f

Review

Factors affecting electrode-gel-skin interface impedance in electrical impedance tomography E. T. M c A d a m s I

J. Jossinet 2

A. Lackermeier 1

F. Risacher z

1Northern Ireland Bioengineering Certre, University of Ulster at Jordanstown, Co. Antrim, BT37 OQB, UK 21NSERM Unit 281, 151 Cours Albert Thomas, 69424 Lyon, Cedex 03, France

magnitude, mismatch and temporal variations of the electrode-gel-skin interface impedance can cause problems in electrical impedance tomography (E/T) measurement. It is shown that at the high frequencies generally encountered in EIT the capacitive properties of the electrode interface, and especially those of the skin, are of pr/mary importance. A wide range of techniques are reviewed that could possibly be used to minimise these problems. These techniques include the use of skin preparation, penetration enhancers, temperature and electrical impulses. Although several of these techniques appear very attractive, they are not without serious potential drawbacks. A combination of some of these techniques may well hold the key to success. Abstract--The

Keywords~EIT, Electrodes, Electro-Osmosis, Electroporation, Penetration Enhancers, Skin Preparation, Temperature, Med. & Biol. Eng. & Comput., 1996, 34, 397-408

J ! Importance of electrode-gel-skin contact

~LECTRICALimpedance tomography (EIT) is the production of mages of the distribution of electrical impedance, or more :ommonly resistivity, in a two-dimensional slice through a ~ody segment by means of impedance measurements made on :he surface of the body part under study. The impedance is ?rooed with a series of two- or four-electrode measurements asing an array of electrodes attached to the skin and encircling :he area of interest. A reconstruction algorithm then computes -he electrical conductivity or permittivity distribution within :he body slice using the applied current waveforms and the aaeasured voltages. In a two-electrode measurement, the same electrode pair is ased to applied the current and to measure the resultant voltage. In a four-electrode measurement, current is applied ay a pair of electrodes and the voltage is measured using a .different pair of electrodes located elsewhere on the body mrface. A four-electrode technique is used in most EIT ~ystems. For example, in the Sheffield Mark 1 EIT system, ~r is the most often used for clinical trials, current is applied sequentially to the body segment using a pair of tdjacent electrodes (BARBER et al., 1983). While current is acing applied to a given pair of electrodes, the voltages aetween the adjacent pairs of the remaining non-current "arrying electrodes are measured. This process is repeated for each pair of adjacent electrodes, and a complete set of voltage data is thus obtained.

Zirst received 22 June ~995and in final form 20 August 1996 IFMBE:1996 Vledical & Biological Engineering & Computing

In order to achieve sufficiently high spatial resolution in EIT, a relatively large number of small electrodes are needed to surround the body segment under investigation (TAKTAK et aL, 1995a). The Sheffield system, for example, is based on the use of 16 electrodes (BARBER et aL, 1983), and these allow 104 independent measurements to be obtained for each data set. Systems using 32, 64 and even more electrodes have been used in an effort to fixrther improve the spatial resolution of EIT (FUKS et al., 1991; GOBLE and IAACSON, 1989). The accurate measurement of underlying tissue impedance is often rendered difficult by the presence of sizeable electrode 'contact' impedances. (Contact impedance is used here to describe the total electrode-gel-skin impedance.) Contact impedances are dominated by the impedance of the outer layers of the skin and often have characteristic frequencies around 10 Hz (MCADAMS and JosSrNET, 1991a). They therefore tend to dominate measured impedances at low frequencies. Although contact impedance decreases with increasing frequency, it can still have a value of several hundred Q at frequencies as high as 100 kHz (ROSELL et al., 1988). Tissue impedance, on the other hand, is generally observable between a few kHz to tens of MHz. Initially, EIT systems used only one frequency in the tens of kHz range. For example the Sheffield Mark I EIT system uses a frequency of 50 kHz. More recently, images have been produced by comparing measurements made at two frequencies, one of which was of the order of a few hundred kHz (JosSINET and TRILLAUD, 1992). Efforts are underway to develop multi-frequency systems whose frequency ranges encompass as much as possible of the band of frequencies that characterise the tissues under study (kHz and MHz) (BROWN et al., 1994). In this way, "it is hoped that the advantages of EIT can be combined with those of bio-

November 1996

397

electrical spectroscopy, and thus increase the effectiveness and versatility of LIT systems. It can therefore be seen that the electrode contact impedance can adversely affect the accuracy of existing mono-frequency LIT systems which are operated in the tens of kHz range and the accuracy of the lower end of the frequency range of the new multi-frequency systems. Theoretically at least, it would be highly desirable to completely surround the body part under study with many closely spaced small electrodes. Decreasing electrode size will, however, increase the electrode-gel-skin impedances and thus can cause problems during LIT measurement. Large contact impedances can give rise to large potential drops at the electrode-skin interface, which may mask voltage changes within the underlying tissues. The presence of reactive contact impedances generates phase shifts which can cause misleading measurements (TAKTAKe t al., 1995b). The mismatch of contact impedances can be responsible for significant errors that result in common mode voltages in impedance imaging systems (SMITH, 1990). Gersing et aL, have shown that even impedance mismatches due to differences in the lengths of current pathways in an electrolyte tank can give rise to the appearance of positive phase angles (GERSINGe t al., 1995) Contact impedance can also be a source of noise (GONDRAN et al., 1995) and drift (BAISCHet al., 1995), and it has been shown that these may cause artefacts in LIT measurements (BOONEand HOLDER,1995a;b). In the two-electrode technique, the contact impedances of both electrodes are measured along with the desired tissue impedance. The two-electrode technique is therefore very susceptible to problems due to the magnitude, mismatch or drift of contact impedances. In a four-electrode technique, the voltmeters do not measure electrode contact impedances, and hence should be insensitive to problems arising from electrode contact. Although the four-electrode technique is less sensitive to contact impedance problems, problems still do arise. Riu et al.; showed theoretically that large mismatches in the resistive and/or capacitive properties of the voltage detection electrodes, in particular, can result in errors in both the real and reactive parts of the measured impedance (RIu et aL, 1995). The contact impedances in the current injection circuit develop a common-mode voltage on the body, while those in the voltage measuring circuit reduce its common-mode rejection ratio (CMRR) and thus prevent the elimination of the common-mode voltage (BOONE and HOLDER, 1995b). It is interesting to note that, as the contact impedances form part of a larger overall circuit which is complicated by the presence of stray capitances etc., the effects due to contact impedance

lead

I

electrode

]

ZC,~~ mA(E)

RCT

Re,,

epidermis

dermis

ZcpAm ~

R~,

Fig. 1 Simple equivalent circuit model of the electrode-gel-sMn

interface 398

mismatch are not solely confined to the low-frequency range. Mismatch of the series-resistive component of the contact impedance, for example, was found to reduce the CMRR at higher frequencies (around 1 MHz) (RIu et al., 1995). It is therefore important to gain an understanding of the factors influencing the contact impedance so that its undesirable contributions in LIT can be reduced. The effects on the contact impedance of electrode metal, electrolyte composition and concentration, skin site, skin abrasion, penetration enhancers, temperature and applied signal amplitude are briefly reviewed here. Some of these effects are well known and exploited in biosignal monitoring. Other techniques for decreasing contact impedance are not widely used, and some appear to be interesting suggestions which need a considerable degree of development before they can, if ever, be clinically exploited during LIT measurements. It is hoped that this work will stimulate interest in this important area of research. Important topics related to the subject but judged to be beyond the scope of this paper include the current density distribution under current-carrrying electrodes, and the effects of the geometries and relative positioning of the currentinjecting and the voltage-detecting electrodes (YORKEY et al., 1985; PAULSON et al. 1992; WANG, 1994).

2 'Contact' impedance The impedance loci of the electrode-electrolyte interface, the epidermal layer of the skin and the underlying tissues have each been found to be characterised by 'depressed' semicircular arcs (MCADAMS and JOSSlNET, 1995). Depending on the relative magnitude of the impedances, all three arcs may be discernible on a complex impedance plot when a sufficiently wide frequency range is used. The electrode-electrolyte impedance arc dominates at low frequencies ( f < 1 Hz), the impedance of the skin dominates at 'mid' frequencies (approx. 1 H z < f < 10 kHz) and the impedance of the underlying tissues dominates at high frequencies (approx. 100 kHz < f < 10 MHz). The total electrode-gel-skin interface can be simplistically represented as shown in Fig. 1. The electrode-gel interface is represented by an empirical 'constant phase angle' impedance ZcPACE): in parallel with a resistance RcT (MCADAMS, 1989a;b), where

ZcpA(e) = XE(Jco) ~

(1)

co is the angular velocity equal to 2nf f i s frequency, KE is the magnitude of ZceAm) at co = 1 and fl is a constant such that 0<3<1. ZcPAm) is thought to represent the non-faradaic properties of the interface, generally attributed to the presence of the double-layer capacitance distorted by specific adsorption and surface roughnes effects (MCADAMS, 1989a). Rcr, the charge transfer resistance, represents the faradaic reactions taking place at the interface. The impedance of the electrode gel pad can be generally represented by a small resistance Rgel (MCADAMS, 1990; MCADAMS and JOSSINET, 1991a). The stratum comeum is generally 10-15 gm thick. On some body areas it can, however, be several hundred #m thick. Thickness will vary with the number of cell layers making up the stratum corneum and the state of hydration. The stratum comeum is composed of approximately 40% protein (large keratin), 40% water and 20% lipid (KELLY, 1985). The stratum comeum consists of many layers of compacted, flattened, non-nucleated dehydrated cells (called comeocytes) which are filled with cross-linked keratin. The intercellular spaces between comeocytes are occupied by arrays of bilami-

Medical & Biological Engineering & Computing

November 1996

nar membranes with the morphological features of polar lipids (PRAUSNITZ et al., 1993). These lipid bilayers are composed primarily of ceramides, cholesterol and fatty acids. This matrix appears to serve to bind the cells to form a fabric. The stratum comeum has therefore been described in terms of comeocyte 'bricks' surrounded by lipid 'mortar' (BROWN and LANGER, 1988). The underlying layers of the epidermis are, in contrast, a relatively aqueous environment. The transition from an essentially non-conductive lipophilic membrane (the stratum corneum) to an aqueous tissue (viable epidermis and dermis) gives rise to the skin's barrier properties. A lipophilic molecule may be able to penetrate the stratum corneum lipids but cannot partition into the aqueous environment of the epidermis. On the other hand, a hydrophilic molecule may never be able to partition from the skin surface into the stratum comeum lipids. Molecules that most easily permeate through the skin are those with both appreciable lipid and water solubilities. As the stratum comeum is relatively non-conductive, it presents a high impedance to the transmission of electrical currents. However, due to its delectric properties and its thinness, it permits capacitive coupling between a metal electrode placed on the skin surface and the underlying conductive tissues. The 'capacitive' properties of the straum comeum can be represented by an empiracal 'constant phase angle' impedance ZcpA(s), and these are closely associated with the lipid-protein matrix: Zcpa(s) = Ks(joo) -~

(2)

where Ks is the magnitude of ZcPa(s) at ~o= 1 and c~ is a constant such that 0 < c~< 1. Some ions manage to traverse the stratum comeum via paracellular pathways and through the skin's appendages (hair follicles, sweat ducts, sebaceous glands, imperfections in the integrity of the skin) (CULLANDER and GUY, 1992). An average human skin surface is believed to contain between 40 and 70 hair follicles and 200 to 250 sweat ducts on every square centimetre (CHIEN, 1987). The diameter of the sweat ducts ranges from 5 to 20 #m. As skin appendages extend through the stratum comeum barrier, they can act as shunts to the interior. Although water-soluble substances can penetrate into the skin via these appendages at a faster rate than through the 'intact' stratum comeum, as these appendages occupy only approximately 0.1% of the skin surface they are regarded as relatively unimportant routes for the passive diffusion ions through the skin (CHIEN,1987; CULLANDER,1992). (In iontophoretic transdermal delivery of ionised drugs, however, the DC appears to seek out paths of least resistance and a significant portion of the current flows through the skin's appendages.) The flow of ionic current can be represented electrically by a large resistance Re in parallel with Zcencs). Given that the frequencies of interest in EIT are relatively very high, it is the series resistances and the parallel capacitive impedances in the overall contact impedance that are of primary interest here. Unfommately, much of the information published in the literature has been measured over frequency ranges which are not wholly appropriate to EIT, and meaningful deductions have to be carefully made. As the skin impedance tends to be much greater than that of the electrode-electrolyte interface, the effects of certain of the parameters are only studied on skin impedance and their effects on the electrode-electrolyte interface ignored.

2.1 Effect o f electrode material and surface topography The impedance of the skin is generally the largest component in the overall 'contact' impedance (MCADAMS and Medical & Biological Engineering & Computing

JOSSINET, 1991a). The impedance of the electrode-gel (or electrode-electrolyte) interface can, however, be significant (depending on the electrode system involved, the frequency range of interest etc.) and obviously can play an important role when electrodes are inserted directly into biological tissues/ fluids or are used in electrolyte tanks (GERSING, 1995; ZHOU et al., 1994; R[GAUD et al. 1995). The first biomedical electrodes were made from readily available, low-cost easily machined metals and alloys. Stainless steel is still often used as an electrode material in reusable systems as it is resistant to corrosion. With the need for smaller electrodes in more demanding biosignal recording applications, high electrical performance has become of major importance. The electrical properties generally considered desirable for an electrode-electrolyte interface, especially for those used for biosignal recording, include (i) low, stable offset potentials. (ii) low, matched interface impedances. (iii) low polarisation (MCADAMS, 1990; MCADAMS et al., 1992a). Criteria (i) and (iii) depend solely or principally on the interfacial electrochemistry, i.e. on the metal, the electrolyte and the reactions taking place at their interface. Criterion (ii), which concerns the interface impedance, depends not only on the interfacial electrochemistry but also, to a considerable extent, on the surface topography of the electrode (MCADAMS 1989a). An electrode system that meets all of the above criteria, and which as a result is often used in many biosignal recording applications, is the silver-silver chloride (Ag-AgC1) electrode. The potential of this electrode is determined by the activity (related to concentration) of the chloride i0n,and hence the electrode potential is quite stable (as well as small) when the electrode is placed in contact with an electrolyte containing C1- as the principal anion (these electrolytes include electrode gels, tissue fluids and the electrolytes commonly used in in vitro experiments) (MCADAMS et al., 1992b). Ag-AgCI electrodes tend to have low interface impedances due, at least in part, to the surface topography of the deposited layer (MCADAMSet al., 1992b SHOU et al. 1994). Simplistically, the pores on a rough-surface electrode can be thought of as behaving somewhat like transmission lines (DELEVlE, 1965) and, as a result, the interface impedance of a rough interface is proportional to the square root of that of a smooth interface. Based on this simple theoretical model, it can be seen that the magnitude of the rough interface impedance can be much smaller than that of a smooth one. The decrease in impedance is due to more than a simple increase in effective surface area. In cases of extreme porosity, the interface phase angle at high frequency can decrease from around 90 ~ (for an ideally smooth surface whose impedance is dominated by the double layer capacitance) towards a limiting value of 45 ~ When an initially smooth-surface Ag electrode is chlorided, the AgCI deposit can give rise to a rough surface and thus to a small interface impedance (MCADAMS et al., 1992b). A similar metal-metal salt electrode system sometimes used is tin-tin chloride. Electrode systems such as silver-silver chloride are often (wrongly) described in the literature as 'non-polarisable' electrodes. Polarisation has been defined as 'the departure of the electrode potential from the reversible (or equilibrium) value upon the passage of faradaic (or dc) current' (BARD and FAULKNER, 1980). Expressed very simply, when a DC iac flows through an electrode-electrolyte interface, it flows

* Lec Tec Corp. US Patent 4,674,512, 23 June 1987

November 1996

399

through the faradaic charge transfer resistance Rcr, and thus gives rise to an additional voltage (termed 'over-potential' in electrochemistry) equal to the product iacRcr. As a result, the electrode potential deviates from its equilibrium (or 'reversible') value and the electrode system is said to be polarised. A perfectly polarisable electrode would not permit the flow of any DC (or 'faradaic' current) as the charge transfer resistance in this case would be infinite. An ideal non-polarisable electrode would have no resistance to faradaic current and charge could thus be transferred across the interface with ease and without any voltage being dropped across the interface. Real biomedical electrode systems are however, all more or less polarisable with varying values of charge transfer resistance. Ag-AgC1 electrodes have relatively low values of charge transfer resistance and are thus relatively non-polarisable, operating close to their equilibrium or reversible potentials. Electrodes that come closest to behaving as perfectly polarisable electrodes are those made of noble metals (MCADAMS et al., 1992a). As these metals are considered to be inert, they tend not to react chemically with the surrounding electrolyte or tissue. The charge transfer resistances for such electrodes tend therefore to be relatively large and the electrodes are highly polarisable. As there is no simple means by which charge can be transferred across the interface, the noble electrode assumes an indeterminate potential with respect to the electrolyte which can drift randomly. Noble metals are therefore generally not now used in the construction of biosignal monitoring electrodes. It would appear widely believed that the requirements for an EIT electrode system are as demanding (if not more so) as those for biosignal electrodes, depending on the application. This may not be the case. Given that EIT systems are generally capacitively coupled to the electrode-patient interface, electrode offset potentials and polarisations do not often give cause for concern. It would therefore appear that it is the magnitude and stability of the interface impedance which are of primary importance in EIT. Furthermore, as EIT generally involves measurements at relatively very high frequencies, it is only the magnitude of the non-faradaic component of the electrode-electrolyte interface impedance that is of relevance. The high-frequency interface impedance of even noble metal electrodes can be greatly reduced by simply roughening their surfaces. For example, the well known 'platinum-black' electrode is simply a platinum electrode covered or 'platinised' with a finely divided deposit of platinum, thus generating a very rough surface and a small interface impedance. As discussed above, the interface impedance of a noble Ag electrode can be decreased by the use of a 'depolarising' layer such as AgC1 (MCADAMS et al., 1992a;b). It is interesting to note en passant that most biomedical electrode systems which have been used successfully for biosignal recording or for electrical stimulation have either inherently or intentionally roughened surfaces. We can therefore conclude that for EIT applications it may be advantageous to ensure an optimal roughness of the electrode surface (ZHOU et al., 1995; GERSING et al., 1995; RIGAUD et aL, 1995) and, somewhat surprisingly, that the actual electrode material used may be relatively unimportant (JosSlNET et al. 1995). 2.2 Effect of electrode gel Electrode gels serve to ensure an optimal electrical contact between the electrode and the patient's skin, and to decrease the high epidermal impedance. There exist two main types of electrode gel; 'wet' gels (often described as pastes, creams or jellies) and hydrogels.

400

Standard electrode gels are usually composed of water, a thickening agent, a bactericide, fungicide, an ionic salt and a surfactant (CAIUM, 1988). The ionic salt serves to ensure the electrical conductivity of the gel. As the major portion of ions present in tissue fluids and sweat are sodium, potassium and chloride (CI-), in order to ensure bio-compatibility the ionic salts most commonly used in electrode gels are NaC1 and KCI. Generally, a relatively high concentration of electrolyte is used to decrease the value of the charge transfer resistance (thus rendering the electrode more non-polarisable) and to decrease the skin impedance. When a standard pre-gelled electrode is applied to the skin, the gel rapidly fills up the pores and wrinkles in the skin under the electrode, thus ensuring maximum effective contact area. The magnitude Ks of the capacitive component of the epidermal impedance (where ZcPAcs)= Ks(jog) -~) is therefore initially observed to drop rapidly in value following electrode application and then to remain relatively constant (MCADAMS and JOSS1NET, 1991a). Although Ks does not exhibit a strong time dependence, it does vary with the electrolyte composition and concentration (GATZKE, 1974), decreasing with increasing concentration (OH et al., 1993). The skin's parallel resistance Rp is generally observed to decrease with time in a pseudo-exponential manner as the ions in the gel gradually diffuse through the skin and make it more conductive. The time constant for this decay appears to be inversely proportional to the concentration of the gel (MCADAMS and JOSSINET, 1991a). As a result, aggressive gels tend to be used in biosignal monitoring applications such as 'stress testing' where 'instant' high-quality traces are required (MCADAMS 1990). Biological tissues cannot tolerate, in particular long-term, exposure to salt concentrations which depart significantly from physiological levels (0-9% NaC1 for body fluids, 0- I-0.4% NaC1 for human sweat) (REILLY, 1992). Aggressive gels (>> 5% NaC1) should not therefore be used, for example, for the monitoring of neonates or for the longterm monitoring of bed-ridden patients. As the contribution of Re is negligible at EIT frequencies, there is perhaps less need for such aggressive gels. The skin's capacitive component does, however, depend on the concentration and hence on the conductivity of the electrolyte (GATZKE, 1974; REILLY, 1992). Hydrogel-based electrodes have become popular for numerous biomedical applications. Hydrogels are 'solid' gels which incorporate natural (e.g. karaya gum) or synthetic (e.g. polyvinyl pyrrolidone) hydrocolloids (CAPdM, 1988). The use of such 'solid' gels entails numerous advantages when they are used in conjunction with screen printing technology (MCADAMS et al., 1994a;b), especially for EIT and related applications (PdSACHER et al., 1993). For example, it is possible to construct thin, lightweight highly flexible electrode arrays with accurately defined electrode/gel areas, shapes and inter-electrode distances. The use of an adhesive hydrogel pad dispenses with the need for the standard gel-impregnated sponge, gel-retaining ring and surrounding disc of adhesive foam. Hydrogels tend also to cause less skin irritation than 'wet' gels and standard adhesive backings. Hydrogels, being hydrophilic, are poor at hydrating the skin and may even absorb surface moisture. It is for this reason that hydrogels are now used for wound dressings to absorb exudate. As hydrogels tend to accommodate the skin surface's contours and irregularities, they are good at increasing the effective contact area between the electrode and the skin compared to dry electrode systems. With hydrogel (and metal plate) electrodes, Rp is observed to fluctuate while Ks remains relatively constant (MCADAMS and JOSSINET, 1991 a). The variations in Rp are largely dependent on sweat gland activity, with Rp decreasing during increased activity and

Medical & Biological Engineering & Computing

November 1996

gradually increasing again as the hydrogel absorbs the excess surface moisture (MCADAMSand JOSSINET, 1990; MCADAMS et al., 1994c). Values of Rp and Ks tend to be larger for hydrogels than for standard 'wet' gels. Typical values of Rp for hydrow can be as high as 15 M~ cm 2 and as high as 5 Mf~ cm for 'wet gels. Ks can typically range between 2 and 10 Mf~ s -~ cmz for 'wet' gels and have a value of up to 50 Mf~ s -~ cm 2 for hydrogels. One way of decreasing the value of Ks is to use thinner hydrogel pads (MCADAMS and JOSSINET, 1991a). The stratum corneum and the hydrogel layers can be imagined forming the dielectric layer between the two conductive 'plates', i.e. the electrode and the underlying tissues. Reducing the thickness of the stratum corneum or the hydrogel layers will lead to an increase in the capacitance. Larger hydrogel pad areas can also be essayed. This need not necessarily entail the use of a larger overall electrode area as the adhesive hydrogel may not require the use of a large surrounding disc of adhesive foam. Some researchers believe that in order to maximise the distinguishability of inhomogeneities in a two-dimensional image, electrodes should be as wide as possible so as to nearly fill the available circumference (ISAACSON, 1986). Newell showed that long electrodes can result in greater accuracy (NEWELL, 1995). Newell and his colleagues used hydrogel-based electrodes in their experiments as such tests would be difficult with wet electrodes. Long, narrow fingerlike hydrogel electrodes have been used by other EIT researchers (BAISCH et al. 1995); although they are narrow, they have significant surface areas. The fact that hydrogel electrodes can be made and cut to whatever size or shape required is a major advantage which significantly compensates for their higher skin impedances. Many such electrodes can be placed close together on a body site with little risk of electrical shorting between the electrodes, a major problem with 'wet' electrodes. Hydrogels tend to be more resistive than standard 'wet' gels. Typical resistivities for 'wet' gels are in the order of 5 500 f~ cm-~ (the higher the salt concentration, the lower the resistivity) compared to 800--8000 f~ c m - l for hydrogels (the higher resistivity hydrogels tend to be used in cardiac pacing electrodes). 'Wet' ECG electrodes have gel layer thicknesses of around 0.3 cm and typical areas of 3 cm 2. The resistance of a 'wet' gel layer is therefore generally in the range 0.5-50 fL Although hydrogels have higher resistivities, this disadvantage may possibly be compensated for by the use of larger gel areas. Many of the EIT electrode harnesses constructed at the Northern Ireland Bioengineering Centre (NIBEC) for various research groups working in this field have hydrogel pad resistances of around 10 f~. This surprisingly low value is due not only to the appropriate choice of the hydrogel and pad area used, but also to the thickness of the hydrogel layer. It is interesting to note that gel layer thickness is a variable generally ignored in electrode design, even though it can have a significant effect on electrical performances. As many of the commercial hydrogels used in biosignal monitoring electrodes have layer thicknesses of around only 1 mm (compared to around 3 mm for pre-gelled 'wet' electrodes), it is therefore not surprising that the pad resistances can be so low. Based on unpublished data, we believe that further improvements can be made to the performances of hydrogelelectrodes (and 'wet' electrodes) by the use of even thinner gel layers. It must be remembered that the electrolyte resistance is not solely determined by the dimensions and properties of the gel pad. When a large gel padis used in conjunction with a small sensor, the dimensions of the sensor will help determine the magnitude of the electrolyte resistance as the large overlapMedical & Biological Engineering & Computing

ping section of gel pad will carry relatively little current. An extreme example of this is the case where a small spherical electrode is placed in a large tank filled with electrolyte. The electrolyte resistance in this case is given by RELECTROLYTE = p/4rcro

(3)

where ro is the radius of the electrode and p is the resistivity of the electrolyte. The electrolyte resistance therefore depends on the dimensions of the small electrode rather than on those of the large electrolyte tank. 2.3 Effect of inter- and intra-human variations The impedance of the skin is known to vary greatly from patient to patient and from body site to body site. Generally, due to the frequency ranges used (Hz to tens or hundreds of kHz), these reported variations are largely caused by differences in the values of Rp (MCADAMSand JOSSINET, 1991a). The skin's resistance is dependent on the presence and activity of sweat glands and on the presence of other appendageal pathways. The density of sweat glands varies over the body surface with a value of approximately 370 per cm 2 on the palms of the hands and the soles of the feet and a value of approximately 160 per cmz on the forearm (REmLY, 1992). The diameter of the ducts can range from 5 to 20 gm, which may also help explain the wide range in Rp values. Owing to the high frequencies used in EIT, the capacitive properties of the skin are the most important. The capacitance of the skin has a typical value in the range 0.020.06 gF cm -2 (EDELBERG, 1971; YAMAMOTO and YAMAMOTO, 1978). Its value is related to the thickness and composition of the stratum comeum. In general, fair-skinned subjects have thinner stratum comeum layers than those with dark skin. The stratum corneum in the dark-skinned group is denser and has more cell layers (KLINGMAN. 1984). The stratum comeum of pale-skinned subjects who sunbathe tends to thicken with exposure to sunlight. Not surprisingly, skin impedance tends to be much higher for dark-skinned subjects. Considerably larger Ks values for dark-skinned subject have been reported (MCADAMS and JOSSINET, 1991a). According to Klingman, the thickness of the stratum corneum does not vary appreciably with age or sex (KLINGMAN, 1984). A slightly higher average value of skin impedance was reported by Lawler et al. for females compared to males, although it was concluded that the results were not statistically significant (LAWLER et al., 1960). Schrnitt and Almasi measured the skin impedance of a relatively large number of subjects and obtained straight lines when they plotted their data on log-normal cumulative probability plots (SCHMITTand ALMASI, 1970). They reported that the average female skin impedance is about 50% higher than that of male skin. They asserted that, although the basis of the male-female difference was not clear, the trend in their data was undeniable. The size of male and female epidermal cells tend to differ and this has been suggested as a possible cause of the observed differences in impedances. Schmitt and Almasi also pointed out that there is considerable daily variation in a given subject, and seasonal changes have also been observed (YAMAMOTOand YAMAMOTO, 1978). These variations may help explain some of the discrepancies in the literature. The number of cell layers in the stratum comeum can range from 12 to 30 (KLINGMAN,1984). Stratum corneum thickness can therefore vary greatly for different body sites within the range of about 10 lam to well over 100 pan. The stratum corneum can be as thick as 400 to 600 gm in the palm and plantar areas, for example, and as little as 10 to 20/am on the back, legs and abdomen (CHIEN, 1982). Gerstner and Gerb-

November 1996

401

stadt demonstrated that areas of the skin with a thick stratum corneum had higher impedances than those with thin layers (GERSTNERand GERBSTADT,1949). As the stratum comeum is typically at least ten times as thick on the palms of the hands and soles of the feet as other body areas (REILLY, 1992), the skin capacitance at these points is considerably smaller (i.e. Ks is larger) than at other sites on the body. The presence of an increased density of sweat glands at these sites, however, can result in low values of Rp depending on their activity. Lawler et al. observed that impedances (measured between 1 kHz and 20 kHz and hence dominated by Ks) tended to be highest on the palms, greater on the dorsal forearm than on the ventral forearm and greater on the dorsal arm than on the ventral arm. (LAWLER et al., 1960). They therefore concluded that the impedance of normal skin is highest in areas with a thick stratum corneum. The stratum comeum on the face and scalp is not as substantial as that on other body parts (KLINGMAN,1984). This and the presence of a high density of hair follicles (which act as low resistance shunts) gives rise to a low value of skin impedance. Of the body sites used in reported skin impedance measurements, the forehead appears to have the lowest impedance value (GRIMNES, 1983c). It is hard to determine from such reports if this low impedance value is due to a small Ks value or a small value of Rp or both. 2.4 Effect of skin preparation technique The presence of minute cuts on the skin surface or the deliberate puncturing of the skin can short-out the epidermal impedance, resulting in very low and relatively stable skin impedances. Previously, 'multi-points' electrodes resembling nutmeg graters have exploited this advantage when they were pressed against the skin prior to biosignal recording. Abrasion of the skin effectively removes a large proportion of the epidermal impedance. Several ECG electrodes are currently supplied with abrasive pads built in to the electrode release backing. A Skin Raspt which resembles a strip of Velcrot can be obtained from Medicotest for this purpose. The Quinton Quick-Prep Applicatort rotates the abrasive centre of the Quick-Prept electrodes, causing a marked decrease in skin impedance. The LIT group at University College London uses a similar idea to the Quinton Quick-Prep Applicatort in their breast electrode harness (HOLDER,1995). Adhesive tapes have been used to progressively strip away layers of the epidermis (LAWLERe t al.; 1960, YAMAMOTOand YAMAMOTO, 1976). Such skin stripping can result in a dramatic decrease in the skin impedance. Lawler et al. observed that the removal of a single strip of the superficial keratin may, however, result in an increase in the skin resistance, which they believed was due to the removal of conductive' surface electrolytes. With continued stripping there is a dramatic decrease in the value of Rp and an increase in Cp (inversely proportional to Ks). The outermost layers of the statnma comeum are the most resistive. As a result, the most significant decreases in Re are obtained with the first few strippings (YAMAMOTO and YAMAMOTO, 1976). As stated above, the stratum comeum can be imagined forming the dielectric layer between the two conductive plates of a capacitor, i.e. the electrode and the underlying tissues. Reducing the thickness of the stratum corneum will lead to an increase in the capacitance (a decrease in Ks). The impedances of three sites in parallel over the frequency range 5-100 kHz have been monitored in vivo using a threeelectrode technique (PdSACHER, 1995). Site 1 was the reference site, site 2 was stripped ten times using Scotch tape and t" registered trademark

402

site 3 was stripped 20 times. Ks was found to decrease slightly with ten strippings relative to the reference. With 20 strippings the stratum comeum appears to have been completely removed as there was a drastic decrease in Ks. ~ also tended to decrease with skin stripping (SALTER, 1981). The value of Rp for the X10 site was less than that for the reference site and the value for the X20 site was almost zero due to the complete removal of the stratum corneum. Although the complete removal of the stratum comeum gave rise to highly desirable skin impedances, such a procedure would obviously be clinically unacceptable due to the discomfort caused to the patient during and following the experiment. The outer layer of the skin serves to protect the body from irritating substances, such as electrode gels and adhesives. When the outer layer is substantially removed, not only is the procedure painful, but the skin is also subsequently more susceptible to irritants. For a given gel composition, the more the skin is abraded, the sooner discomfort develops and the more severe the irritation. The level of irritation also varies with the salt concentration and the additives in the gel. Mild abrasion or stripping which does not totally remove the stratum comeum should only be used. Such preparation will decrease skin impedance without giving rise to pain, bleeding or irritation. Such mild abrasion will cause a marked decrease in Rp and a significant decrease in Ks. Lawler reported that washing the skin with ether resulted in a decrease in resistance, which they attributed to the removal of poorly conducting lipid substances from the surface of the skin (LAWLER et al., 1960). Routine degreasing of the skin with alcohol may initially increase the impedance of the skin by dehydrating the outer layers of the skin. This does not appear to be widely appreciated. Electrolyte does, however, penetrate more readily once a 'wet' gel electrode is applied to the skin for several minutes. This may not be the case with hydrogel electrodes, and this form of skin preparation should therefore be used with care. Accompanied by vigorous rubbing, however, this preparation technique could result in low initial impedances due to the additional mild abrasion. An alternative method of rapidly decreasing skin impedance is to rub the skin site with a high-concentration electrolyte. This causes significant decreases in Rp and Ks, especially if it is accompanied by vigorous rubbing. Some gels contain abrasives such as crushed quartz, which greatly reduce skin impedance when rubbed into the skin prior to electrode application. The effects of treating skin sites with methanol and Redox paste prior to the application of a hydrogel electrode have been compared (RISACHER, 1995). The modulus and the phase angle of the overall impedance were monitored at an applied signal frequency of 64 kHz after the application of the electrode. Mod. Z decreased only slightly with methanol pre-treatment, but decreased markedly with Redox gel and appeared almost constant and independent of time. The phase angle of the overall impedance was observed to be larger for the Redox gel than for the other two surface treatments. This was possibly due to a decrease in RTOTAL accompanying the decrease in K . It is suggested that degreasing the skin with alcohol accompanied by rubbing with Redox or similar paste could optimally decrease the high-frequency skin impedance for EIT studies. 2.5 Penetration enhancers By virtue of its barrier qualities, the skin imposes constraints which have limited the number of drugs that are appropriate for transdermal delivery, for example. Much w Redox, registered trademark, Hewlett-Packard

Medical & Biological Engineering & Computing

November 1996

effort has been made to establish a safe reversible means of reducing this barrier to percutaneous penetration. One chemical approach to this problem is the use of 'penetration enhancers'. A penetration enhancer is an agent which increases the permeability of the skin to a given species. The mechanisms by which penetration enhancers act are not well understood. In some cases they may alter the hydration of the stratum corneum or alter the packing structure of the ordered lipids in the intercellular channels (KNEPP et al. 1987). More aggressive agents appear to operate by destroying or dissolving the lipids, and thus compromising the barrier function of the skin. The safest penetration enhancer is water. Hydration of the skin causes its cells to swell and the normally tight packing of the cells is loosened, thus rendering the skin more permeable (KELLY, 1985). Simply occluding the skin site for a period of time causes the normal flux of water from the skin surface (trans-epidermal water loss (TEWL)) to accumulate under the occluding patch, and thus progressively hydrate the skin. Although increasing the hydration of the stratum corneum greatly decreases the magnitude of Rp, it appears to have relatively little effect on the skin's capacitive properties. (FOLEY et al. 1992). Ks and a do, however, tend to decrease with skin hydration (SALTER, 1981). A class of penetration enhancer commonly used in electrolytic gels are surfectants. Such surface active agents are adsorbed at water-oil interfaces as a result of their hydrophilic (or polar) groups and lipophilic (or nonpolar) groups. By orientation at a water-oil interface, the molecules of the surfectant facilitate the transition between polar and nonpolar phases. One of the most widely used enhancers in transdermal drug delivery is the dipolar aprotic solvent dimethylsulphoxide (DMSO). Other sulphoxides such as decylmethysulphoxide (decyl MSO) have been used as penetration enhancers, particularly the alkylmethyl derivatives (KNEPP et aL, 1987). Chemically related to dimethylsulphoxide are the amides dimethylamide (DMA), dimethylacetamide (DMAC) and dimethylformamide (DMF). Cyclic amides that have been used include 2-NMP, 2-P and Azone* (WIECHERS, 1992). Along with DMSO, Azone is one of the most popular penetration enhancers reported in the literature (CHIEN,1987). The use of esters of saturated and unsaturated fatty acids such as oleic acid, myristic acid and capric acid have also been cited (CHIEN, 1987; CHIEN and LEE, 1987) as have, for example, propylene glycol (WIECHERS, 1992), ethanol (NOLAN et al., 1993), urea (CAMPBELL et al., 1977), sodium lauryl sulphate (NOLAN et al. 1993) and certain natural oils or components thereof (eucalyptus, chinopodium, carvone, 1,8cineole)(WIECHERS, 1992). Although the above researchers were concerned primarily with increasing the skin's permeability to various transderreally applied drugs, some carried out AC impedance studies on the effects of some of the various penetration enhancers cited, and these studies are of particular interest to our review. Foley et al. observed a dramatic decrease (by as much as 95% in Rp following treatment of the skin site with DMSO (FOLEY et al, 1992). The capacitive element of the skin's impedance was also affected. The dramatic effect was only observed when DMSO comprised more than 70% of the applied DMSO/ water mixture, and no evidence could be found that the observed changes were reversible. Ollmar suggesting using sodium lauryl sulphate on skin sites prior to EIT measurements as it could be used to obtain a dramatic improvement in the conductive properties of the skin (OLLMAR, 1995). The effects of sodium lauryl sulphate were

* registered t r a d e m a r k

Medical & Biological Engineering & Computing

investigated by Nolan et al. (NOLANet al., 1993). The parallel resistance of the stratum corneum was typically reduced by approximately 25% on exposure to a 0-1% solution and 95% for a 0.2% solution. The capacitance was also affected. Nolan et al. found that the parallel resistance of the stratum corneum was considerably reduced following treatment with a 1% solution of oleic acid in ethanol. The skin's capacitance tended to increase following treatment. It is interesting to note, however, than an increased concentration of enhancer does not necessarily lead to a correspondingly increased change in the measured impedance. There can be an optimal concentration which is the most effective at reducing the barrier properties of the skin. In addition, although the use of a given penetration enhancer may increase the skin's permeability to certain drugs, it does not always follow that the skin's electrical impedance will be decreased. For example, Kontturi et al. found that dodecyl N,N-dimethylaminaocetate (DDAA) and Azone, although both are used to enhance the penetration of various drugs, gave rise to an increase in the value of Rp (KONTTURI et al., 1993). Kontturi et al. attributed this to the blocking of aqueous pores and other hydrophilic channels by the enhancers. Ks and ~ were, however, observed to decrease and this was attributed to an increase in the heterogeneity of the skin surface due to the production of new transdermal penetration routes and to the increased disorder of the lipoidal matrix. Unfortunately, this interesting concept was not investigated further. The use of penetration enhancers holds much promise and certain surfectants are already included in electrolyte gels for biosignal monitoring electrodes (CARIM,1988). The challenge lies in achieving this without severe irritation or damage to the skin. Enhancers which do not damage or compromise the skin irreversibly are unfortunately not common. 2.6 Effect o f temperature On occasions, it has been observed that when a cold gel is first applied to a skin site, the measured value of Rp is observed to initially increase (ROSELL et al. 1988). This is thought to be due to the cold gel causing the sweat pores etc. to contract. Once the gel has warmed up, the value of Rp is observed to decrease pseudo-exponentially as the electrolyte ions diffuse through the epidermal layer. Klingman has pointed out that the use of occlusive dressings in transdermal drug delivery not only increases the hydration of the stratum corneum, but also helps raise the surface temperature of the skin surface to core temperature (37~ thus enhancing penetration by adding to the thermodynamic driving force (K.LINGMAN, 1984). Oh et al. and Nolan et al. observed a decrease in the skin's parallel resistance and an increase in the parallel capacitance as the temperature of human stratum corneum was varied in vitro from 20 to 80~ (303 to 353~ et al. 1993; NOLAN et al. 1993). This response exhibited several transitions. The gradual decrease in Rp with temperature over th e range 2060~ is thought to reflect an increase in the mobility of ions in the ion-conducting pathways. Up to around 60~ the observed changes in impedance were found to be quite reversible, given sufficient time. EPR (electron paramagnetic resonance) spectroscopy showed a progressive increase in lipid disordering, with one or more phase changes occurring within this temperature range (NOLAN et al., 1993). Above 60~ a specific phase change occurs which has been attributed to lipid melting. It would appear that ions can then pass more freely through the lipid pathways, resulting in a dramatic decrease in Rp and a significant increase in Cp (OH et al., 1993). Oh et al. postulated that the large increase in the disorder of the lipid alkyl chains at the phase transition results in a considerable

November 1996

403

enhancement of the skin's dielectric constant, and thus in the observed increase in capacitance. The changes in the skin's impedance above 60~ were found to be irreversible (NOLAN et al., 1993). This is an interesting observation as often the epidermal layers of excised skin samples for in vitro testing are separated by heating the samples in distilled water at 60~ Although it is obviously not clinically acceptable to heat patients to such temperatures, keeping them as warm as reasonably possible should prove beneficial to electrical performances. The skin sites could be heated directly, for example, by means of heating elements incorporated into the electrode systems.

2.7 Effect of applied signal amplitude The AC impedance of the skin has been found to depend on the applied signal amplitude. The major source of the observed non-linearity tends to be the skin's parallel resistance Re (OH et al. 1993). A plot of direct current versus direct voltage appears exponential (KASTING,1992), similar in form to those observed at electrode-electrolyte interfaces and described by the Butler-Volmer equation (MCADAMS and JOSSINET, 1991b). Kasting noted the similarity between I-V curves obtained for skin and the current-overpotential curves characteristic of many electrochemical reactions which are successfully described by kinetic theory. Such exponential I-V behaviour is thought to be indicative of a potential energy barrier which hinders the movement of charged particles across the interface and which can be modified in height by the interfacial potential difference. Kasting therefore postulated the reorientation of the lipids comprising the epithelial cell walls of the sweat ducts by the electric field as a possible source of such a 'potential-dependent energy barrier' to ion transport in skin. This phenomenon occurs in synthetic lipid bilayers, for example, and is known as reversible dielectric breakdown. Although Kasting's model can successfully interpret observed I-V characteristics, it is relatively simple and does not, for example, include electro-osmotic effects. Rp, which is given by the inverse slope of the direct current-voltage curve, is obviously very nonlinear (MCADAMS and JOSSINET, 1991b; 1992; 1994) and will decrease with applied signal (AC or DC) amplitude (YAMAMOTOand YAMAMOTO,1981). Compared to Rp, ZcpA is relatively linear and often assumed constant (OH et aL, 1993). We have, however, observed in unplished data that Ks does gradually decrease with applied signal amplitude. Moderately increasing the applied AC signal amplitude will not, however, significantly reduce the contact impedance in EIT. The application of a DC signal prior to EIT measurements could possibly help decrease the skin impedance. The effects of DC signals on the skin's impedance have been of particular interest in the field of iontophoretic transdermal drug delivery. Iontophoresis, in this context, may be defined as the facilitation of ionisable drug delivery across the skin by an applied electrical potential. In conventional topical treatment by iontophoresis, a reservior of ionised drug is placed in contact with an electrode and the drug is driven through the skin because the electrode has the same polarity as the drug ions. The direct current is carried by the drug ions (and any competing ions in the formulation or in the skin). The applied electrical field can, in certain cases, drive the ions through the skin far more effectively than pure diffusion or 'passive' transdermal delivery. The transport of a given ion across the skin will depend on the strength of the electric field, the concentration of all ions in the electrolyte and skin, and on the mobility of these ions (PHIPPS and GYORY, 1992). Not only 404

does the ion-electric field interaction provide an additional force which drives the ions through the skin, but iontophoresis also enhances the transdermal drug delivery as a result of the current flow increasing the permeability of the skin and of electro-osmosis producing bulk motion of the solvent which carries the ions (PIKAL, 1992). It is generally believed that during passive diffusion of ions through the skin the skin's appendages play little part as they constitute approximately only 0-1% of the surface area of the skin. During iontophoretic transdermal delivery, on the other hand, the ionic current appears to seek out the channels of least resistance and is more concentrated at hair follicles, sweat glands etc. (CUELANDER, 1992). Although most work in this area aims to increase the flow of ionised drug molecules into the body for therapeutic effect the same mechanisms could, it is suggested, be used to rapidly decrease the skin's impedance prior to EIT measurement by actively pushing the gel's ions into the skin. AC impedance measurements carried out following the application of a DC signal to a reservoir containing a simple electrolyte indicate that Rp decreases significantly as a result of the DC signal (up to 0-2 mA cm-2)(KALIA and GUY, 1995). The larger the DC density or the larger the electrolyte concentration is, the more pronounced the effect. The value of Rp does, however, gradually recover with time following removal of the DC signal. This recovery takes longer for larger applied current densities or larger electrolyte concentrations. The reported in vivo data appear consistent with Kasting's kinetic model (KASTrNG, 1992). The loss of skin resistance is thought to be attributable to (i) the reorientation of lipid molecules in hair follicles and sweat glands resulting in the enlargement- of pre-existing channels. (ii) the reorientation of lipids or keratin bundles in the bulk stratum comeum and the formation of transient conduction channels. (iii) an increase in the ion concentration in the resistive pathways (KALIA and GuY, 1995). Unfortunately, in the literature the skin's capacitive properties appear to have been largely ignored. Where they have been considered, they appear to remain relatively unaffected by the prior use of a DC signal, or the presented data are ambiguous. This area therefore requires and warrants further research to clearly establish the potential of this technique and to reduce the high frequency skin impedance for EIT studies. The observed decrease in resistance as a result of applied DC signals is often only partially reversible (KASTING, 1992), especially for larger signal amplitudes. Rosendal, for example, noted that for voltages greater than 2.5 V, the skin resistance decreased irreversibly (ROSENDAL, 1943). When a large electrical impulse is applied to the skin, its high electrical impedance is drastically reduced due to some form of 'breakdown' phenomenon. The breakdown can be associated with either Re ('electro-osmotic' breakdown or electro-poration) or Zcp~ ('dielectric' breakdown). Dielectric breakdown is relatively slow to occur and not pronounced for voltages below 300-500 V. Such dielectric breakdown has been reported in the literature for dry excised skin. With intact living skin a sudden impedance breakdown is often noted at significantly lower voltages than those needed to breakdown exised samples (REILLY, 1992). Grimnes therefore questioned whether it was generally relevant to consider a dielectric breakdown mechanism at such low voltages and suggested instead an electro-osmotic breakdown mechanism (GRIMNES, 1983 a;b). He pointed out, however, that with large electrolytic currents, heat and mechanical tissue destruction should also be considered.

Medical & Biological Engineering & Computing

November 1996

Electro-osmosis is a capillary transport mechanism of bulk liquid caused by viscous forces when the mobile part of the charged double layer is moved by an electric field (GRIMNES, 1983b). The pores in the skin can be thought of as capillaries. An electrical double layer exists at the interface between the electrolyte and a capillary wall. The skin appears to have an isolectric point between 3 and 4. For lower pH values the skin will be positively charged, whereas for larger pH values (as tends to be the case with electrode gels) the skin will be negatively charged (PHIPPS and GYORY, 1992). The pore walls therefore generally have immobile negative charges, and the mobile ion atmosphere in the electrolyte on the capillary walls has a net positive charge. These positive counter-ions are free to move in response to the imposed electric field. If an electrical field is applied across a pore, this field produces a force on these mobile positive ions in the double layer (ion atmosphere) causing them to migrate towards the negative pole. Viscous forces exerted by these migrating ions cause the bulk liquid in the pore to also be transported, and hence electro-osmotic flow occurs in the same direction as counterion migration (PIKAL, 1992). AS counter-ions tend to be cations (positive ions), electro-osmotic flow generally occurs from anode to cathode. With an electro-osmotic mechanism, breakdown can occur at relatively low voltages, as observed experimentally. Grimnes and Yamamoto et al. believed that only electroosmosis could explain the wide range of phenomena reported in the literature associated with the non linear behaviour and breakdown of the electrical properties of the skin (GRIMNES, 1983b; YAMAMOTOet aL, 1986). The current is proportional to the exponent of the applied voltage during electro-osmotic transport (GRIMNES, 1983b). As a consequence, the value of Re will decrease with applied signal amplitude and current 'runaway' is observed. In the initial phase this current runaway is reversible. The subject experiences a pricking sensation, which is most probably due to the high current densities concentrated in the pores. In the literature there has been considerable interest in the possible electroporation of the skin, especially for use in transdermal drug delivery. Whereas iontophoresis acts primarily on the drug, involving skin structural changes as a secondary effect, eletroporation appears to act directly on the skin, causing transient changes in tissue permeability (PRAUSNITZ et al. 1993). Electroporation of isolated single cells is well established. When the electric field strength across a cell membrane is sufficiently large, aqueous pores will form in the membrane, thus giving rise to an increase in membrane permeability. The creation of such pores in the membrane, termed electroporation(REILLY,1992), can be a reversible process. At largervoltages, however, pore size increases and/or several pores fuse causing irreversible rupture of the membrane. Weaver and his colleagues have published a range of papers on electroporation, including the electroporation of the skin to increase the skin's permeability to ions (WEAVER, 1993; PRAUSNITZ et al. 1993). Transdermal transport through the stratum corneum takes place largely through the inter-cellular lipids, which are mainly organised in bilayers. Although it is still not well understood, it is believed that electroporation of the stratum comeum can create aqueous pores in the lipid bilayers by a mechanism involving only transient, reversible structural changes. Such 'reversible electrical breakdown' would appear to occur when the transmembrane voltage reaches 0-5-1 V in a time of/~s to ms. Increases in the transdermal flux of polar molecules and dramatic reductions in the skin's 'resistance' have been observed (PRAUSNITZ et al. 1993). Although interest in this area now appears to be on the wane, the use of electroporation in-combination with some of Medical & Biological Engineering & Computing

the other techniques outlined here should prove an interesting and potentially fruitful area of study for those interested in reducing skin impedance for optimal EIT measurements.

3 Summary and conclusions In EIT, we are generally interested in relatively high frequencies. The contributions of the parallel resistances in the equivalent circuit model of the electrode-gel-skin interface shown in Fig. 1 are therefore relatively unimportant and the capacitive properties of the eletrode-gel interface and the epidermis have a more important role. At very high frequencies, even the series resistance of the gel pad (which is generally ignored) may become significant. The high-frequency capacitive properties of the electrodegel interface are largely determined by the roughness of the electrode surface and the electrode material may be relatively unimportant. Variations in skin impedance will occur around a given body segment under investigation due to variations in epidermal layer thickness, condition etc. It is therefore important to minimise skin impedances, their mismatch and their fluctuations with time as far as is reasonably possible. Although the use of 'wet' electrode gels results in lower skin impedances and gel resistances, the use of hydrogels can be advantageous in multi-electrode EIT harnesses. In any case, gel concentration should be as high as possible without causing skin irritation problems. Where possible, the gel pads should be as thin as possible and have as large an area as other constraints will allow. These measures will minimise the gel pad resistance and will also help decrease the magnitude Ks of the skin's pseudo-capacitive impedance. Relatively simple techniques which can be used to minimise skin impedance problems include keeping the patient as warm as possible. When 'wet' gelled electrodes are used, they should be left on the patient for as long as possible (at least 10 min) before commencing measurements. Skin abrasion, if used, should be moderate to avoid causing trauma to the patient. A potentially attractive technique involves rubbing the electrode sites with an 'abrasive' electroIytic get such as Redox paste to abrade the skin and render it more conductive. (Obviously, care must be taken when preparing adjacent sites to avoid the possibility of electrical shorting via the conductive paste). This simple technique appears to have a considerable effect on the high-frequency skin impedance. Other techniques of potential interest include the use of penetration enhancers, localised temperature, pressure and electrical impulses. Some of these techniques are not without their difficulties, however, and a combination of several of them may hold the key to success. The following examples are used to illustrate some of the possibilities. It has been suggested that for short-term, resting EIT measurements the use of 'cradles' or 'cages' of reusable spring-loaded electrodes or pneumatic electrode systems appears well suited (MCADAMS et al., 1994a). Such systems have been used for EIT monitoring of the breast. (JOSSINET, 1988; HOLDER,1995). Krenzke et al. have patented a similar electrode system for use in cardiac mapping (KRENZKE et al., 1985). The electrodes in such systems could conceivably incorporate a heating element to warm the skin site, and thus decrease skin impedance. The application pressure of the electrodes (possibly with roughened surfaces) could be optimised to further decrease the skin impedance without causing discomfort to the patient. Holder has exploited electrode surface roughness by rotating roughened electrodes in situ to abrade the skin (HOLDER,1995). A potentially attractive extension of this concept would be to monitor the skin

November 1996

405

impedance during skin abrasion and to use this information as feedback to control the rotating mechanism. The Quinton Quick-Prep Applicator apparently operates using this principle. Persons or body sites with high skin impedances could cause problems because such a device would continue to abrade until the theoretically desirable impedance value was obtained (the subject may not find this procedure highly desirable). An alternative would be to apply some form o f electrical impulse to the electrodes to decrease contact impedance. Feedback could ensure that suitably low impedances are achieved and even minimise impedance mismatch. Once again, although this concept is theoretically very attractive, the safety and comfort of the subject will be critical factors in determining the suitability of such a system. We would like to reiterate one further advantage of these systems. Given the orientation and extension o f the pistons (and their variations with time), the relative positions of the electrodes could be computed (MCADAMS et al., 1994a). Inter-electrode distances and the dynamic shape and volume of the body segment under study may then be derived from the above data, and used in conjunction with the EIT algorithm to increase the accuracy o f the EIT image.

References BAISCH, F. J., HAHN, G., SIPINKOV,~,I., BEER, M., and HELLIGE, G. (1995): 'Comparison of electrode belts with 'spot' electrodes in electrical impedance tomography'. Proc. Concerted Action on Impedance Tomography (CAIT) Workshop on The ElectrodeSkin Interface in Electrical Impedance Tomography, Belfast, UK, Innov. Techno. Bio. Med., 16, (2), pp. 119-125 BARBER, D. C., BROWNB. H., and FREESTON,I. L. (1983): 'Imaging spatial distributions of resistivity using applied potential tomography', Electron Lett., 19, pp. 933-935 BARD, A. J. and FAULKNER,L. R. (1980): 'Electrochemical methods' (John Wiley) BOONE, K., and HOLDER, D. S. (1995a): 'Assessment of noise and drift artefacts in electrical impedance tomography measurements using the Sheffield Mark 1 system', Proc. Concerted Action on Impedance Tomography (CAIT) Workshop on The Electrode-Skin Interface in Electrical Impedance Tomography, Belfast, UK, Innov. TechnoL Biol. Med., 16, (2), pp. 49-60 BOONE, K., and HOLDER, D. S. (1995b): 'A model of the effect of variations in contact and skin impedance on electrical impedance tomography measurement artefacts'. Proc. Concerted Action on Impedance Tomography (CAIT) Workshop on The Electrode-Skin Interface in Electrical Impedance Tomography, Belfast, UK, Innov. Techno., Bio. Med., 16, (2), pp. 61-70 BROWN, B. H., BARBER,D. C., WANG,W., LU, L., LEATHERHEAD,A. D., SMALLWOOD, P,.. H., HAMPSHIRE, A., MACKAY, R., and HATZIGALAMIS,K. (1994); 'Multi-frequency imaging and modelling of respiratory related electrical impedance changes', PhysioL Meas. 15, supplement A, A1-A12 BROWN, L., and LANGER,R. (1988): 'Transdermal delivery of drugs', Ann. Rev. Med., 39, pp. 221-229 CAMPBELL,S. D., KRhNrNG, K. K., SCHIBLI,E. G., and MOMMI,S. T. (1977): 'Hydration characteristics and electrical resistivity of stratum comeum using a non-invasive four-point microelectrode method', J. Invest. Dermatol., 69, pp. 290--295 C~d~dm, H. M. (1988): 'Bioelectrodes' in WEBSTER, J. G. (Ed.): 'Encyclopedia of medical devices and instrumentation' (John Wiley and Sons, New York) pp. 195-226 CHIEN, Y. W. (1982): 'Transdermal controlled-release drug administration' in Swarbrick, J. (Ed.): 'Novel drug delivery systems' (Marcel Dekker Inc., New York) p. 149 CHIEN, Y. W. (1987): 'Development of transdermal drug delivery systems', Drug Devel. Ind. Pharmacy, 13,(4&5), pp. 589-651 CHIEN Y. W., and LEE, C-S. (1987): 'Transdermal drug delivery system with enhanced skin permeability', Am. Chem. Soc., Symp. Series, 34E pp. 281-300

406

CULLANDER, C. (1992): 'What are the pathways of iontopheritic current flow through mammalian skin?', Adv. Drug Delivery Rev., 9, pp. 119-135 CULLANDER, C., and GUY, R. H. (1992): 'Visualization of iontophoretic pathways with confocal microscopy and the vibrating probe electrode', Solid State lonics, pp. 53-56, 197-206 DE LEVIE, R. (1965): 'The influence of surface roughness of solid electrodes on electrochemical measurements', Eletrochim. Acta, 10 pp. 113-130 EDELBERG, R. (1971): 'Electrical properties of the skin' Jn Elden, H. R. (Ed.) 'A treatise of the skin' (John Wiley & Sons) FOLEY D., CORISH J., and CORRIGAN O. I. (1992): 'Iontophoretic delivery of drugs through membranes including human stratum comeum', Solid State Ionics, pp. 53-56, 184--196 FUKS, L. F, CHENEY, M., ISAACSON,D., GISSER,D. G., and NEWELL, J. C. (1991): 'Detection and imaging of electric conductivity and permittivity at low frequency', IEEE Trans., BME-38, pp. 11061110 GATZKE, R. D. (1974): 'The electrode: a measurement systems viewpoint' in MILLER, H. A., and HARRISON, D. C. 'Biomedical electrode technology' (Academic Press, New York) pp. 99116 GERSINGE., SCHAFER,M., and OSYPKA,M. (1995): 'The appearance of positive phase angles in impedance measurements on extended biological objects'. Proc. Concerted Action on Impedance Tomography (CAIT) Workshop on The Electrode-Skin Interface in Electrical Impedance Tomography, Belfast, UK, Innov. Technol. Biol. Med., 16, (2), pp. 71-76 GERSTNER H., and GERBSTADT,H. (1949): 'Der wechselstromwiderstand der menschlichen Haut', Arch. f d. ges. PhysioL, 252 p. 111 GOBLE, J., and ISAACSON,D. (1989): 'Optimal current patterns for three-dimensional electric current computed tomography', Proc. Ann. Int. Conf. of IEEE Engineering in Medicine and Biology Society, 11, pp. 463-464 GONDRAN, C., SIEBERT, E., YACOUB, S., and NOVAKOV, E. (1995): 'Dry electrode based on nasicon ceramic for surface biopotenti~.l measurements'. Proc. Concerted Action on Impedance Tomography (CAIT) Workshop on The Electrode-Skin Interface in Electrical Impedance Tomography, Belfast, UK, lnnov. Technol. Biol. Med., 16, (2), pp. I13-118 GRIMNES, S. (1983a): Dielectric breakdown of human skin in vivo', Med. Biol. Eng. Comput., 21 pp. 379--381 GRIMNES, S. (1983b): 'Skin impedance and electro-osmosis in the human epidermis' Med. BioL Eng. Comput., 21, pp. 739-749 GRIMNES S. (1983c): 'Impedance measurement of individual skin surface electrodes', Med. Biol. Eng. Comput., 21, pp. 750-755 HOLDER, D. S. (1995): Design and electrical characteristics of a multi-level electrode array for electrical impedance tomography of the female breast'. Proc. Concerted Action on Impedance Tomographay (CAIT). Workshop on The Electrode-Skin Interface in Electrical Impedance Tomography, Belfast, UK, Innov. Technol. BioL Med., 16, (2), pp. 143--150 ISAACSON,D. (1986): 'Distinguishability of conductivities by electric current computed tomography', IEEE Trans., M-5,(2), pp. 9295 JOSS1NET,J. (1988): 'A hardware design for imaging the electrical impedance of the breast', Clin. Phys. PhysioL Meas., 9, supplement A, pp. 25-8 JOSSlNET, J., MCADAMS,E., MCLAUGHLIN,J., and JARRY, R. (1995): 'The interface impedance of metal electrodes for electrical impedance tomography'. Proc. Concerted Action on Impedance Tomography (CAIT). Workshop on The Electrode-Skin Interface in Electrical Impedance Tomography, Belfast, UK, Innov. TechnoL Biol. Med., 16, (2), pp. t03--112 JOSSINET, J., and TRILLAUD,C. (1992): 'Technical improvement of a dual frequency LIT tomograph for bioelectrical characterisation'. 9 Proc. 14th Ann. Int. Conf. of tEEE EMBS Society, 1712 KALIAY. K., and GUY, R. H. (1995): 'The electrical characteristics of human skin in vivo', Pharm. Res., 12 (11), pp. 1605-1613 KASTING. G. B. (1992): 'Theoretical models for iontophoretic delivery', Adv. Drug Delivery Rev., 9, pp. 177-199 KELLY H. W. (1985): 'Controlled-release transdermal drug delivery', Cutis, 35, pp. 204--207

Medical & Biological Engineering & Computing

November 1996

KLINGMAN,A. M. (1984): 'Skin permeability: dermatologic aspects

OLLMAR, S. ( 1995): ' Factors inflencing the electrical properties of the

of transdermal drug delivery', Am. Heart.', 108, (l), pp. 200207 KNEPP, V. M., HADGRAFT,J., and GuY R. H. (1987): 'Transdermal drug delivery: problems and possibilities', CRC Crit. Rev. Therapeutic Drug Carrier Syst., 4 (1), pp. 13-37 KONTTURI, K., MURTOM/iKI,L., HIRVONEN, J., PARONEN, P., and URTTI,A. (1993): 'Electrochemical characterization of human skin by impedance spectroscopy: The effect of penetration enhancers' Pharm. Res., 10, (3), pp. 381-385 KRENZKE,G., DUSTERHOFTT., and SCHWANKE,R. (1985): 'Vorrichtung zur Albeitung yon Herzpol'. German Patent DD 225 333 AI LAWLER, J. C., DAVISM. J., and GRIFEITH,E. C. (1960): 'Electrical characteristics of the skin', J. Invest. Dermatol., pp. 301-308 MCADAMS, E. T. (1989a): 'Effect of surface topography on the electrode-electrolyte interface impedance (i) the high frequency, small signal, interface impedance - - a review', Surface Topography, 2, pp. 107 -122 MCADAMS, E. T. (1989b): 'Effect of surface topography on the electrode-electrolyte interface impedance (ii) the low frequency, small signal, interface impedance - - a review', Surface Topography, 2, pp. 223-232 MCADAMS, E. T. (1990): 'Surface biomedical electrode technology', Int. Med. Device Diagnostic Ind., pp. 44-48 MCADAMS, E. T., and JOSSINET, J. (1990): 'Hydrogel electrodes in biD-signal recording'. Proc. 12th Ann. Int. Conf. of IEEE, Engineering in Medicine and Biology Society, Philadelphia, USA, pp. 1490-1491 MCADAMS, E. T., and JOSSINET, J. (1991a): 'The importance of electrode-skin impedance in high resolution electrocardiography', Automedica, 13, pp. 187-208 MCADAMS, E. T., and JOSSINET,J. (1991b): 'DC nonlinearity of the elcectrode-electrolyte interface impedance', Innov. Technol. Biol. Med., 12 (3), pp. 329-343 MCADAMS, E. T., and JOSSlNET,J. (1992): 'A physical interpretation of Schwan's limit current of linearity', Ann. Biomed. Eng., 20 (3), pp. 307-319 MCADAMS, E. T., and JOSSINET,J. 1994): 'A physical interpretation of Schwan's limit voltage of linearity' Med. Biol. Eng. Comput., 32, (3), pp. 126-130 MCADAMS, E. T., and JOSSINET, J. (1995): 'Tissue impedance: a historical overview', Physiol. Meas., 16, AI-AI3 MCADAMS, E. T., MCLAUGHLIN,J. A., and HOLDER, D. S. (1992a): 'Neurosensors: A review of some fundamental electrode parameters'. Satellite Symp. on Neuroscience and Technology, 14th Ann. Conf. of the IEEE Engineering in Medicine and Biology Society, Lyon, France, pp. 226-234 MCADAMS, E. T., HENRY, P., and ANDERSON, J. McC. (1992b): 'Optimal electrolytic chloriding of silver ink electrodes for use in electrical impedance tomography', Clin. Phys. Physiol. Meas., 13, supplement A, pp. 19-23 MCADAMS, E. T., MCLAUGHLIN,J. A., ANDERSONJ. McC. (1994a): 'Multi-electrode systems for electrical impedance tomography', Physiol. Meas., 15, A101-A106 MCADAMS,E. T., MCLAUGHLIN,J. A., BROWN,B. N., and McARDLE F. (1994b): 'The NIBEC LIT hamess' in HOLDER, D. (Ed.) 'Clinical and physiological applications of electrical impedance tomography' (UCL Press, London, 1993) Chap. 8, pp. 85-92 MCADAMS, E. T., LACKERMEIER,A., and JOSSINET,J. (1994c): 'AC impedance of the hydrogel-skin interface' 16th Ann. Int. Conf. of IEEE Engineering in Medicine and Biology Society, Baltimore. USA, pp. 870-871 NEWELL, J. C. (1995): 'Influence of electrode size on in-vivo impedance images'. Concerted Action on Impedance Tomography (CAIT). Workshop on The Reproducibility of in vivo Electrical Impedance Tomography Images, 12-14 May, Keele, UK NOLAN, L. M. A., CORISH,J., and CORRIGAN,O. I. (1993): 'Electrical properties of human stratum comeum and transdermal drug transport', J. Chem. Soc. Faraday Trans., 89, (15), pp 28392845 OH, S. Y., LEUNG,L., BOMMANNAN,D., GUY R. H., and POTTS,R. O. (1993): 'Effect of current, ionic strength and temperature on the electrical properties of skin', J. Controlled Release, 27, pp. l 15125

skin'. Proc. Concerted Action on Impedance Tomography (CAIT) Workshop on The Electrode-Skin Interface in Electrical Impedance Tomography, Belfast, UK, lnnov. Technol. Biol. Med., 16 (2), pp. 136-142 PAULSON K., BRECKON, W., and PIDCOCK, M. (1992): 'Electrode modelling in electrical impedance tomography', SIAM J. Appl. Math., 52, pp 1012-1022 PHIPPS, J. B., and GYORY,J. R. (1992): 'Transdermal ion migration', Adv. Drug Delivery Rev., 9 pp. 137-176 PIKAL, M. J. (1992): 'The role of electroosmotic flow in transdermal iontophoresis', ibid., 9, pp. 201-237 PRAUSNITZ,M. R., BOSE, V. G., L&S:ANGER, R., and WEAVER,J. C. (1993): 'Electroporation of mammalian skin: A mechanism to enhance transdermal drug delivery', Proc. Nat. Acad. Sci., 90, pp. 10504-10508 REILLY, J. P. (1992): 'Electrical stimulation and electropathology' (Cambridge University Press) REIN, H. (1924): 'Eperimentelle Studien fiber Elecktroendosmose an iiberlebender menschlicher', Haut. Z Biol., 81 pp. 125--140 RIGAUD,B., HAMZAOUI,L., GRANIE,M., CHAUVEAUN., MARTINEZ E., and MORUCCl, J. P. (1995): 'Electrode-electrolyte interface impedance estimation beyond 10 Hz for different electrolytes and electrodes'. Proc. Concerted Action on Impedance Tomography (CAIT) Workshop on The Electrode-Skin Interface in Electrical Impedance Tomography, Belfast, UK, Innov. Technol. Biol. Med., 16, (2), pp. 87-94 RISACHER, F., JOSSINET, J., MCADAMS, E. T., MANN, Y., and SCHMITT,M. (1993): 'Impedance plethysmography for the evaluation of pulse wave velocity in limbs', Med. Biol. Engl. Comput., 31, (3), pp. 318-322 RISACHER,F., (1995): 'Etude de la propagation de l'onde de pouls par pl~thsmographie d'impedance 61ectrique'. PhD Thesis, l'Universiti~ Claude Bernard, Lyon, France RIU, P. J., ROSELL, J., LOZANO,A., and PALLAS-ARENY,R. (1995): 'Multi-frequency static imaging in electrical impedance tomography: Part 1 Instrumentation requirements', Med. Biol. Eng. Comput.', 33, pp. 784-792 ROSELL, J., COLOMINAS,J., RIU, P., PALAS. R., and WEBSTERJ. G. (1988): 'Skin impedance from 1 Hz to 1 MHz', IEEE Trans., BME--35, pp. 649--651 ROSENDAL, T. (1943): 'Studies on the conducting properties of the human skin to direct current' Acta. Physiol. Scand., 5, pp. 130151 SALTER, D. C. (1981): 'A study of some electrical properties of normal and pathological skin in vivo'. D Phil Thesis, University of Oxford SCHMITT,O. H., and ALMASI,J. J. (1970): 'Electrode impedance and voltage offset as they affect efficacy and accuracy of VCG and ECG measurements'. Proc. Xlth Int. Vectorcardiography Symp., New York, pp. 245-253 SMITH, R. W. M. (1990): 'Design of a real time impedance imaging system for medical applications'. PhD Dissertation, University of Sheffield, pp. 45--49 TAKTAK, A., RECORD, P. M., GADD R., and ROLFE, P. (1995a): 'Interface impedance measurement of paediatric electrodes' Proc. Concerted Action on Impedance Tomography (CAIT) Workshop on The Electrode-Skin Interface in Electrical Impedance Tomography, Belfast, UK, Innov. Techno. Biol. Med., 16 (2), pp. 126135 TAKTAK, A., RECORD, P. M., GADD R., and ROLFE, P. (1995b): 'Practical factors in neonatal lung imaging using electrical impedance tomography', Med. Biol. Eng. Comput., 33, pp. 2022O5 WANG, M. (1994): 'Electrical impedance tomography on conducting Walled process vessels'. PhD Dissertations, UMIST, UK WEAVER J. C. (1993): 'Electroporation--a general phenomenon for manipulating cells and tissues', J. Cell. Biochem., 51, (4), pp. 426435 WIECHERS J. (1992): 'Use of chemical penetration enhancers in transdermal drug delivery--~ossibilities and difficulties' Acta Pharm. ?Cord. 4, (2), pp. 113---128 YAMAMOTO,T., and YAMAMOTO,Y. (1976): 'Electrical properties of the epidermal stratum corneum', Med. Biol. Eng., pp. 151-158

Medical & Biological Engineering & Computing

November 1996

407

YAMAMOTO,Y., and YAMAMOTO,T. (1978): 'Dispersion and correlation of the parameters for skin impedance', Med. BioL Eng. Comput., 16 pp. 592-594 YAMAMOTO,T., and YAMAMOTO,Y. (1981): 'Non-linear electrical properties of skin in the low frequency range', Eng. Comput., 19, pp. 302-310 YAMAMOTO, T., and YAMAMOTO, Y. and YOSHIDA A. (1986): 'Formative mechanisms of current concentration and breakdown phenomena dependent on direct current flow through the skin by a dry electrode', IEEE Trans, BME-33, (4), pp. 386404 YO~C.EY, T. J., WEBSTER,J. G., and TOMPKINS,W. J. (1985): 'Errors caused by contact impedance imaging'. Proc. 7th Ann. Int. Conf. of IEEE Engineering in Medicine and Biology Society, pp. 632-637 ZHOU, D. M., MCADAMS. E., LACKERMEIER,A., and JONES J. G. (1994): 'AC impedance of Ag-AgC1 reference electrodes for use in disposable biosensors'. Proc. 16th Ann. Int. Conf. of IEEE Engineering in Medicine and Biology Society, Baltimore, USA, pp. 832-833 ZHOU, D. M., JONESJ. G., MCADAMS.E., LACKERMEmR,A., (1995): 'Study of the electrode-electrolyte interface of screen printed AgAgC1 electrodes'. Proc. Concerted Action on Impedance Tomo-

408

graphy (CAIT) Workshop on The Electrode-Skin Interface in Electrical Impedance Tomography', Belfast, UK, Innov. Techno. Biol. Med., 16, (2), pp. 77-86

Author's biography Dr. Eric McAdams obtained his PhD at the University of Leeds while working at Leeds General Infirmary. He is head of the Bio-Medical Electrodes group at the Northern Ireland BioEngineering Centre (NIBEC), University of Ulster. He is a visiting lecturer on biomedical electrode technology to postgraduate students in several French universities in the Rhrne-Alpes region, and an associate researcher at the Institut Nationale de la Sant~ et de la Recherche M~dicale (INSERM), Unit 281, Lyon, France. His present areas of interest include electrode systems for cardiac mapping, electrical impedance imaging, impedance plethysmography, cardiac pacing and defibrillation, transcutaneous electrical nerve stimulation and fetal monitoring.

Medical & Biological Engineering & Computing

November 1996