Microprocessor-based four-channel electronystagmography system

Med. & Biol. Eng. & Comput., 1980, 18, 104-108

Technical note Microprocessor-based four-channel electronystagmography system Keyword--Electronystagmography

I Introduction

TH~ microprocessor-based instrument, which is described here, is intended for the online computation of eye movement responses produced in vestibular function testing. In this connection the matter of interest is nystagmus, which is involuntary to-and-fro movement of the eyes. In recording eye movements, an electrical method, electrooculography (e.o.g.), is often used. This is also called electronystagmography (e.n.g.), when specifically used for the recording of nystagmus. Nystagmograms are usually analysed manually. Computer-based methods have also been developed (HERBERTS et al., 1968; TOLE and YOUNG, 1971; KTONAS et al., 1975; A~LUM et al., 1975) for this task, but the high cost has limited their number. Modern 1.s.i. (large-scale integration) technology, however, is now offering a chance to construct highly intelligent and flexible devices at a relatively low cost. As evidence of this, a microprocessorbased instrument for the analysis of nystagmus was presented two years ago (MICH~ELS and TOLE, 1977). In our project a microprocessor-based e.n.g, analyser is also under test and development. It is mainly intended to perform numerical analyses of e.n.g, recordings in conjunction with caloric tests. Although the movements of the eyes are usually conjugate, in some instances (NACLE and ANDERSON, 1972) disconjugate movements also occur in normal patients. A later work by KIMM and MACLEAN (1975) reports disconjugate eye movements in conjunction with central nervous system disorders, thus arguing in favour of

Fig.

1 E.N.G. processor N Y S T A G I

First received 13th November 1978 and in final form 1st May 1979 0 1 4 0 - 0 1 1 8 / 8 0 / 0 1 0 1 0 4 -I- 05 $01 95 0 / 0

9 IFMBE: 1980 104

separately recording the movements of the right and left eyes. The fourlchannel e.n.g, analyser (Fig. 1) is equipped with a numeric display. This approach, which is intended to be a step towards an automated multichannel processing of routine clinical as well as research e.n.g, data, still has problems, however, in selecting the most informative nystagmus parameters for output. Our instrument currently calculates 10 nystagmus parameters per channel. These include the total counts and total amplitudes of the fast phases of nystagmus in both directions, the average slow and fast phase velocities in both directions, and the value and point of time for the maximum slow-phase velocity corresponding with the dominant direction of nystagmus. 2 Schematic and basic operations

Fig. 2 shows the general construction of the instrument and also some essential features of its operation. The eye-movement signals are measured using skin electrodes placed around the eyes. The preamplifiers used so far are based on operational amplifier stages with a high differential and common-mode input impedance and a common-mode resolution ratio of more than 80 dB. The bandwidth is limited to 0-02-20 Hz to eliminate the electrode-contact-potential variations and the e.m.g. noise in the eye-movement signals. The gain is adjustable between 1500-17000 to compensate for variations in the measured signal. For the calibration of the preamplifiers there is an eye-movement meter on the front panel of the instrument. The calibration may also be performed automatically by the processor, which leads to an appropriate scaling of the digital sample values. The microprocessor system consists mainly of Motorola's M6800-components, such as a microprocessor unit (m.p.u.), three peripheral interface adapters (p.i.a.s) and a programmable timer (p.t.m.). The latter is used to generate the sampling commands for the 12-bit hybrid integrated circuit analogue-digital convertor. The eye-movement signals are each sampled sequentially at a frequency of 50Hz. Because the sampling frequency is practically in phase with the mains frequency (50 Hz), no mains interference is expected in the digital samples. For system programs there are 4 kbytes (1 byte = 8 bits) of ultraviolet erasable r.o.m, and for storage of data during processing 1 kbyte of r.a.m. The results of analysis, 10 nystagmus parameters per channel, may be examined in a numeric display after processing. The 'raw' e.n.g, signals, together with fastphase identification marks produced by the processor, are recorded on paper as a reference. A counter is included with the processor that may be used to automatically initiate and end the analysis.

Medical & Biological Engineering & Computing

January 1980

3 Description of the algorithm for detecting fast phases The analysis of nystagmograms is based on a proper differentiation of the fast and slow phases of nystagmus. The algorithm, which is designed to do this, actually detects saccades (fast phases of nystagmus or other abrupt eye movements) and assumes the remaining eye movements to represent the slow phases of nystagmus.

High values of instantaneous acceleration are characteristic of the start and end of saccadic eye movements (CARVENTER,1977). Therefore, points of peak acceleration are inspected by the algorithm (a simplified flow diagram is shown in Fig. 3) to detect the fast phases in the signal. The approximation of the acceleration is calculated using the same method (a simple digital f.i.r, filter) (MIcaA~,LS and TOLE, 1977).

OOg99N l!un

BU !SSa3OJdOJ3!UJ

x

~._

o~x ~E

E 0 U

o4 oo r

E

0 8 0=

N O

E co O

oq

E oE

-CL

I

r

-4"

C r

E C

C E

E C

O J~u

0 JC u

O .C u

0 tu

tO

Medical & Biological Engineering & Computing

January 1980

105

However, we also found it necessary to add a 'velocit~ test' in our algorithm in order to differentiate the peak accelerations corresponding with the start and end of saccadic eye movements. When five successive samples are multiplied by coefficients, as shown in Fig. 4, and

summed, an approximation to the instantaneous velocity is produced for the velocity test. The velocity test is performed after the acceleration has exceeded the manually adjustable threshold setting. If the speed during peak acceleration exceeds the former eye

i ,ort o,0-coov..,oo t scale the sample I

I calculate instantaneousvelocity 1 (store maxima)and acceter J yes

- - ~ ~

no j

~

c.or:::P -tlt~otrald t ~ s

,I

L

c ~ fnlt~CJn e~n~Jd [I

~

c'e~rl~ t; amnt:ne~

1 yes calculate s.p.velocity (store maximum and time of maximum) add s.p.period to corresp, total sum |

I

indicate f.phases (a flash of green or red I.ed.) prepare for next channel

! wait for c~command to take a n e w sample (200Hz clock )

f.p.: fast phase of nystegmus s.p.:slow phase of nystagmus

i

jump to start J (analysin9 next channel ) J(same channel,iS analysed J every 20ms) Fig, 3 Data flow during processing

106

Medical & Biological Engineering & Computing

January 1980

speed maxima (values from the time before the abrupt change in the velocity), then a saccade is assumed to have occurred. Saccades are classified according to their direction, which is termed either positive (eye movement directed to the right or up) or negative. If the velocity test fails, a 'flag' is set in the program. This causes the previously stored eye speed maxima to be cleared next time the acceleration falls below the threshold setting. The method assists in synchronising the program operation with the rhythm of the e.n.g, signal.

a

b

c

d

e

a=to-/.At b=to_3At c = t o - 2At d = t o - at e=t o

phase are determined for the calculation of the saccade amplitude and these fast- and slow-phase period lengths and their total sums. These points correspond (depending on the direction of the saccade) to local minimas and maximas in the 10 sample value shift register of the algorithm. At the beginning of each detected fast phase the average velocity in the previous slow phase is calculated by dividing the slow-phase amplitude with its period length. The value and the point of time corresponding with the maximum of this velocity is stored for output. Fig. 5 shows a sample of saccade identification executed by the algorithm. There are a few misinterpretations, but these are mainly concentrated in periods of the signal where the eye movements do not represent pure nystagmus. 4 Slow- and fast-phase velocities

The speed of the eye in the slow phase of nystagmus is commonly regarded as one of the most important measures of nystagmus intensity, since it is known to be Fig. 4 Coefficients for multiplication of sample values to calculate instantaneous velocity of eye : At is sampling period of 20 ms a to--4At b to--3At c to--2At d to-- At e

to

The eye-speed maxima are also cleared when the direction of the eye movement has changed, which normally occurs at the end of a nystagmus saccade. If no such reversal is found within 240 ms, the saccade period is automatically interrupted (240 ms should normally exceed the period for the longest possible saccades) (CARPENTER, 1977). There is also a 100 ms settling period after the end of each detected saccade in which no new saccade is recognised. This 100 ms period is less than the minimum refractory period of saccadic movements (CARPENTER, 1977). At the beginning and end of each saccade period, the exact starting and finishing points of the assumed fast

~-rrrr~-r,:,, ,w'w-r-r% fast phase identification

.... - a - r , - w % - ~

I10"

2s ,

,

total amptitude of negative fast phases

Fig. 5 Identification of fast phases and cumulative sum of amplitudes of negative direction fast phases M ed i c a l & Biological Engineering & C o m p u t i n g ~

I10 " ~S I

i

f

Fig. 6 Cumulative slow-phase position (c.s.p.p.) produced by extracting fast phases from the signal in a direct relationship with the cupular displacement and vestibular stimulus (HeNRIKSSON, 1955; 1956). Instantaneous slow-phase velocities may be computed using temporal variations of the cumulative slow-phase eye position (c.s.p.p.). The c.s.p.p, may be produced by a n algorithm that detects the saccades and removes them from the eye-m0vement signal. Fig. 6 shows a typical c.s.p.p, recording that was produced using 10 bit digitalanalogue convertor in the development stage of our e.n.g, algorithms. For simplicity, the maximum slow-phase velocity is detected, however, by calculating the average eye speeds in individual slow phases, as already explained. The average slow-phase velocities are determined using the total amplitudes of the saccades and the total sums of the positive and negative slow-phase periods. This is possible since the total amplitude of the saccades must equal the total amplitude of the slow phases. The average fast-phase velocities are also included in the set of ten output parameters of our instrument, although this information is known to be less relevant when analysing vestibular nystagmus (MEVILL JONES, 1973). These approximations are computed using the total sums of saccades and the total sums of the fast-phase periods. January 198,0

107

The computed values are somewhat diminished due to the low sampling frequency and the lowpass filtering used in conjunction with the preamplifiers.

CARPENTER,R. H. S. (1977) Movement of the eyes, Pion

Ltd., London. HENRIKSSON, N. G. (1955) The correlation between the speed of the eye in the slow phase of nystagrnus and vestibular stimulus. Acta Oto-Laryng, 45, 120-136. 5 Discussion HENRIKSSON,N. G. (1956) Speed of slow component and At present the methods described here are still under duration in caloric nystagmus. Acta Oto-Laryng., test and development. The purpose of the project is to Suppl. 125, 3-29. conclude with a compact data-processing tool for research as well as clinical nystagmography. A special interest will HERBERTS, G., ABRAHAMSSON,S., EINARSSON,S., HOFFMANN, H. and LINDER, P. (1968) Computer analysis of be concerned with disconjugate eye-movement comelectronystagmographic data. Acta Oto-Laryng., 65, parisons. Future investigations will also concentrate on 200--208. developing output methods using a printer and digitalKIMM, J. and MCLEAN, J. B. (1975) Disconjugate eye analogue convertors in addition to a numeric display. movements during electronystagmographic testing in T. RAHKO patients with known central nervous system lesion. Ann. Otol., 84, 368-373. Department of Audiology KTONAS, P., WEINTRAUB,B., SMITH, J. and BLACK, F. O. Tampere Central Hospital (1975) Computer aided nystagmus analysis. Comput. Tampere, Finland P. KARMA Programs Biomed., 5, 153-157. MELVILL JONES, G. (1973) Nystagmography--a useful Institute of Clinical Sciences Medical Faculty tool in basic and applied investigations. A G A R D ConTampere University ference on use of nystagmography in aviation medicine, Tampere, Finland T. TORIKKA preprint 128, A12-1-A12-13. J. MALMIVUO MICHAELS, D. L. and TOLE, J. R. (1977) A microprocessor-based instrument for nystagmus analysis. Proc. Laboratory of Biomedical Engineering IEEE, 65, 730-735. Tampere University of Technology NAGLE, D. W. and ANDERSON,R. G. (1972) Method for Tampere, Finland establishing electronystagmograms for normal humans subjected to caloric stimulation. Laryngoscope, 82, References 1671-1702. ALLUM, J. H. J., TOLE, J. R. and WEISS, A. D. (1975) MITNYS I I - - A digital program for on-line analysis of TOLE, J. R. and YOUNG, L. R. (1971) MITNYS. A hybrid program for on-line analysis Of nystagmus. nystagmus. IEEE Trans., BME-22, 196-202. Aerosp. Med., 42, 508-511.

108

Medical & Biological Engineering & Computing

January 1980