of the limited dynamics that ate available to a system that has suffered a neurolog:cat insult, tt ls hypothesized tha children with CP assume a gait pattern that allows them to take advantage of the spdng properties of muscle and connective tlasue In the absence of adequate forcing due to weak musculature. In a nutshell, an understanding of ordered and disordered gaits as alternative functional solutions to the dynamical requirements for continued cycling of a pedodioally toreed oscillator Is explored. Applications of the approach to clinical evaluation and therapeutic Intervention are proposed.

538

541

THE CAST PROTOCOLFOR GAFFANALYSIS: FEATURESAND AN EXAMPLEOF COMPARISONWITHANOTHER COMMERCIALEXPER/MENTALPROT(X3OL.

GAIT PERFORMANCE AND R E L A T E D FACTORS 1N H E M I P L E G I C

Ugo Ddla Cto~

IKtitatoFisiologiaU r ~ , "I.a Sapicnza"Univcmi~,,Rome,Italy

S- 115

SUBJECTS Maureen K. Holden, Ph.D., P.T.

M G H Institute of Health Professions and Speulding Rehabilitation Hospital, Boston,/viA

In order to make movcn~mt analysis cffective in tim solution of clinical problems, a strnctomdconcelmml background is nccckd m additionto sazMatdiz~ dcfinitiom~ d methods Tl~ Calibmt~l An~omicat System T(w.hnique (CAST) is an pretzel that deals with thc~ prc~cmtsinmlmion with the r ~ o ~ o n of tl~Ix~itioa and odemmion of 11~ human l~Ivisend the lower limb bones in SlmOC din'ragtl~ e,xcc~on of l~om~tion and physicslcxcrcis~usinga stemophotognmmae~c systmnin ord~ to make pcs~'ate a com#ctc quamitam~ dcserilsionof joint kinematics and dynamics forboth researcha,d applicatioaThe r ofanatomicalIsn&mark isused aml bone crab.Idealanatomicaltctcmncc systemsare dcfm~.

The aim of this studywas to comparethis protocol and the Saflo clinical protcx;olfor gait anal)~isused withtheELITE (BTS, Milan) s3'stems.This was done usingthe same dam acquisitions~,sians.Markers were placed on the body so thatthe i"~mrm~nts of both

protocols ~ satisfied. For tl~ identification of the anatomical landmarks antixr~ome(ricexternal manual mea.mtememsare required for the SafioprotoQolwhile, using the CAST protocol two additional acquisitions, at leastone flarae tong, were made noomsaryfor the landmark cah~orat]on.Gait t~fiats,~zrr acqvh'ed and the kinematic degreesof fi~dom va:n: ~ for both the protocols.

539

Characteristic problems in the gait patterns of patients with hemiplegm secondary to stroke and other causes ~ be r e v i e w ~ and discussed from a historical perspective and in light of more recent findings from cases seen at the Spanlding Gait Laboratory. The role of different impairments in producing various gatt problems will be discussed, and related to theories for treating such problems. Examples of problems discussed include gait asymmetry, stance instability, knee hyperexteusion and altered dynamic E M G patterns. Evidence that, for selected patients, treatments aimed at strengthening of the plantarflexor muscles, as opposed to the traditional approach of inhibiting the plantarflexors muscles, will be considered. The role of posture and balance control in the upright position, a requisite skill for many functional activities such as sit to stand, turning and reaching, as a factor in locomotor ability will also be examined Factors such as symmetry of two foot standing and limits of stability patterns for weight shifting tasks in different postures will be compared to functional gait performance demonstrated by hemiplegic subjects.

542 SHOE LIFT A N D STIFF KNEE GAIT Hussein Abdulhadi, M.D. D. Casey Kerrigan, M.D.,M.S. Paul Larala, M.D. H a r v a r d Medical School, Division of Physical Medicine &Rehab.

Stiff k n e e gait is a c o n s e q u e n c e o f m a n y p a t h o l o g i c a l conditions. Patients with this type of gait a r e unable to flex their knees in e a r l y o r late s t a n c e . In addition, these patients a r e unable to voluntarily flex their knees d u r i n g swing. In o r d e r to clear the involved e x t r e m i t y during swing, these patients e m p l o y energy consuming maneuvers : H y p e r e x t e n s i o n of t h e u n i n v o l v e d s u p p o r t i n g l i m b at m i d s t a n c e , v a u l t i n g of t h e contralateral supporting foot, a n d ipsilateral hip hiking. In this article, we look at the e n e r g y cost of walking in n o r m a l subjects w i t h t h e i r k n e e s i m m o b i l i z e d u n i l a t e r a l l y in full e x t e n s i o n t h r o u g h out t h e gait cycle. This was c o m p a r e d to t h e w a l k i n g e n e r g y cost of the same subjects after the addition of a shoe lift to contralateral foot of the fully extended knee. Also p r e s e n t e d , a r e the kinetic and kinematic changes of gait after the immobilization of one k n e e in full e x t e n s i o n . T h e s e kinetic a n d k i n e m a t i c changes are c o m p a r e d to those b r o u g h t about by the addition of a shoe lift to the contralateral foot of the fully extended knee.

A F O R W A R D D Y N A M I C S M O D E L OF G A I T T O P R E D I C T M O T I O N F O L L O W I N G T R E A T M E N T A N D TO U N D E R S T A N D T H E B A S I S OF SPASTIC PARETIC STIFF-LEGGED GAIT Roth, R o b e r t S., PhD, K e r r i g a n , D. Casey, M D We p r e s e n t a f o r w a r d d y n a m i c m a t h e m a t i c a l m o d e l of two d l m e n s i o n a l g a i t . T h e e n t i r e body, feet, legs, thighs, and t o r s o is m o d e l l e d as a l i m p e d m a s s - r i g l d l y llnked s y s t e m to c a p t u r e the i n t e r a c t i o n s w h i c h o c c u r d u r i n g m o t i o n . A full set of m u s c l e m o d e l s a r e p r o v i d e d to accurately slmulate the physical torque mechanisms. E a c h m u s c l e is m o d e l l e d as a s p r i n g - d a m p e d contractile s y s t e m w h i c h c a n be c o n t r o l l e d b y o n - o f f s w i t c h e s . T h e dlfferential e q u a t i o n s of m o t i o n are d e r i v e d f r o m Hamilton's Princlple. This forward model provides the g a i t m o t i o n w h e n the s p e c i f i c m u s c l e a c t i o n is d e f i n e d . As a b y - p r o d u c t , j o i n t f o r c e t i m e h i s t o r i e s are d e t e r m i n e d f r o m the m o t i o n . T h e s p e c i f i c p r o b l e m of s t i f f l e g g e d g a i t is a d d r e s s e d in t h e paper.

S-116 543 THE ROLE OF CLINICAL GAIT ANALYSIS IN THE TREATMENT OF PERSONS WITH NEUROMUSCULAR DISORDERS

S. 0unpuu. M.Sc., R.B. Devis, Ph.D. and P.A. DeLuc& M.D. Gait Analysis Laburatury, Det~ment of Orthopaedics, Newington Children's Hospital, Newington,Connecticut Gait analysis has become a routine clinical decision-making tool in the treatment of gait abnormalities in many center. Gait analysis is most commonly used for surgical decisionmaking but is also useful in the evaluation of orthotics and other cons~vafive treatment plans. The primary purpose of clinical gait analysis is twofold: to define the exact gait abnonnaiities and to help deteenaine the possible causes of these gait abnormalities through the objective docuraenmtion of gait. With this infmmation, a more substantial basis for t r ~ t m ~ t decisions is possible. The intetlnetation of gait analysis data is an involvedprocess which requiresthe integration of many types of information including one's visual impression of gait, passive joint range of motion, estimation of bony defotmitien, muscle strength, muscle conl~ol and muscle activity (EMG), joint kinemalics and joint kinetics during gait. Each of these components has a vea'yimportant role in determining the causes of guit abuormatities, but each component is singularly incomplete. The skills for these data interpretation are developed over time with an understa~ling of the methods of data conectlon, with experience and with the knowledge developed through the evaluation of gaitdatabothbefore and after treatment. The post treatment gait analysis is anothe~ major utility of gait analysis as it allows quantification and facilitates systematic review of the pre to post treatment changes. Through this evaluation, modiftcation and ultimately improvement of U~mnent appmache~ are mote likely to occur. Gait analysis dnes not dictate a specific ~e.atment response, as this is dependent on the philosophies of the Izenting physician, but it provides additional information so that more informed treatment decisions are possible.

Gait Analysis

Maintaining Employability through Engineering 544

546

THE ADA, JOBS,

AND REHABILITATION ENGINEERING John H. Leslie, Jr. Pli.D. Executive Vice President The Cerebral Palsy Research Foundation of Kansas

A NON-INVASIVE METHOD FOR THE EVALUATION OF CARPAL TUNNEL SYNDROME Zeynep Erim, Ph.D. NeuroMuscular Research Center, Boston University,Boston, MA

~lie ADA is a marvelous piece of civil rights legislation. However, for people with severe physical disabilities, it is not bee*~ the panacea particularly in the area of employmen%. Recent statistics indicated that most individuals seeking tlie protection of the ADA have been employed and have suffered injuries through c u m u l a t i v e t r a u m a disorders, ie., carpal tunnel syndrome, and~or low back injury/pals. The Cerebral Palsy Research Foundation of Kansas has 23 years experience helping to employ severely disabled persons through the media of rehabilitation engineering. This paper will examine the requirements necessary for persons with severe disabilities to become productive on tlie Job in white and blue collar environments. The elements of engineering, ergonomics, business policy, financial issues and value engineering will be examined from a holistlc standpoint. Selected slides will be utilized to demonstrate worksite accommodations. The future will be examined in light of the 1994 election.

The ptapose of this study was m employ surface electromyography (EMG) and force measurements in the design of an objective, quantitative, and non-invasive method to evaluate impairment resalting from the carpal tunnel syndrome (CTS). The method was based on the premise that given a task which could be accomplished by contributions from two muscles, impaltmcot in one muscle will be reflected in the increased preferential use of the other, unaffected muscle. The Abductor Pollicis Brevis (APB) muscle which receives ianervation from the median nerve damaged by CTS, and the Opponeas Digiti Minimi (ODM) which is innervated by the ulnar nerve unaffected by the syndrome, were studied. A testing device was built to allow for the measurement of force and surface EMO on both muscles while the subject performed a pinching motion with the thumb and fifth digit. The subject was provided with visual feedback of only the summed result oftbe force contributions from the two muscles and was asked to maintain this sam at a displayed level. No instruction was given as to the relative coh~bations from the individual muscles. It was hypothesized that higher levelsof involvement with CTS would cause the subject to produce a larger pocdco of the force with the unaffected ODM muscle than would be seen in healthy controls. Preliminary analysis of the individual force and EMG signals from the two muscles of five normals and five CTS patients revealed different trends in patients and normals as expected. Regressionanalysis on these signals yielded several parameters that had statisticallydifferentmeans in the two populations. These parameters could ultimately be used to classify a subject in the normal or CTS groups, and to devise an index to provide a scaled measure of severity of damage,

S-117

547

545 JOB A C C O M M O D A T I O N F O R E M P L O Y E E S A G I N G W I T H A DISABILITY D.R. M c N e a l , N . S o m e r v i l l e , D. w i l s o n Rehabilitation Engineering P r o g r a m R a n c h o Los A m i g o s M e d i c a l C e n t e r , D o w n e y , C A

As employees with disabilities age, their needs for job accommodation may change. If these needs are not met, the employee's job may be in jeopardy. This presentation describes a study in progress that is investigating the job accommodation needs of employees with a disability to determine how their needs have changed over their work history and what has been done to meet these need~ Information is being gathered through a mailed questionnaire and structured telephone interviews of 150 individuals who either had polio, have rheumatoid arthritis or had a spinal cord injury. Preliminary results indicate that these individuals do have a number of problems that require accommodation and that these problems change during their work history because of new work assignments and continuing functional declines. These data will be reviewed along with information about accommodations received (types of accommodation, cost, and the impact that the accommodations have had on the employee's ability to do their job).

AN EMG-BASED SYSTEM FOR IDENTIFICATION OF MUSCLE IMPAIRMENT IN LOW BACK PAIN Buijs, R.J.C, De Luca C.J., Gilmore, L.D., Maloof, P., Oddsson, L.I., Roy S.H.

NeuroMuscular Research Center, Boston University, Boston, MA 02215 This presentation describes the underlying basis and clinical applicability of the Back Analysis System (BAS). The BAS is a computerized surface electromyographic (EMG) system under development to objectively diagnose and monitor muscle impairment in patients with low back pain. The system uses spectral EMG data from six muscle sites on the lower back detected by surface EMG electrodes during sustained isometric contractions. Preliminary results using a laboratory prototype of the BAS have demonstrated that the system provides a reliable measure of muscle performance associated with different impairment conditions. The clinical implementation of the BAS is designed to allow clinicians with minimal experience in EMG assessment to identify and monitor back muscle impairment in patients. It accomplishes this goal by means of a simple, easy-toadminister procedure that automatically generates a clinical report afser completion of a back evaluation test. Some of the key features of the clinical system are: (1) a newly designed pastural restraint apparatus that positions the subject in a comfortable, optimized posture, (2) a user-friendly '%Vindows" interface that hides the underlying complexities of the digital signal processing hardware and soi%Ware to the operator, and (3) a relational database management system that ensures that collected data can be automatically stored and retrieved by the system.

S- 118

Poster Presentations

548

551

EFFECTS OF AGE IN THE MODELING OF ISOMETRIC LIFTING STRENGTI~ Trina A. Bubr and Charles B. Wooltey The University of Michigan

ESTIMATION O P B O D Y SEGMENT KINEMATICS A N D D Y N A M I C S FROM ]tILATERAL FORCEPLATE D A T A IN STA/qDING SWAY J. Mizrahi, O. l.zvin, M. Shohem, E. Isakov and Z. Susak Techeien-IIT, Hails and Loewenstein Rehabilitation Center, Raanene, Israel

Significant decreasesin the muscM force productioncapabilities of humse beings have been shown to begin in the fifth decade of life. The magnitude of this decline in strength depends on the individual as well as the specific muscle groupstested within the iodividual. Because of the varied age-related effects on muscles throughout the body, it is unclear bow the combined changes in muscle force production capabilities could alter whole-body lifting performance. This study examined selected isometric exertion scenarios is an older population to determise how whole-body strengths were affected by the specific strength decreases obtained from a popalation-hosod study of older individsals. In the first phase of this study, a computerized strength prediction program was used to predict changes in whole-body strength with age. For the second phase of the study, the postures for the simulated lifting tasks were altered to examine how older individuals could modify their postures to improve their maximum exertion capability. Results indicated severely compromised lifting capabilities of the older individuals due to the declines in specific muscle strengths. In particular, knee and hip extension strength declines played a significant role in reducing whole-body exertion capabilities for the simulated postures. It was further predicted that older individuals could increase their lifting capabilities by slightly modifyieg their postures. Future research will focus on recording a,d analyzing these modified postures in older adults.

In this study we developed a method for the estimation of postural standing dylaaznics of the body, without direct kinematic measurements. The ~ s t i a s of the body segments and, conseqtmady, the forces and moments acting in the ankles and hips were evaluated from bilateral fore,plate measurements and sethiopemctrlc data only. A 31), 4-joint, 5--scgmcot model of the human body was used. The model was identified as s spatial 4-bar mechanism, connected together by 4--isis% each allowing rotation about two axes. This model was capable of describing the motion of the center of ~csvity of the body separately in the enmnal and esgittal planes. A developed invca'se kinematics algorithm was used to evaluate the kinematics of the body segments. The forces and moments acting in the joints were next resolved. The model results served to compe~ the foot-lgouod reaction forces on sue foot to those actually measured. The source of errors included estimation of the exact location of the ankle joint and enthrepome~c data. The relative errors obtained were 16 to 42 percent for the forces and 23 percent for the moments. These re'ors are of the same order of co'ors as reported in direct kinematic measurements. Acknowledgemems: This study was supported by the Segal Foundation and by the Walter and Sandra Kaye Fund.

552

549 STUDY OF THE ENERGETICS OF BIARTICULAR AND U N I A R T I C U L A R A C T U A T O R S Dariush, Mohamad Parnianpour, and Hooshang Hemami T h e Ohio State University

APPLICATION OF NARMA MODELLING TO RECONSTRUCT CENTER OF PRESSURE ATTRACTORS Mulavara, A.P. , Qammar, H.', Verstraete, M.C., and Simon, B.N., Dept. of Biomedical Engineering, The University of Akron Dept. of Chemical Engineering, The University of Akron

A mnsculoskeletal model based on the important properties and parameters of the hum~a b o d y has been developed. T h e body is represented as three rigid segments in the sagittal plane: leg, thigh and toreo. T h e muscle actuators are represented by a nonlinear, viscoelastic model. T h e kinematic and kinetic redundancies in nettromusculosketetni systems provide opportunity to optimize performance. In this study, the kinetic redundancies are investigated by developing different control strategies during performance of two functional tasks: Sqanting and. Bowing. T h e identical movement trajectories were i n p u t to a three link model of t h e leg, thigh, and trunk. Ten u n i a r ticulas and biarticular muscles were partitioned into two functional groups: movement generators and gravity compensators. Eleven simulations were performed to evaluate the role of biarticular muscles by comparing the c o m p u t e d adjusted mechanical works. T h e results indicated t h a t the biarticular muscles provide a more efficient load sharing amongst t h e actuators, Moreover, the simulations show t h a t majority of the work performed by the muscle a c t u a t o r s is used to compensate for gravity. T h e aasigament of actuators to each functional group should be considered as a dynamic phenomena, since the optimal partitioning was shown to be task specific.

The mammalian nervous system has been suggested to exhibit chaotic behavior. Recent studies, using chaos time-series techniques, have indicated that the human postural control system may be modelled as a random walk process, that is, the data are correlated noise. However it is known that "in general the detailed diagnosis of chaotic dynamical systems requires long time series of high quality- (Reulle, 1987). Such difficulties ~ay be overcome b y estimating models for the system and then using these models for generating longer sequences of data. One of the techniques that has been successfully applied to the generation of nonlinear chaotic models are NARMAX (Nonlinear AutoRegressive Moving Average models with eXogenoue inputs) models. The advantages of this modelling technique are its ability to utilize short data sets for str~cture selection and parameter estimation and also the robustness of the deterministic part of the model to moderate amotL~tS of white and correlated noise. NARMA models, for systems without exogenous inputs, have also been successfully utilized in identifying some chaotic systems. In this paper N A R M A m o d e l s were developed for center of pressure data obtained for seven subjects under four different sensory conditions. Preliminary results indicate that the N A R M A m o d e l s yield short time series models, but the long time series models are of insufficient quality to apply chaos analysis techniques. The generation of such N A R M A m o d e l s m a y provide another approach to identifying human postural dynamics, which have wide ranging applications.

A THEORETICAL

Behzad

550

553 A

NEW QUANTITATIVE APPROACH FOR TREATMENT EVALUATION OF PARKINSON DESEASE

R.M. Hasherni, M. Zardoshti-Kermani, E. Fatemizadeh Biomedical. Eng. Lab., Amittabir University of Technology, Tehran, Iran A .ew quantitative approach has been

usedfor both estimation of Parkinson desease levelS (i,II,III,IV) and its treatment evaluation by utilizing interpretation of handwriting. The approach is based on both bandwdtingdynamics modeling and acquired dinioal data. A set of optimal handwriting test patterns have been designed based on Paddnson charactedstics and its symptoms. Pattern recognition techniques such as dominant points detection, principle curves, neural networks and genetic algodthms have been used to extract optimal features for quantifying three major Paddnsonsymptoms, namely rigidity, tremor and Idnesia which subs~uently used for desasse evaluation. The approach has been evaluated on handwriting data scqui~l from 70 persons with Parkinson symptoms and 30 healthy ones. The estlrnatod levels of r by our approach have been compared with an expert's report which show better than g0% match. Furthermore by follow up the patients, a quantitative treatment evaluation has been reported which matches the expert's repoct.

TRANSCUTANEOUS MODELING FROM VOLTAGE AND CURRENT PULSE INVESTIGATIONS. C. Murray, IL Reilly. Dept. of Elsctronic and Electrical Engineering, University College Dublin, Dublin 4, Irdaod. Human muscle was stimulated transcntanconsly with voltage pulses. Using a short burst monophnsic periodic stimulation train, of pulse width 0.2m% tho corresponding injected cement waveform implied the presence of a memory r It was also noted that the form of the response was amplitude dependent, resulting in two broad uatagories for low and high stimulation intensity. Based on these ebsarvatioas the ~mulator muscle interface was mcceasfidly modeled using equations in the Hodgkin-Hurdey form. To further validate this model simulated stimulation with constant r wa= r out and was found to be in agreement with actual experimental mnssmeraents. The ~sults indiuate an explanation for the frequency response at high and low stimulafice intensities ned give an attentative perspective to the classic Bode representation of impodmu~ with stimulation f~l~aCy. T h e ~ results have yielded a new viewpoint on a feedback mcchonism based on what wc term "impodan~ goniomctr/" for application in fuagfioual electrical stimulation.

Poster Presentations

S- 119

554

557

IN'I'ELMGENTPER~HERAL FOR A U ~ A T I V E COMPUTER A C C K S S M.J, Wang, X.L, Dadmgton l.G. Faculty of Art, Design & Tedumlogy, Cardiff Institute of Higher F,.ducatton, Wales,U.K.

"ATTITUDE VECTOR" & EULER/CARDAN REPRESENTATIONSOF L/E KINEMATICS Sreesha Rao, MS, Ernest Bontrager, MS, Chns Powers, MPT, Jacquelin Perry, MD Rancho Los Amigos Medical Center, Downey, CA.

Severely handicapped Individuals find it difficult to access standard c o m m e r ~ software that uses graph/ca] inte~acns because of the need for precision hand mow~nts. A 'Swi~se', developed in the Faculty, allows sevesely handicapped users to overcome this Interface problem. However, the device is based on danple switches which have uncompromLdng passive characteristics that are intolexant to tremor and grons movesnm~ts.

Euler/Cardan methods [1,2,3] for determining 3D joint angles are popular among clinicians within the biomechanlcs r because interpretation of the results is physically meaningful. The "affitude vector" approach has not gained wideapread clinical acceptance despite obvious advantages related to orthogonal component axes, unambiguous measurement properties and an absence of singularitiea [4]. The resistance to the adoption of the 'attitude vector" approach appears to be a lack of physical (geometric) representation of the three attitude vector components in terms of rigid-body rotations. The present study demonstrates a clinically meaningful variation of the "attituda vector" approach applied to the measurement of lower extremity joint and segment kinemmics. A 6 camera VICON motion analysis syatam was used to test 9 normal subjects during walking at a self-selected free pace. The foot, shank, thigh and pelvis attitude vector components were determined relative to the laboratory axes. 3D joint angles at the ankle, knee and hip were obtained by subtracting the attitude vector components of adjacent segments about the laboratory axes. Joint angles were also computed with the Euler/Cardan method. The results showed excellent agreement between the two methods for sagiltal plane motions, Frontal and transverse plane Euler angles showed cross-talk with the sagittal plane motion at the knee and ankie between 55% and 85% of the gait cycle.

(YCom~

A new approach has been used which involves the continuous monitoring of the mowanents of the communicator. An artificial Intelligence system learns to make judgements about the Intention of the communicator, alleviating the need to actually press the switches. The communication process is speeded up, Rducing the frustration of the user, and makes effective use of m u l t ~ l i a computer products a reality. The user works In an area whene a number of targets are defined. These targets replace the switches. A learning phase is used wtereby the user's hand movements towards a specified target are tracked. The changing coordinates are fed Into a neural network which 'Mares' to recognise typical paths to each target. In the operational phase the system uses its knowledge of the user's movement patterns to predict which target is b e r g sought. The switch risked to that target can then be activated prior to the target being h i t The system is tolerant of gross movecnant and ~ because the learned characteristics of the u s ~ are taken into account by the neural network when predk'tions are made. The characteristics of each new user can be learned and stored for future referemce by the system.

555

[1] Chao, E. Y. S. J. Biomechanics 13: g89-I006, 1980. [2] Grood et aL J. Biomech. Engng 105:136-144, 1983. [3] Kadaba et aL J. orthop. Ras. 8: 383-393, 1990. [4] Woltring et aL J, Biomechanics 27: 1399-1414, 1994,

558

TACTILESENSATION~ O L D S :

A COMPAPJSON B E T W E E N PERPENDICULAR

A N D TANGENTIAL ~ A ~ O N

Claudio .-%.Perez and Marcels E. Osorlo Dept. of Electrical Engineering, Uni'awsityof Chile. Av. Tapper 2007, Santiago, CHILE

MINIATURIZED MULTI]PROGRAMMABLE NEUROMUSCULAR STIMULATOR USING A NEW TELEMETRY TECHNIQUE K. Arabi and M. Saw-an Department of Electrical and Computer Engineering,lh:olePolytedmJqnode MontrEal P.O.Box 6079, Station Cen~-ville, Manueul, Qno., Canada H3C 3A7

Tactile stimulation is a means of trartsferring information to visually handicapped individuals. Part of the effort in developing new devices for the blind community requires an appropriate specification of the vibratory stimulus matching it to the tactile system characteristics. The present study compares v/brotaetile sensation thresholds in five subjects for perpendicular and tangential excitation. Sensation thresholds were measured using the method of the limits. The device used as a vibrator was a rectangular piezoeeramic bender with dimensions 21.0 rma by 3,45 mm and 05 mm thickness mounted in a cantilever manner. The experiments were performed on the index finger while the hand rested on s table. The vibrator contacted the skin through a 5.45 mm hole on a metal plate. The bimorph was excited with sinusoidal waveforms of 25, 40, 60, 70,100,150, 200, 250, 300, 350, 400, 500, and 700 Hz. Displacement was measured w/th an optical sensor to avoid contact with the vibrator. The results show that sensation threshulds are in the range 1.83.1/am for tangential excitation and between 2.0-5.3 ~am for perpendicular exciradun. As expected, for both modes of excitation the minimum threshold resulted for 250 Hz. In genera], sensation thresholds were lower for the tangential excitation compared to perpendicular excitation. Statistically significant differences (p
It is well known that the majority of commercially available in:plantable systems are dcd/catcd to a specific applicationand they do not completelysat/:;fythe reqinrcmcnt of other applications. They are also designed to deliver a limited stimulationpattern and are not able to support the new stimulationstrategies for selectivesthnulatian of serves nod muscles. This paper presents a new implantsble microstimulator system for neuromuscular stimulation. The implantable circuiuy is externally controlled and powered by a single encoded radio frequency cartier ,rod e'ay have up to 16 independently controlled bipolar channels. A secure communJcaion protocol permits the implant to recoperate safely the data and clock coming from the exlermd conlmller. A conventionalforward error correction (FEC) tedmiq~ has b~n employed and a n~m,'cl implementation has been proposed. An optical telemetry link ,ransmitsthe stateof patient to the cxternal controller. Thc telemetry link allowsthe physicianto optimizethe stimulation parameters after implantation and verify the functionality of the implant. The implant system is adaptable to patient needs and the futore development in stimulation algndthins. The microstimelatar is able to generate a wide range of stimulation patterns including selective stimulation wavefcLms. .The chip uses circuit ~chniques to reduce power consumptionand insure long-term stability. The implantable system can be used to any neuromuscular applications. It is intended to restore normal bladder functions to spinal cord injured patients. The character)sties of the implant satisfy the exigencydemand of implanteddeviensand exceed the requirements of actual funotioanlelectrical stimulation(FES) applications.

556

559 A COMPARISON OF EMG AND ENERGY IN GAIT INITIATION Barry N. Simon, Ajitkumar P. Mulavara, Mary C. Verstraete, Chris A. Miller" Dept. of Biomedical Engineering, The University of Akron "KRUG International, NASA Houston Space Center

Abstract N o t Available

and

The dynamic characteristics of a mechanical system are most easily determined as the system moves from one steady state to another. Thus, the interrelation of system parameters for human gait analysis may be studied during gait initiation. The ability to quantify the changes in mechanical energy of the body from the ~MG activity of the major muscles would provide a simple tool that has wide ranging applications in both clinical and sports biomechanics. This abstract presents the preliminary results of that study. Seven adult males participated in the study. Using rigid body mechanics, the kinetic (translational and rotational) and potential energies of each body segment were calculated from a quiet standing position (steady state condition l) to a uniform walking condition five steps later (steady state 2). EMG activity was collected from the tibialis anterior, gastrocnemius, rectus femoris, vastus lateral)s, vastus medial)s, biceps femoris, semitendonosis, gluteus maximus and gluteus medias. The results of a qualitative analysis indicated that a relationship between EMG activity and body segment energies may be developed. Two trends were easily identified: the period of the muscle activity decreases with each step, while the period of the variation in segment energies increases with each step. It was also concluded from the qualitative analysis, that in order to quantify this relationship, a multi-link model musu be utilized.

S- 120

Poster Presentations

560

562

A RELATIONSHIP BETWEEN EXTERNAL AND INTERNAL ENERGY DURING DORSI/PLANTARFLEXION Barry N. Simon, Ajltkumar P. Mulavara, and Mary C. Verntraete Dept. of Biomedical Engineering, The University of Akron

AN A S S E S S M E N T O F H O W T H E B R I T I S H A S S I S T I V E T E C H N O L O G Y S E C T O R C A N BE S T R E N G T H E N E D IN O R D E R T O F A C I L I T A T E TECHNOLOGY TRANSFER Carolyn Young and Jim Sandhu S p e c i a l H e e d s R e s e a r c h Unit, U n i v e r s i t y of N o r t h u m b r i a at N e w c a s t l e

The work-e~ergy approach towards the analysis of human motion has wide ranging applications in both clinical and sports biomechanics. This study is part of an ongoing project investigating the relationship between the energy in the EMG signals from the tibialis anterior and the gastrocnemius muscles and the external work performed during dorsi/plantarflexion movements. Four subjects participated in the study. A BIODEX 2000 was used to measure torque produced, angular velocity, and angular position at a rate of 100 Hz. Nine cycles of a dorsiflexion/plantarflexion movement were collected at each of four speeds. Preliminary results indicated that there exists a relationship between the external work done and the energy in the EMG signal. External work and internal emergy exhibited similar responses co variations in testing speed and applied force, with one subject's responses being opposite the responses of the other subjects. However, intersubJect variations in muscle cross sectional area and intrasubject variation in the level of effort during testing hinder the development of a direct relationship between external work performed by a muscle and the internal EMG energy of the muscle. Similar results are seen for both the tibialis anterior and the gastrocnemius, with a greater degree of variation for the tibialis anterior. This was likely due to the subjects' stated opinion that the dorsiflexion portion of the movement was not common, indicating a less refined level of motor control for the tihialis anterior in comparison to the gastrocnemius.

561 SENSORS SIMULATION FOR FES RECIPROCAL GAIT FROM UNIMPAIRED SUBJECTS WALKING WITH STRATHCLYDE-DUNDEE RGO ORTHOSIS Y Yang, J P Paul, M H Granat, L Yang Bioengineering Unit, University of Stmthelyde, Glasgow, UK Abstract Computer simulation has been applied to a variety ol research in Bioengineering for many years. It is important to obtain different sensors' OUtput signals which indicate the status of gait and evaluate them for FES gait restoration. In this study, sensor signals have been simulated for leES reciprecal gait from unimpaired subjects walking with Strathclyde-Dundee RGO orthosis. The kinematic and kinetic as well as EMG signals were detected from three normal subjects with the above ortbosis and crutches by the VICON gait analysis system and Kistles force plates and an EMG system. About 20 real & imaginary sensors which may he utilized to indicate the state transitions of gait have been simulated by means of com~ter program "MIMIC" according to the kinematic and kinetic data. The results of sensor simulation will he further used in research for the development of a hierarchical finite state rule-based eontroll~ for a hybrid FES orthosis system.

T h e B r i t i s h A s s i s t i v e T e c h n o l o g y m a r k e t is ~ m i n a t e d by s m a l l and m e d i u m s i z e d e n t e r p r l s e s . E v i d e n c e s u g g e s t s that the n u m b e r of B r i t i s h A s s i s t i v e T e c h n o l o g y S M E s is in d e c l i n e . D e s p i t e the v a s t m a r k e t for a s s i s t i v e t e c h n o l o g y w i t h i n E u r o p e m o s t h i g h t e c h n o l o g y p r o d u c t s u s e d in the UK are i m p o r t e d from t h e US. D e s p i t e the r e c e n t i n n o v a t i o n s in r e h a b i l i t a t i o n e n g i n e e r i n g , t h i s e x p e r t i s e w i l l not be e x p l o i t e d to its full p o t e n t i a l if the f i r m s do not e x i s t to u t i l i s e this t e c h n o l o g y . T h i s paper d e s c r i b e s a p i e c e of r e s e a r c h u n d e r t a k e n to s t u d y the p r o b l e m s e x p e r i e n c e d by B r i t i s h A s s i s t i v e T e c h n o l o g y s m a l l and m e d i u m s i z e d e n t e r p r i s e s to d i s o c o v e r how t e c h n o l o g y t r a n s f e r can be f a c i l i t a t e d w i t h i n the a s s i s t i v e t e c h n o l o g y s e c t o r . It l o o k s at the c o n s t r a i n t s w i t h i n the a s s i s t i v e t e c h n o l o g y s e c t o r in a E u r o p e a n c o n t e x t , the p r o b l e m s e x p e r i e n c e d by small firms, A m e r i c a n l e g i s l a t i o n and J a p a n e s e m e t h o d s .

ANNALS OF BIOMEDICAL ENGINEERING Author Index V o l u m e 23, S u p p l e m e n t 1, 1995

(Numbers refer to abstract number, not page number.)

Aarsland A, 359 Abdulhadi HM, 539 Abdullaev YG, 325 Abrams, Jr. CF, 500 Abushanab H, 514 Achakri H, 169 Ackerman JL, 314, 479 Adamicza A, 43 Adamson SL, 214 Agali P, 209 Ahmad CS, 498 Akay M, 139, 445,446 Albright J, 503, 504 Alcock A, 431 Alexander R, 227 Aljuri N, 28, 59 Alloway KD, 412 Alpert NM, 315 Alt W, 243 Altman K, 380 Ambrose CG, 499 Amin VR, 328 Amory D, 126, 190 Amster I, 14 Anand S, 93 Andersen TT, 220 Anderson DJ, 402, 404 Anderson KR, 115 Anderson KP, 121 Anderson RJ, 202 Andreo MA, 499 Andreou AG, 371 Andrews B, 510 Antaki JF, 108, 111, 112, 215, 216 Aoki T, 116 Arabi K, 558 Arramon YP, 480 Aston H, 440 Atala A, 246 Atanasov P, 459 Ateshian GA, 463,497,498 Atkinson WA, 115 Atwater I, 242 Audi SH, 47

Aufdemorte TB, 218 Avignone E, 525 Aviram A, 508 Ayappa I, 46 Baba KR, 501 Babb AL, 32 Babich JW, 358 Bachrach NM, 462 Bada H, 80 Bae YH, 290 Baevsky RM, 180 Bai S, 461 Baish JW, 125, 168 Baker DA, 361 Balaban RS, 353 Baldwin SP, 234 Baptist-Nguyen SE, 40 Barahona M, 457 Barbee KA, 118 Barineau DW, 362 Barnes CA, 407 Barocas V, 249 Basser PJ, 303, 378 Bassingthwaighte JB, 78, 189, 317, 331, 336, 435,436 Bates JH, 41, 42, 61 Baudry M, 391 Bauman JW, 195 Baxter LT, 21, 83, 125, 167, 254 Bay B, 484 Beach JM, 122, 156 Beard DA, 189 Ben-Jebria A, 33 Bennett JC, 263 Bennett T, 207 Benni PB, 126, 190 Berg RA, 107 Berger TW, 391, 393, 394, 396, 398,423 Berk BC, 295 Berk DA, 125, 166, 255 Berkenblit S, 491 Berkman EP, 503 Bernard C, 449 Berry JL, 173 S-121

S- 122 Bertram C, 147 Betts F, 472 Bey P, 10 Bhattacharyya S, 309 Biegler FB, 466, 467 Biemann K, 12 Bijak M, 505 Billich T, 88 Biron RJ, 258 Bizios R, 127, 220, 222, 224 Black JP, 103 Blaise J, 397 Blankevoort L, 498 Blumenstock FA, 222 Bombard D, 491 Bonab AA, 315 Bonadonna R, 338,342 Bongard RL, 48 Bontrager E, 557 Booker F, 362 Borg T, 185 Borovetz HS, 81, 108, 111, 112 Borrello MA, 74, 75 Bosan S, 50 Boschetti F, 329 Boskey AL, 472 Boston JR, 216 Boucher J, 0 Boucher Y, 83, 167 Bowlin GL, 256 Brady TJ, 313 Braun RD, 155 Breinan HA, 257, 280 Breitenstein DS, 216 Brekken C, 167 Brodrick CW, 494 Bromberek BA, 249 Bronzino JD, 397 Brooker A, 333 Brown LV, 46 Brown N, 339 Brown R, 57 Brown TD, 464A, 493 Brownell WE, 287 Bruce EN, 54, 84 Brunengraber H, 0 Brunner HR, 135 Buckey JC, 424 Buckley JM, 187 Buerk DG, 191 Buettner HM, 270 Buhr TA, 548 Buijs RJ, 547 Bundy JM, 330

Author Index Buonocore M, 318 Burgreen GW, 112 Burke EJ, 89 Burke JF, 358 Burns J, 0 Burns MP, 175 Burrin DJ, 360 Bursac P, 258 Burstein D, 305, 311 Bush ML, 33 Butera, Jr. RJ, 441 Butler JP, 34 Butler KC, 108 Byrne JH, 441 Cabrera ME, 352 Cagan J, 112 Caldwell JC, 317,331 Caldwell MD, 249 Calvo WJ, 228 Camacho NP, 472 Cao Y, 307 Cape EG, 201, 301 Capitini LA, 171 Capps SG, 500 Cariani P, 386 Carney LH, 422 Carson RE, 316 Cartee LA, 377 Chait BT, 17 Chamberlain LJ, 259 Chan L, 106 Chance EM, 349 Chang YS, 160 Chao E, 333 Chao EY, 0 Chapin JK, 408 Chapman GB, 96 Chapman R, 0 Charles JB, 207 Chatham JC, 349 Chatterjee S, 76 Chauvet GA, 391 Cheever EA, 196 Chen AC, 489 Chen B, 126, 190 Chen W, 22 Chen X, 94 Cheng LL, 479 Chernyak YB, 124 Chesla S, 297 Chesler DA, 314 Chesler NC, 179 Chian MT, 396

Author Index Chien S, 145, 199 Chinkes D, 359, 363 Chittur KK, 474 Choi B, 120 Choi Y, 62 Cholewicki J, 534 Chomey GS, 462 Chow CC, 518 Christian BT, 315 Cima L, 24, 236, 237,272, 273,279 Cima M, 24, 273 Clark G, 431 Clark JW, 441 Clary T, 372 Clopton BM, 372 Clough AV, 60 Cmolik BL, 195 Cobelli C, 338, 341,342 Cogan SF, 403 Cohen M, 443,444 Cohen RJ, 124 Cokelet GR, 96 Collins DM, 151 Collins JJ, 439 Collins T, 115 Comte B, 350, 364 Conley KE, 523 Conover CD, 5 Constable T, 310 Conway P, 432 Cook AD, 260 Cook GA, 261 Cooper SL, 233 Cooper, IV G, 187, 188 Corbett S, 372 Corcoran PJ, 537 Corcos DM, 517 Cordeau S, 364 Cornhill J, 129, 332 Cosgrove, III DM, 101 Cowin SC, 470, 480, 483 Cowley, Jr. AW, 436 Crago PE, 507 Craig C, 213 Crane GM, 218 Cranstoun SD, 191 Crisafulli B, 329 Cuckler J, 495 Cullom S, 320 Curry FE, 194 Cuzwala P, 495 Dacey, Jr. RG, 165 Dai J, 192

Dailey KJ, 488 Daley M, 80 Damiano ER, 95 Dancy RN, 261 Dariush B, 549 Darlington JG, 554 Datta S, 411 Daubenspeck J, 446 Davies CR, 193 Davies MR, 509 Davies PF, 118 Davis R, 510 Davis III RB, 1,543 Dawson CA, 47, 48, 60, 66, 68, 77 Dawson M, 304 De La Fuente EAE, 77 De Luca CJ, 425, 439, 526, 527, 547 Dee KC, 220 DeHart M, 333 DeJuan E, 0, 388 Del Fabbro M, 86 Del Vecchio P, 127 Delabays A, 302 Delafontaine P, 206 Delcroix M, 67 Delhaas T, 184 Deligeorges SG, 369 Della Croce U, 538 Dellian M, 130 DeLuca PA, 1,543 deMelo E, 363 DeNicolao G, 341 DePaola N, 114, 175 Des Rosiers C, 350, 364, 365 Detres LA, 127 Deussen A, 204, 355 Deutsch S, 98 Devaney L, 406, 417 Dewey C, 115, 117, 133 Dhadwal AS, 71 Diamond SL, 93,226, 228 DiCarlo J, 409 Dick EK, 176 Dickey B, 40 Dietrich HH, 165 DiLorenzo DJ, 417 DiMilla PA, 262 Ding H, 232 Ding J, 392 DiStefano III JJ, 339, 340 Dolan P, 536 Dolman C, 61 Dongr6 AR, 13 Dorgan SJ, 511

S- 123

S- 124 Dorson W J, 82 Drazen JM, 26 Drues ME, 128 Drumheller PD, 235 Drzewiecki GM, 181, 182, 183, 212,448 Du W, 225 Duerk J, 134 Duling BR, 95, 122 Duncan JS, 310 Durand D, 366 Durfee W, 506 Dustman T, 121 Easterling M, 14 Eberhardt AW, 464, 495 Eckberg DL, 84 Eckman JR, 286, 294 Eckmann DM, 35A Eckstein EC, 94 Edell DJ, 406, 415,417 Edelman ER, 252 Eidelman D, 61 Eisenberg SR, 501 Eller S, 23 Elliot AR, 429 Ellis JT, 99 Ellsworth ML, 151 Emley M, 425 Engin A, 535 Epstein MF, 263 Erecinska M, 346 Erim Z, 546 Ershler PR, 121 Espinosa FF, 25 Espy-Wilson C, 370 Evancho MM, 174 Evans JM, 84 Fagette, Jr. PH, 3, 4 Fan M, 168 Fatemizadeh E, 550 Federspiel W J, 81 Feigl EO, 335 Feldman AB, 124 Fenton BM, 272A Fernandez CF, 0 Fernie G, 0, 530 Field S, 182, 448 Finegold DN, 201 Finley CC, 377 Fischman AJ, 358 Fischman GS, 473 Flaherty JF, 171 Flaherty JE, 497

Author Index Flesche CW, 204 Florez L, 114 Foley JM, 347 Fontaine AA, 99 Forder JR, 349 Forrester T, 151 Frame MD, 158 Frangos JA, 160 Frangos JA, 197, 223,264 Frank E, 491, 502 Frazer E, 360 Fredberg JJ, 36, 37, 38, 39 Freed LE, 258,265,291 Frei C, 147 Freitag L, 28 French AS, 418 Frey EC, 324 Frey U, 440 Friberg TR, 201 Friedman M, 422 Friedman MH, 171 Fritton JC, 471 Fritton SP, 469,485 Fronticelli C, 8 Fry DL, 171 Fu B, 194 Fukamachi K, 193 Fukumura F, 193 Fuller M, 121 Fung L, 23 Fyhrie DP, 481 Gadaleta S, 472 Gaddipati A, 129 Gallagher G, 225 Gan RZ, 69 Ganobcik S, 299 Gao BZ, 144 Garcia A, 502 Gardner JF, 200 Gardner TR, 498 Garrido L, 314,479 Gaumond RP, 200 Gaver, III DP, 27, 35A Gavriely, N, 35A Gazit Y, 125 Geha AS, 195 Gemmell CH, 103 George DT, 195, 196 George SC, 32 Gerritsen ME, 133, 224 Geselowitz DB, 98, 200 Gessert J, 96 Ghadiri R, 18

Author Index Gharib M, 136, 137, 138 Ghattas O, 112 Gillenwater K, 478 Gilmore L, 547 Gilon D, 301 Gimbrone MA, 113, 115, 133,225,232 Giordano RA, 273 Giorgio TD, 269 Girsch W, 505 Girton TS, 266 Giszter S, 509 Glagov S, 170 Gleason EF, 406 Gleason KK, 406 Glick SJ, 321 Glickson JD, 349 Glimcher MJ, 314 Glismann CC, 267 Godleski J, 34 Goeckeler Z, 162 Goetz DJ, 232 Goldberg DP, 244 Goldberger AL, 451 Goldfarb M, 506 Goldring SR, 476 Goldstein SA, 482 Goldstick TK, 155 Golub AS, 153 Gooch KJ, 197 Goodell M, 10A Gore RW, 159 Goresky CA, 356 Gottlieb GL, 514 Grabec D, 268 Grabiner MD, 519 Granat MH, 561 Gratzl M, 449 Gray ML, 305 Greenberg NL, 101 Greenwald SD, 450 Gregory CD, 304 Grelsamer RP, 498 Griffith B, 108 Grimshaw P, 502 Grodzinsky A J, 491,502 Grood E, 292 Groop L, 338 Grossman S, 23 Grotberg JB, 30, 35A, 63, 64 Gu W, 490 Guan J, 186 Gudi SR, 264 Guglielminotti P, 525 Guiseppi-Elie A, 433

Gulani V, 304 Guo E, 482 Guo G, 136, 137, 138, 140 Gupta BS, 296 Gutowska A, 290 Guyton DL, 456 Guzman SJ, 385 Habib RH, 29 Hajduszok G, 228 Hakimian S, 340 Hall CL, 92 Halpern D, 27, 30, 63 Halter JA, 382 Haman JD, 474 Hambrecht T, 0, 375 Hammer DA, 186, 232 Hancock KE, 399, 414 Handschumacher MD, 301 Hanson SR, 170 Hansson T, 471 Hantos Z, 43, 45 Harasaki H, 193 Harrigan TP, 164, 198,467, 468 Harris TR, 49, 50 Hart R, 485 Harty H, 108 Haselton DJ, 78 Haselton FR, 269 Hashemi RM, 550 Hattler BG, 81 Hausdorff JM, 451 Hauser DW, 333 Hawkins SS, 270 Haworth ST, 68 Haxhiu MA, 366 Hayden M, 45 Hayes WC, 0 Hayoz D, 135 H e B , 112 He J, 521 Healy KE, 219 Healy TM, 99 Hederick E, 129 Hedman GE, 0 Heegaard JH, 496 Heigenhauser GJ, 351 Heller A, 367 Hellums J, 152 Helmlinger G, 130, 271 Hemami H, 549 Henning G, 249 Henry FS, 34, 178 Henry J, 360

S- 125

S- 126 Herderick EE, 332 Hem D, 235 Hetke JF, 402 Hezar R, 308 Higgins WT, 82 Hill CH, 132A Hill M, 44 Hipp JA, 0 Hirahara A, 24 Hlastala MP, 32 Hobbs SK, 272 Hockberger PE, 219 Hoffman E, 85 Hoffman NG, 9B Hoffman S, 132A Hogan N, 516 Holden MK, 541 Holle J, 505 Holmes MH, 497 Holmes T, 309 Holt KG, 540 Hong C, 104 Hopfer U, 449 Hopper DL, 128 Horton AF, 267 Houdayer TP, 510 Houlind K, 132 Houtz P, 46 Howorka K, 505 Hradek GT, 377 Hrousis CA, 26 Hsiao SS, 409 Hsu HA, 315,358 Hsu H, 257, 259, 275,276, 288 Huang T, 431 Hubbard AE, 368,413 Hubbell JA, 235,250 Hubmayr R, 39, 44 Hudson DL, 443,444 Hudson SM, 296 Hughson RL, 55,434 Hung GK, 395 Hunter DG, 456 Hunter WC, 0 Hutchinson CL, 430 Huxley VH, 142, 161 Hwang NH, 100, 144, 146, 267 Hwang Y, 199 Iatridis J, 487 Ibrahim SF, 141 Ijspeert AJ, 163 Ingber DE, 36, 38, 188

Author Index Ingenito EP, 65 Isakov E, 508,551 Ishuang SL, 218 Izquierdo M, 328 Jackson A, 29 Jahoor F, 360 Jain RK, 21, 83, 125, 130, 166, 167, 168, 254, James K, 405 James SD, 362 Jaron D, 150A Jarrett PK, 106 Jassawalla JS, 88 Jeffries GA, 452 Jen M, 502 Jensen OE, 63 Jette M, 364 Jiang A, 314 Jiang H, 314 Jiang H, 453 Jiang L, 301 Johnson DL, 223 Johnson IM, 260 Johnson K, 409 Johnson M, 65 Johnson M, 412 Johnson P, 104 Jolly CN, 372 Jones GS, 376,403 Jones JL, 13 Jones KA, 37 Jones M, 298 Jones RB, 376, 403 Jung U, 230 Juusola M, 418 Kalhan SC, 352 Kalia K, 104 Kameneva MV, 111, 215 Kamm R, 25, 26, 65, 71, 150, 179 Kan YW, 0 Kandarian S, 526 Kantor H, 313 Kantor S, 431 Kargo W, 509 Karl W, 141 Kashyap R, 449 Kaufman JW, 79 Kay AR, 379 Kay BK, 9B Ke H, 176 Kearney RE, 419 Keilson SE, 384 Kelkar R, 463 Kelleher JK, 348

Author Index Kelley NC, 200 Kelly SM, 41 Kenner T, 180 Kepler GM, 31 Kerrigan C, 539, 542 Keynton RS, 174 Khachigian L, 115 Khalsa PS, 501 Khoo MC, 56 Kiani MF, 272A Kim BS, 464 Kim HW, 6 Kim S, 78 Kim S, 248, 290 Kim T, 283 Kim W, 132 Kimbell JS, 31 Kinahan PE, 322 King MA, 321 King RB, 317 Kingsbury C, 140 Kinzie DJ, 357 Klarman ML, 234 Klein SS, 330 Klinger MM, 477 Kmiecik JA, 304 Knapp CF, 84 Knudson O, 298 Koegler WS, 273 Koenig GC, 274 Konstantopoulos K, 229 Kopchick J, 22 Korenberg M J, 418 Koris M, 417 Kormos R, 108 Kovacs G, 405 Kralj-Iglic V, 268 Krenz GS, 47, 66 Krewson CK, 234 Krishnan P, 494 Krohn KA, 317, 331 Kroll K, 317, 331, 335, 336, 357 Kruse R, 504 Ku DN, 170 Kuban BD, 171 Kuhl PR, 236 Kukreti S, 229, 253 Kundich RC, 80 Kung RT, 90 Kunkel E, 230 Kupa E, 425,525, 526 Kushmerick MJ, 524 Kwak SD, 498

Kwon J, 298 Ky B, 275 Labadie K, 104 Laden Z, 451 LaForge DH, 88 Lai W, 490 Lai-Fook SJ, 46 Lalush DS, 324 Landstrom AE, 259, 275 Lange DN, 128 Langer RS, 245,260, 277, 283, 291 Langille BL, 214 Lanmiiller H, 505 Lapedes A, 9A Lapidot A, 358A Laplante A, 350, 365 Laraia P, 539 Larson BC, 415 Laruelle M, 319 Lauffenburger D, 238 Lauterbur PC, 304, 306 Leake PA, 377 Lee AA, 184 Lee ET, 201 Lee JS, 72 Lee JM, 202 Lee J, 88 Lee LP, 72 Lee R, 179 Lees RS, 141 Lehtovirta M, 338 Lei P, 14 Lemay MA, 507 Lemon JC, 244 Lemons DE, 217 Lemons JE, 475 Leong J, 184 Leslie JH, 544 Less J, 0 Leung DY, 101 Leunig MK, 255 Levin DN, 307 Levin O, 508,551 Levine RA, 301 Levy DJ, 136, 137, 138 Lewis J, 465 Ley KF, 230 Li JK, 126, 149, 182, 183, 190, 192, 212 Li J, 314 Li Z, 317, 331, 336 Liang Z, 304, 306 Liaw J, 391

S- 127

S- 128 Lichtenbeld HC, 271 Light WR, 152 Lim K, 145 Lim LK, 446 Limb S, 406 Lin H, 233 Linberg R, 5 Linderrnan J, 239 Linehan JH, 47, 48, 60, 66, 68, 77 Link J, 331 Linsenmeier RA, 155 Litwak P, 108 Lo Conte L, 525 London AP, 296 Loree, II HM, 179 Lorenz CH, 330 Loughlin PJ, 522 Louie LK, 276 Lovell JR, 399 Lovett EG, 203,522A Lucas LC, 474, 477 Lukatela K, 410 Lund JL, 401 Lund LW, 81 Luscinskas FW, 232 Lutchen KR, 43,421,460 Lux RL, 121 Lyman MD, 106 Lynn G, 504 Ma PX, 277 Macia NF, 82 MacKinnon R, 0 Magid D, 333 Makaroun MS, 143 Maksym GN, 42 Malloy C, 0 Maloof P, 547 Mann JR, 406, 415 Mannion AF, 536 Mari A, 337, 343 Markham J, 47 Markou CP, 170 Markus EJ, 407 Markwald RR, 132A Marmarelis VZ, 393,394, 396, 420, 423 Marras WS, 0 Martin BO, 326 Martyniuk J, 372 Marzian S, 204 Massia S, 105 Matsumoto T, 116 Maxian TA, 0, 493 May-Newman K, 131

Author Index Maymir JC, 98 McCauley P, 309 McCulloch AD, 184 McCullough WT, 151 McDonald I, 10 McDowell CL, 488 McEver RP, 231 McGill SM, 534 McGuire, Jr. HH, 488 Mclntire LV, 40, 229, 253, 285 McKay CB, 152 McKeown PJ, 219 McKinsey J, 114 McLafferty FW, 15 McLeod K, 469, 471 McMahon TA, 0 McNaughton BL, 407 McNeal DR, 545 Mears DJ, 242, 433 Meford RM, 227 Mehta BV, 22,442,447 Meister J, 135, 147, 163, 169, 205, 208 Melanson D, 106 Melder RJ, 254, 271, 274, 278 Melot CA, 67 Melvin DB, 110 Menapace C, 387 Mente PL, 465 Menter DG, 285 Merfeld DM, 425,428,520 Merker MP, 48, 77 Merletti R, 525 Meyer RS, 98 Meyer RA, 347 Michel RP, 41 Michele JJ, 196 Mikos AG, 218, 244 Milam D, 454 Milet SF, 132 Miller CA, 559 Miller C, 20 Miller M J, 218, 244 Miller P, 88 Milo S, 136, 137, 138 Mironov VA, 132A Miserocchi G, 86, 329 Mitzner W, 29, 57 Mizrahi J, 508, 551 Montes RA, 294 Mooney DJ, 245 Mooradian DL, 266 Moore JE, 170, 173 Moore KL, 231 Morgan KT, 31

Author Index Moriarty JA, 30, 64 Morris ED, 301,315 Mortlock A, 510 Moser M, 180 Mountain DC, 369, 413 Mow VC, 462, 487,490, 497,498 Mowery J, 251 Mueller K, 332 Muldaerjee DP, 503,504 Mukherjee N, 494 Mulavara AP, 552, 559, 560 Mulder E, 445 Mulier JP, 180 Mulligan RC, 10A Muluk SC, 143, 176 Mundel T, 118 Munn LL, 274, 278 Murray CD, 553 Mussivand TV, 87 Myers ER, 492 Myklebust BM, 522A Myklebust JB, 0, 203,522A Naegeli A, 88 Naeije R, 67 Nagel TE, 115, 133 Naim KL, 150A Naimark WA, 202 Nair HH, 13 Nair MS, 134 Nakajima HH, 413 Nau WH, 454 Neer RM, 314 Neff J, 35 Negrini D, 73, 86 Nelson L, 458 Nemcek T, 431 Nemerson Y, 92 Nerem RM, 206, 227, 295 Netti PA, 83, 167, 168, 271 Neufeld G, 35 Neunlist M, 123 Newman DJ, 426 Newman VS, 173 Nguyen TT, 340 Nicolellis M, 408 Nicolson GL, 285 Niemi PP, 313 Nikolovski J, 245 Nitowski KN, 201 Nollert M, 231 Noordergraaf A, 149, 180, 181 Nordhausen CT, 374 Noflie J, 431

Norman KE, 230 Normann RA, 0, 374,405 Norregaard TV, 275, 288 Norwich KH, 334, 390 Nozue M, 125 Nugent M, 272 O'Connell M, 554 O'Hara DA, 149 O'Leary S, 207 O'Malley M, 511 Oddsson L, 527,547 Odgaard A, 468 Ohashi T, 116 Okabe K, 34 Okano T, 248 Olivier LA, 240 Olsen BR, 280 Olson E, 413 Olson LE, 48 Oman CM, 427 Orsorio ME, 555 Osterberg UL, 453 Ott DE, 2, 455 Otto S, 387 Ounpuu S, 543 Ozawa ET, 150 Paganini AT, 347 Page TC, 152 Paithankar DY, 430, 458 Palfreyman M, 10 Palladino JL, 180, 181 Palsson B, 354 Pan J, 431 Pan T, 321 Pandian NG, 300, 302 Panettieri R, 39 Pang KS, 356 Pantalos GM, 207 Parfitt M, 481 Park A, 237 Park W, 85 Parnianpour M, 535, 549 Partridge LD, 3 Paschalis E, 472 Pascual R, 269 Patel KD, 231 Patel SN, 456 Patil MM, 366 Patrick J, 510 Patrick, Jr. CW, 253 Patterson B, 359 Patterson M, 339

S- 129

S- 130 Patwardhan AR, 84 Paul JP, 561 Paulsen KD, 453 Paydarfar D, 53 Peattie RA, 172 Pedersen EM, 132 Peng CK, 451 Penn RD, 517 Percival DB, 435 Pereira CA, 202 Perepelkin PD, 400 Perez CA, 555 Perron CY, 398 Perry J, 557 Perun ML, 27 Pet~k, 43, 45 Peterson DR, 263 Pfann KD, 517 Pfleiderer B, 479 Pia Saccomani M, 342 Pierce WS, 109 Pilla J, 182 Pincus S, 438 Pishko MV, 367 Pitsenberger C, 14 Pittman RN, 153 Planus E, 39 Pliura D, 0 Polacek D, 114 Poncelet BP, 313 Poon C, 52, 457 Pope M, 471 Popel AS, 154, 287 Popovic D, 513 Porret C, 135, 169 Portner PM, 88 Posner MI, 325 Pou NA, 49 Powell KA, 299 Powers C, 557 Powers M J, 279 Prakash YS, 37 Prandi JL, 501 Prasad PV, 312 Presson, Jr. RG, 70 Prieto TE, 522A Prince CW, 477 Prisk GK, 429 Prulamp G, 505 Puleo DA, 221,478 Puppala K, 146 Purdon PL, 451 Putnam CT, 351 Pythoud F, 147, 205,208

Author Index Qammar H, 552 Quick CM, 149 Quin Y, 407 Rabelo LC, 447 Raboudi S, 37 Radziuk JM, 0 Raghavan ML, 143 Rahemtulla F, 477 Rajagopal KR, 215 Ramanathan C, 314 Ramappa AJ, 257, 280 Ramasamy N, 88 Rambod E, 136, 137, 138 Ranjan V, 226 Rao SS, 557 Raphael RM, 281 Raymond GM, 336, 435 Reddy CC, 238 Redfem M, 522 Redling JD, 139 Reeds PJ, 360 Reeves SJ, 308 Rehkopf PG, 201 Reichert WM, 240 Reilly R, 553 Reinhold BB, 16 Reinhold V, 16 Resnick N, 115, 133, 225 Resnik C, 333 Restivo M, 120 Reuben JD, 466, 467, 468,499 Rezania A, 219 Rhee W, 107 Rhodes C, 317 Richards VM, 384 Richardson KE, 187 Richardson PD, 282 Rickaby DA, 68 Rider JE, 9B Riehle TJ, 172 Riley L, 333 Ringer GW, 486 Rintoul T, 108 Rippley RK, 241 Ritter C, 144 Rittgers SE, 134, 174, 256 Riva CE, 191 Rizzo JF, 389 Rizzo RS, 417 Robblee L, 0, 376, 403 Roberts R, 328 Robicsek F, 209, 210 Rodarte JR, 40

Author Index Rodrigues A, 425 Rodway N, 174 Rogers N, 452 Rohr S, 119 Rojas E, 242 Romero CM, 140 Roselli RJ, 49,454 Rosen M, 0, 528, 532 Rosenberg G, 109 Roth R, 542 Rousche PJ, 374 Rouse G, 328 Routh WD, 173 Rowley J, 245 Roy P, 22 Roy SH, 425,525,526, 527, 547 Roy TK, 154 Rubin C, 469, 471 Rubinstein JT, 381 Rudolph A, 9 Rumshitzki DS, 145, 199 Sachs MB, 383 Sacks MS, 102 Sadisivan K, 504 Saglam MA, 393 Sah RL, 489 Sahn DJ, 302 Saidel GM, 0, 193,352 Saint-Amand R, 512 Salama G, 120 Saltzman CL, 464A Saltzman WM, 23,234 Salzberg B, 119 Sandhu JS, 562 Santamore WP, 150A Sarelius IH, 97, 158 Sarge TW, 333 Satcher R, 117 Sato M, 116 Satterfield WC, 244 Savaria Y, 512 Sawan M, 512, 558 Sawhney A, 106 Scaringe WA, 394 Schaffler M, 481 Scharlack RS, 433A Scherer PW, 35, 79 Schmidt CE, 283 Schnetzer KA, 206 Schnitzer JE, 157 Schotland J, 515 Schuessler T, 61 Schuster D, 47

Schwab AJ, 356 Schwartz EA, 224 Schwartz SL, 300, 302 Schwirtlich L, 513 Sclabassi RJ, 396, 423 Scott W, 333 Seagrave RC, 76 Sears C, 301 Sedlatschek L, 5 Sefton MV, 103 Selent WP, 97 Selvaraj P, 297 Setton LA, 487 Severyn DA, 176 Sevick-Muraca E, 430, 458 Shafritz TA, 284 Shandas R, 298 Shannon R, 387 Shapiro RA, 143 Sharp MK, 207 Shastri VR, 283 Shea LD, 239 Sheft S, 385 Shephard MS, 497 Sheppard, Jr. NF, 242,433 Sheu BJ, 423 Shi P, 310 Shih JC, 132A Shimony JS, 304 Shin HY, 222 Shiota T, 302 Shirazi-Adl A, 535 Shofner W, 385 Shoham M, 551 Shoichet MS, 247 Shore SA, 37, 39 Shorr RG, 5 Shulman GI, 0 Shum K, 5 Sidossis L, 345 Sieck G, 37 Siemann DW, 272A Sigalovsky I, 414 Sill HW, 160 Silverman M, 440 Simaan M, 216 Simms RL, 174 Simon BN, 552, 559, 560 Simon BA, 58 Simpson DG, 185 Singh S, 369 Sinusas A, 310 Skalak R, 83 Skalak TC, 95

S- 131

S-132 Slack SM, 261 Sledge CB, 280 Slepian MJ, 105 Sly PD, 45 Smith C, 229 Smith DB, 102 Smith TW, 285 Smolinski PA, 286 Snyder AJ, 109, 200 Soares AM, 406, 415 Solheim J, 301 Somerville N, 545 Somogyi A, 13 Sonka M, 85 Sowyrda P, 0 Spangler N, 333 Sparacino G, 341 Sparks AB, 9B Spector AA, 287 Spector M, 257, 259, 275,276, 280, 288 Spelman FA, 372 Spilker MH, 288 Spilker RL, 497 Spodnick GJ, 296 Sporn LA 272A Spotnitz AJ, 139 Sprague EA, 251 Srinivas SV, 156 Srinivasan RS, 148 Stadler RW, 141 Staller S, 387 Stamenovic D, 38 Starcher B, 132A Staub, Sr. NC, 69 Steed DL, 143 Steffensmeier SJ, 464A Stephanazzi J, 67 Stepp DW, 335 Stergiopulos N, 135, 147, 205,208 Stine SS, 262 Stokes C, 241 Strasser JF, 23 Strobel JT, 301 Stroupe S, 431 Stubblefield HM, 464 Suh J, 461 Suki B, 43,421,440, 460 Sumpio BE, 225 Susak Z, 508, 551 Suzuki B, 202 Swadlow HA, 410 Swanson J, 372 Swartz MA, 166 Systrom DM, 0

Author Index Szarowski D, 309 Szeto HH, 51 Szymanski BK, 497 Tagawa H, 188 Tang D, 177 Tarbell JM, 98, 160 Tardy Y, 163 Taubman MB, 92 Taylor L, 108 Teich MC, 437 Teien D, 302 Tepavac D, 513 Terracio L, 185 Thipakorn B, 142 Thoma H, 505 Thomas CH, 219 Thomas D, 108 Thomas JD, 101,211, 299 Thompson CV, 406 Thompson DR, 196 Thompson GJ, 221 Thubrikar MJ, 209, 210 Tiedeman JS, 156 Tompkins RG, 358 Tonellato PJ, 435,436 Tonelli AE, 296 Townsend DW, 322 Traber DL, 363 Tranquillo RT, 243, 249, 266 Traore M, 449 Treco DA, 0 Trepagnier CY, 0, 529 Treppo S, 491 Troy TL, 430, 458 Truskey GA, 240 Tsai M, 289 Tsuda A, 34 Tsui BM, 324 Tung L, 123 Turitto VT, 92, 261 Turner CH, 482A Turner J, 309 Tuten R, 488 Ultman JS, 33 Unger E, 505 Uusitalo R, 418 Vacanti JP, 245,283 Valdes-Cruz L, 298 Valeri C, 7 Vallurupalli S, 84 van Bibber R, 335

Author Index van den Honert C, 373 Vanaria CM, 406 Vandervoort PM, 101,211, 299 Varma A, 354 Varner SE, 227 Vaslef SN, 155 Venegas JG, 28, 59 Venturoli D, 86, 329 Vera DR, 318 VerLee D, 431 Vermeulen F, 67 Vernon BL, 290 Verstraete MC, 552, 559, 560 Vicini P, 338 Villarreal FJ, 184 Vincent G, 364, 365 Vogt JA, 358 Voie AH, 372 Voigt HF, 392, 399,414 yon Recum HA, 248 Vorp DA, 143, 176 Votaw JR, 320 Vunjak-Novakovic G, 258, 265, 291 Wagner, Jr. WW, 70 Waldman SD, 202 Walker PG, 132 Walker R, 121 Wallace M, 81 Walsh N, 506 Walsworth F, 431 Waiters F, 81 Wang H, 292 Wang J, 535 Wang J, 183, 212 Wang J, 314 Wang N, 36, 38, 39, 188 Wang PM, 46 Wang SK, 267 Wang WC, 293 Wang WC, 261 Wang XL, 554 Ward J, 452 Ward KA, 186 Warnicki JW, 201 Wasserman DH, 344 Waters CM, 213 Waugh RE, 281,289, 293 Wayne JS, 486, 488, 494 Webster MW, 143, 176 Weckstr6m M, 418 Wei JY, 451 Weidenbaum M, 487 Weiland JD, 404

Weinbaum S, 145, 194, 199, 217,470 Weinstein SL, 0, 493 Weiss WJ, 109 Weisskoff RM, 313 Welkowitz W, 139 Wells A, 238 Wells SM, 214 Wenz J, 333 West JL, 250 West JB, 429 Westerhof N, 208 Westgaard RH, 533 Westwick DT, 419 Weyman AE, 301 Whalen RL, 91 White JA, 379 White ML, 247 Whitehurst T, 405 Whitingham TS, 366 Wick TM, 286, 294 Wiegner A, 506 Wielopolski PA, 312 Wiesner TF, 295 Wiggs B, 26, 71 Wilkins ES, 459 Williams DA, 161 Williams III TW, 0, 531 Wilson D, 545 Wilson D, 328 Wilson S, 492 Wilson TA, 44 Winn SR, 247 Wise KD, 401,402 Wolfe RR, 345, 359, 363 Won J, 516 Wong CB, 466 Wong S, 431 Wong W, 390 Woodruff S, 207 Woods SA, 29 Woolley CB, 548 Worsey MJ, 104 Wright DK, 452 Wrighton MS, 406 Wroblewski D, 62 Wu BM, 24 Wu CH, 11 Wu F, 180 Wu GY, 11 Wu J, 155 Wu J, 93 Wu Y, 314 Wu Z, 144 Wyatt J, 389

S- 133

S-134 Wylie KB, 296 Wyman BT, 384 Wysocki VH, 13 Wysolmerski R, 162 Xia J, 122 Xiao YA, 240 Xiao Z, 226 Xu M, 323 Xu S, 304 Xue S, 373 Yamaguchi R, 179 Yamamoto Y, 55,434 Yang C, 79 Yang L, 561 Yang Y, 304 Yang Y, 561 Yannas IV, 257,259, 275,276, 280, 284, 288 Yao L, 307 Yarmush DM, 358 Yasko AW, 218, 244 Yaszemski MJ, 218 YeS, 16 Yeleswarapu KK, 215 Yen J, 240 Yeo EL, 103 Yin FC, 131 Yin Y, 145 Yipintsoi T, 331 Yoganathan AP, 99, 132

Author Index Young C, 562 Young ED, 384 Young LR, 425 Young VR, 358 Yu AA, 146 Yu YM, 358 Yu Y, 216 Yuan F, 130, 255 Yuan H, 421 Yun Z, 285

Zale S, 0 Zardoshti M, 550 Zatorre RJ, 327 Zeks B, 268 Zervos GP, 313 Zhang D, 470 Zhang Q, 421,460 Zhang X, 363 Zhang Y, 11 Zheng LN, 153 Zhou J, 211 Zhu C, 297 Zhu H, 21,254 Zhu H, 395,416 Zhu LC, 217 Zhu Y, 183 Zile MR, 187 Zuelzer WA, 486 Zupan B, 382