Med. & Biol. Eng.& Comput., 1981,19. 247 249
Technical n o t e Keywords-- Time-domain analyser
High-speed time-domain spectrum analyser I Introduction LUNT (1975) has described the various difficulties encountered when attempting analysis of Doppler-shifted ultrasound signals from blood vessels using zero-crossing detection. There are a number of problems associated with
Crossing Detector
/
Generator
I
=
Fig. 1 Time-domain process
predictable use of this method, but the most fundamental problems relate to the fact that the zero-crossing detector attempts to reduce the Doppler-shift signal to a single voltage which may be displayed as a single-line trace upon an oscilloscope or a chart recorder. If artefact should be present, it may not he evident that this is so. Since in essence the Doppler-shift signal is a composite of all the signals backscattered by all the red cells insonated by the ultrasound beam, to produce a single-line display from such a complex signal it is necessary to either discard a great deal of the information contained therein, or to simply take an average of some parameter related to the signal.
Fig. 2 Time-domain analyser as a module for storage oscilloscope
in [Zero ', 11bit L ~ -~Cro ss ] [Det. ',C~ ~]1 I
TIME DOMAIN ANALYSER I
,11 C o r s s :[DetC .' ~ ~ ~ ~ ~ ',
{Ze,o,i ,10,, t_ 1 2-/Cross
(
L:
oJ?'l
'
3msb's Received 4th January 1980
Fig. 3 Block diagram of 2-channel analyser
0140-0118/81/020247+03s01,50/0 OIFMBE: 1981 Medical & Biological Engineering & Computing
March 1981
247
t-"~e|l l | II
!I
I"
iI
Jl 'liJ
l!!
i
*! !
J. J~
j~
I.
~3
2 Discussion A more satisfactory approach to the problem would be to preserve as much as possible of the information present within the signal for display. This may be accomplished in a relatively practical and artefact-free manner using spectral analysis, but normally with a considerable increase in the cost of the instrumentation required. However, an approximation to spectral analysis is possible (BAKF.Ret al., 1974; MOL, 1973) by operating on the unsmoothed output of a zero-crossing detector. The time T between zero crossing is determined, and the reciprocal taken (Fig. 1). This gives a signal, the amplitude of which may be used to indicate that the frequency equivalent to 1/T is present. To evaluate the practical performance of such an analyser a 2-channel analyser was constructed, with the individual channels muRiplexed on demand to a high-speed digital reciprocal generator (Fig. 2). The analyser was mounted along with a phase-domain processor (COGHLANand TAYLOe,, 1976) as a plug-in module for a standard laboratory storage display oscilloscope. A block diagram of the analyser is shown in Fig. 3. The circuit diagram is shown in
Fig. 4.
The time between the zero crossings of an input signal was counted in a binary counter, the output of which is connected to a reciprocal generator consisting of a r.o.m. programmed with the binary reciprocal function. The reciprocal generation was performed in this way very economically and at very high speed (less than 1 ~). The output of the reciprocal generator fed a d.a. converter, and the resulting analogue output connected to the Y-axis of the storage oscilloscope, whilst the oscilloscope Z-axis was simultaneously pulsed on.
,
sive and works well, particularly if both positive-going and negative-going zero crossings are used. It does, however, only perform an approximation to spectral analysis, and a full mathematical treatment is needed to determine the extent to which the technique is susceptible to artefact. When originally evaluated, insufficient separation of forward and reverse flow signals from the directional Doppler velocimeter resulted in noisy directional waveforms (zerocrossing detectors are very sensitive to crosstalk between forward and reverse channels--unless some form of interlocking is introduced, which precludes their simultaneous
display). The greatest potential use for this technique, however, rests in analysis of pulsed-Doppler velocimeter outputs. The channel outputs from a multichannel pulsed-Doppler may be multiplexed on demand to a single time-domain spectrum analyser, In this way a single analyser may suffice for analysis of a large number of signal channels. Given the relatively good signal/noise ratio of modern pulsed-Doppler velocimeters, and the very low probability that both forward and reverse flow velocities could simultaneously occur within a range cell (so allowing the use of interlocking zero-crossing detectors), this arrangement can be expected to perform well. B. A. COGHLAN M. G. TAYLOR R. G. GosuNo Physics Department, Guy's Hospital Medical School London SE1 9RT, England
Acknowledgment--We would like to thank Guy's Hospital Medical School, Medishield Corporation and the Department of Clinical Physics and Bioengineering, Guy's Hospital, for the necessary financial support.
i References
a
c
/9
d
Fig. 5 Two typical displays obtained from the time-domain spectrum analyser with corresponding sonaorams taken simultaneously The resulting display (Fig. 5) resembled a sonagram which had no grey scale. This form of analyser is inexpen-
BAra~R, D. W. et al. (1974) Prospects for quantitation of transcutaneous pulsed Doppler techniques in cardiology and peripheral vascular disease. In Cardiovascular applications of ultrasound (Ed. RL~n~r.tAN,R). North Holland, Amsterdam, chap. 8. COOrmAN, B. A. and TAYLOR, M. G. (1976) Directional Doppler techniques for detection of blood velocities. Ultrasound in Med.& BWI., 2, 181-188. LUNT, M. (1975) Accuracy and limitations of ultrasonic Doppler blood velocimetcr with zero crossing detector. Ultrasound in Med.& Biol., 2, 1-10. MOL, J. M. F. A. (1973) Doppler-haemato tachografisch onderzoek by Cerebrale circulatiestoornissen. Thesis. University of Utrecht.
Medical & Biological Engineering & Computing
March 1981
249