404
Electroencephalography and Clinical Neurophysiology Elsevier Publishing Company, Amsterdam - Printed in The Netherlands
TECHNICAL
CONTRIBUTION
AUTOMATIC SAMPLING A N D AVERAGING OF ELECTROMYOGRAPHIC
UNIT POTENTIALS
A. H. LANG, P. NURKKANEN 1 AND K. M. VAAHTORANTA I)epartment ot Clinical Neurophysiology, Unicersio' Central Ho,spitat Of Turku, 20520 ]urku 52 ~l'7tdand)
(Accepted for publication: April t5, 1971)
Various automatic analysing methods have been developed in order to obtain a rapid and quantitative evaluation of EMG signals (Lenman and Ritchie 1970), but by these methods the primary information achieved by photographing the different motor unit potentials (MUPs) and analysing their parameters separately is lost. The automatic methods give no information as to the spatial distribution of the MUPs in muscle. We have developed an automatic sampling method which in principle gives the same statistical information on the parameters of M U P s as the photographic method. l u r!hcrmore, the method can be used in analysing the average spatial distribution of unit potentials. Only the principles of this mclhoci a~d some illustrative results are presented here. "1he rccordi,~l dezicc. The amplified EMG signal triggers the recording-averaging device at the moment the signal
fulfils the amplitude and peak conditions (see below). The signal is delayed in a delay circuit, in which it is not distorted in shape and the amplitude changes are linear. The delay device makes it possible to record also the part of the signal that occurs before the moment of triggering. The recordingaveraging unit (an analog-digital computer) samples the signal at selected time intervals. The sampling is repeated as many times as necessary to gel a satisfactory average record of the EMG signals. The trigger det:ice. "The trigger input signal is divided into two channels, the discriminator or threshold channel and thc peak channel (Fig. 1). The discriminator gives a logical "1 ""output all the time the signal is over the threshold. In the peak channel the signal is differentiated and the output crosses the zero line at both the positive and negative peaks of
. . . . .
DIFFERENTIATOR
ZERO CROSSING
.~H...ES_H_OL.O.
-.
SIGNAL !
TRIGGER INPUT
I.
! ! !
J DISCRIMINATOR
1
I AND
DISC
TIME
I
OUTPUT
TRIGGER OUTPUT DIFF SIGNAt
ZERO CROSSING
! !
TRIGGER OUTPUT
Fig. 1. Block diagram and working principle of the trigger device. Department of Physics, University ofTurku, lurku, Finland. Electroenccph. eli,. Neurophv:~io/., 1971, 31:404~.06
405
AUTOMATIC EMG ANALYSIS f,
A
I
B
A
Fig. 2. Two original E M G records representing activity in m. interosseus dorsalis by slight (A) and strong (B) muscle contraction. Below, the corresponding averaged MUPs, each representing the sum of 400 signals. Calibrations: 0.5 mV, 1 sec (upper records) and 10 msec (averaged records). the signal. When we select only the zero crossing of the negative slope of the differentiated signal, we get a logical "1" in the zero crossing circuit output only when the positive peak of the signal occurs. The A N D gate output is in " l ' state only when both of its inputs (the discriminator output and the zero crossing output) are true, i.e., "1". In this way we get a trigger signal for the averaging device which coincides with the positive peak of the trigger input signal. The threshold discriminator rejects peaks that are too low and a filter (not present in Fig. 1 ) in the peak channel selects the signal to be differentiated ac-ording to its frequency content. If only an amplitude triggering condition were used, a small signal would reach the threshold level later than a large one despite equal rise time. This would cause a time
A
jitter between the trigger and the peak point which would make the averaged potential lower and broader than when also using the peak of the signal as a trigger condition. Examples. Fig. 2 shows two averaged records together with the original EMG records of the same activity, which were photographed simultaneously and from which the averaged records were taken. Only a short run of a longer EMG record is shown, but the force of the muscle contraction and the EMG pattern did not change throughout the run. Despite the fact that the second averaged record was extracted from EMG activity under strong contraction with interference pattern (B), the shape of the averaged record. is about the same as the averaged record of the activity of mixed pattern (A). Fig. 3 shows two averaged records, the
6'
i
Fig. 3. Averaged M U P s from biceps brachii (A) and m. frontalis (B). Four hundred signals summated. Calibrations: 100/~V, 10 msec.
Electroenceph. clin. Neurophysiol., 1971, 31 : 404-406
406
,4
A.n.
LANG et al.
6'
Fig. 4. Average synchronization of E M G signals at different recording points. All three averaged M U Ps were recorded from a multi-electrode with constant inter-electrode distance. The first record IAI shows an average potential of the trigger signals. The other records are averages of the potentials recorded at a distance of 1.5 m m (B) and 3 m m (C} from the first recording point. The delay between triggering and initiation of sampling was the same for all records. Note the decrease of amplitude and slight increase of duration of the averaged potentials with increasing inter-electrode distance. M. vastus medialis. Three hundred and fifty signals summated. Calibrations: 0.5 mV, 10 msec.
larger (A) from m. biceps brachii and the second (B) from m. frontalis. The E M G activity can also be derived from two or more different recording points within the same muscle and averaged in different sections of the averaging device simultaneously. In this application the signal from one recording site only is used as the trigger signal. The ratio between the amplitudes of the averaged "trigger" signals and the "'synchronous" signals gives a picture of the time relations or degree of synchrony of E M G activities at the two recording points. Fig. 4 demonstrates a result of this recording principle. DISCUSSION The advantages of the presented sampling and averaging technique are its high spee and the fact that the result can be presented in familiar physiological parameters such as duration, amplitude and shape of the M U P , and that no possibility of observer bias exists. A special advantage is that the averaged record is given in digital form, so that a mathematical and statistical analysis can later be made, for instance from a punch tape record. A drawback is the lack of information on the variation of parameters of the single unit potentials, but this is a c o m m o n disadvantage of automatic analysing methods, because most of them give only averaged, statistical information of the recorded signals. A separate problem is how to choose the quantitative values of frequency filtering and amplitude discrimination of the trigger device in order to get the most reliable diagnostic results in various pathological conditions. Ihis topic will be treated in a separate report.
SUMMARY All amomatic method of samptmg and recording motor unit potentials (MUPs) is presented by which a large sample of M U P s can be rapidly recorded and summated to form a kind of average MUP. The method gives also some picture of the average spatial distribution of M U P s in a muscle. The averaged result is presented in digital form to be used for data analysis. RESUME PRISE AUTOMATIQUE D'ECHANTILLONS ET M O Y E N N A G E DES P O T E N T I E L S E L E C T R O M Y O GRAPHIQUES UNITAIRES Les auteurs decrivent une m6thode automatique d'echantillonnage et d'enregistrement des potentiels unitaires moteurs (PUMs) grS~ce g laquelle un grand hombre d'6chantillons de P U M s peut ~tre rapidement enregistr6 et somm6, de fa~on ~t obtenir un P U M moyen. Cette m6thode donne 6galement un aspect de la distribution spatiale moyenne des P U M s dans un muscle. Le rbsultat moyenn6 est pr6sent6 sous forme digitalis6e pour ~tre utilis6 pour l'analyse des donn6es.
REFERENCE LENMAN, J. A. R. and RrrcnlE; A. E. Clinical etectromyography. Pitman, Bath, G. B., 1970, 175 p.
Rt:/erence: LANG, A. H., NURKKANEN, P. and VAAHTORANTA,K. M. Automatic sampling and averaging of electromyographic unit potentials. Electroenceph. clin. Neurophysiol., 1971, 31 : 404-406.