- Email: [email protected]
Control strategies of the elbow antagonist muscle pair during two types of increasing isometric contractions
Journal of Electromyography and Vol. 3. No. 1, pp 3-O 0 1993 Butterworth-Heinemann
Kinesiology
Ltd
Control Strategies of the Elbow Antagonist Muscle Pair During Two Types of Increasing Isometric Contractions J. H. Sanchez, M. Solomonow, Bioengineering
Laboratory,
Department
R. V. Baratta, and R. D’Ambrosia
of Orthopaedic
New Orleans, Louisiana,
Surgery,
Louisiana State University,
U.S.A.
Summary: The purpose of this study was to determine if differences exist between the control strategies of the elbow flexor and extensor muscles performing stepwise and linearly increasing isometric contractions, and to determine their control strategy when active as antagonists to each other. The electromyogram (EMG) from the biceps brachii and triceps brachii were recorded during stepwise and linearly increasing contractions in flexion and extension. The power density spectrum of the EMG was determined and the median frequency (MF) for each spectrum was calculated for assessment of changes in the average conduction velocity, which reflects motor unit recrnitment and derecruitment, and thereby the control strategy of the muscle. The results suggest that differences exist in the control strategies employed by a single muscle during stepwise and linearly increasing contractions. Furthermore, the antagonist triceps recruits motor units up to a higher force level during stepwise contractions than during linearly increasing contractions. The antagonist biceps derecruits motor units up to a higher force level during linearly increasing contractions than during stepwise contractions. Key Words: Electromyogram-Coactivation-Motor units-Recruitment.
In order to improve existing functional neuromuscular stimulation (FNS) systems, it is important to understand the firing rate and motor unit recruitment control strategies of muscles and their antagonist pairs during different types of contractions. Establishing relationships between the antagonist pair control strategies for the contraction type would allow for more efficient stimulation of the muscles, which represents less energy expenditure and longer muscle endurance while maintaining joint stability. The overall result of the implementation of such a relationship in an FNS system would be refined limb control (17). Accepted Address Solomonow University LA 70112.
Lindstrom et al. (13) developed a mathematical model for the power density spectrum that relates the spectrum to the conduction velocity of action potentials along the muscle fibers. Many investigators have noted a shift in the power density spectrum toward the low frequencies during sustained contraction (3,8,9,12-16,21). Lindstrom et al. (13) demonstrated that the shift in the power density spectrum represents a decrease in the average conduction velocity of the active muscle fibers. ArendtNielsen et al. (1) and Bigland-Ritchie et al. (3), among others, provided experimental evidence of the correlation between conduction velocity and the spectral parameters of the myoelectric signal. Stulen et al. (21) determined that the median frequency of the power density spectrum had a linear relationship to the conduction velocity, and that it was more immune to noise than other spectral pa-
November 19, 1992. correspondence and reprint requests to Prof. M. at Department of Orthopaedics, Louisiana State Medical Center, 2025 Gravier Street, New Orleans, U.S.A.
33
J. H. SANCHEZ ET AL.
34
rameters. Using different control strategies to stimulate cat muscles, Solomonow et al. (17,19) studied the spectrum median frequency, observing that the median frequency was affected only by recruitment, and not by changes in the firing rate (19). Bilodeau et al. (4) tracked the median and the mean frequencies of the spectrum obtained from the triceps brachii and the anconeus during voluntary contractions in an attempt to determine if the control strategies of the muscles change during stepwise and linearly increasing contractions. During stepwise contractions, the spectral parameters increased with force up to 20% of the maximal force level, and then either leveled off or decreased for the higher force levels. During linearly increasing contractions, the spectral parameters increased up to 80% maximal voluntary contraction (MVC) and then leveled off or decreased for the higher force levels. Bilodeau et al. (4) concluded that the type of contraction performed (stepwise or linearly increasing) has a major role in the behavior of the median frequency patterns, and thus, the control strategy of the muscles. Zhou et al. (22) studied the median frequency patterns as a function of time for the antagonist muscle pair of the elbow while maintaining a contraction force of 60% MVC. They studied the biceps and triceps brachii in flexion and extension and observed that in a given direction of contraction (flexion or extension), the antagonist muscle had higher median frequency values than the agonist. The objectives of this study are to confirm that a single muscle, acting as agonist, can change its motor unit recruitment strategy during stepwise and linearly increasing contraction; to determine the control strategies it employs when acting as an antagonist; and to delineate any differences in the strategies when the muscle acts as an antagonist in the two types of contractions mentioned above. METHODS AND PROCEDURES
ric flexion and extension. In order to allow conditions to be as close to normal as possible, no restraints were provided for the shoulder or elbow. The wrist was maintained in a neutral position (half way between pronation and supination). Instrumentation A Lebow bidirectional load cell was connected through a metal rod to a cuff around the forearm. A coupling amplifier was used to include the cell in a Wheatstone bridge configuration and to calibrate offset voltages and gain. Two silver/silver chloride pregelled surface electrodes of 1.5 cm in diameter, with a center to center interelectrode distance of 2.5 cm, were applied over the belly of the biceps brachii. A second pair of electrodes was placed over the belly of the triceps brachii. A fifth electrode was attached to unrelated tissue in the same arm, serving as ground. The myoelectric signal from each muscle was detected by a differential amplifier with a gain of up to 200,000, a common mode rejection ratio (CMRR) of 100 dB, and then band-passed in a range of l& 550 Hz. The signal from the load cell and the myoelectric signals from the biceps and triceps muscles were then digitized using a data acquisition card and stored in an IBM 386 computer at a sampling rate of 1,024 Hz for further processing. Two B & K 1477 dual trace oscilloscopes were used, one to display the raw myoelectric signals from each muscle in each channel and one to display the force exerted by the subject and the “target” force to be matched by the subject. A HewlettPackard 3964A FM tape recorder was used to provide the ramp function to be tracked by the subject during the linearly increasing contractions. The ramp function simulated a linearly increasing contraction from 0 to 100% MVC in 3 s.
Subjects
Protocol
Eleven men and one woman were used as subjects in this study, ranging in age from 19 to 46 years. The subjects were seated in a specially designed chair with their elbows flexed to 90” and their upper arms parallel to their sides. A plastic cuff was attached to their forearms, and the cuff was rigidly connected to a bidirectional load cell by a metal rod, which in turn was attached to the chair. This arrangenient allowed the subject to perform isomet-
The subjects were seated on the chair, and their forearms were strapped to the load cell as described previously. The area on the skin where electrodes were to be placed was thoroughly cleansed and slightly abraded with an alcohol pad in order to remove any skin oils, and thus reduce skin impedance. The experiment was divided into two sections, flexion and extension, and these two were further
J Electromyogr
Kinesiol. Vol. 3, No. I. 1993
CONTROL
STRATEGIES
divided into stepwise and linearly increasing contractions. The subjects were trained to obtain their true MVC first, and adjustments were made in the oscilloscope to range full scale from 0 to 100% MVC levels. Beginning the flexion part of the experiment, the MVC force during flexion was determined and eleven steps corresponding to 0 through 100% MVC force levels increasing in steps of 10% MVC were marked. During stepwise flexion the data for each of the eleven target forces was collected in a random order, in order to minimize the possibility of a systematic error introduced by fatigue. Once the subject’s force matched the target line, the force was maintained constant at that level for 2 s while the data were sampled. A 5-min rest period was strictly observed between consecutive trials. During linearly increasing flexion, the subject was to follow as close as possible a ramp signal displayed on the oscilloscope which corresponded to an increase in force from 0 to 100% MVC over 3 s, as described previously. Five linearly increasing trials were recorded for each subject. Five-min rest periods were observed between each trial. The same procedure was used for the stepwise and linearly increasing extension part of the experiment, after proper determination and calibration of the maximal voluntary contraction in extension. It was asserted that cross-talk between flexors and extensors did not exist by monitoring the EMG of one muscle plotted against the EMG of the other to show that constant phase was absent (18).
OF ANTAGONIST
MUSCLES
35
Analysis The stored myoelectric signals from the stepwise contractions were divided into four segments of 0.5 s. Each segment was multiplied by a Tukey window (10% taper) and zero padded by 5 12 points to obtain a resolution of 1 Hz. A fast Fourier transform (FFT) was applied, and an estimate of the power density spectrum was obtained. The median frequency (MF) was estimated by finding the frequency point which divided the power density spectrum into two parts of equal area. To process the myoelectric signals of linearly increased contraction, each trial was divided into overlapping segments of 128 d points. Each of the 128-point segments had 50% overlap with the previous segment. Each segment was multiplied by a Tukey window (10% taper) and zero padded by 128 points to obtain a resolution of 4 Hz. Such a short interval was necessary to minimize the nonstationary effect of the EMG caused by the changing force during the course of each epoch. Then, as in stepwise contractions, FFT was performed on each segment, followed by MF calculation. RESULTS Figure 1 shows a sample of the data obtained for a subject during a stepwise flexion trial at a constant force of 50% MVC maintained for 2 s. Figure 2 shows a sample trial obtained for a subject perform-
FIG. 1. Stepwise flexion at 50% maximal voluntary contraction (MVC). Force is displayed in the top section, and the biceps and triceps myoelectric signals are displayed in the middle and bottom sections. The force is expressed in percentage of MVC, and the myoelectric signals in millivolts. Note the scaling difference in millivolts on the EMG axis for the biceps and triceps.
J Electrom_vogr Kinaid.
Vol. 3. No. 1. 1993
J. H. SANCHEZ
36
ET AL.
100 80 66 !O 20 0 3.0
” ‘- II’
IHcnror EN6
ing a linearly increasing flexion. The force increases linearly from 0 to 100% MVC in 3 s. The mean and standard deviation values of the MF at each force level obtained from the pooled data of all the subjects while performing stepwise and linearly increasing contractions are shown in Figs. 3 and 4, respectively. A second order polynomial model for each datum set was developed, which related median frequency to force level. The peak of each model was determined by finding where the polynomial derivative equated to zero and was compared with the corresponding peak from the averaged data. The peaks obtained statistically were slightly higher than the ones obtained from the average of the data. The statistically obtained peaks could be higher because a continuous model was tit to discrete data. When the biceps was performing flexion in stepwise contractions the median frequency increased up to 50% MVC and then decreased at a slow rate, as observed in Fig. 3. When the biceps performed linearly increasing flexion, the median frequency increased up to 70% MVC and then decreased at a slow rate. When the triceps performed stepwise extension, the median frequency increased up to 50% MVC and then decreased at a slow rate. In linearly increasing extension, the triceps median frequency increased up to 70% MVC and then remained constant for the rest of the force range. When acting as antagonist during stepwise increasing extension (Fig. 3), the biceps exhibits a nearly exponential decrease in the MF up to 50% MVC, after which it levels off up to 100% MVC.
J Electromyogr
Kinesiol, Vol. 3, No. I, 1993
FIG. 2. An example of a trial with linearly increasing flexion used for analysis of the EMG power spectra. A straight line was drawn from 0 to 100% MVC to assess the quality of the linear increase of force.
The triceps, however, shows that, when acting as antagonist during elbow flexion, the MF increases up to 90% MVC, with a sharp drop thereafter. The antagonist activity during linearly increasing contraction is shown in Fig. 4. The biceps exhibits a short increase in MF from 0 to 20% MVC and a gradual decrease thereafter, for the complete force range up to 100% MVC. The triceps, as antagonist, exhibits a gradual increase in MF up to 70% of the force range, and a gradual decline thereafter. DISCUSSION The results suggest that there is a difference in the control strategy of a muscle performing two types of contractions. These results agree in general with those obtained by Bilodeau et al. (4) who studied the EMG power spectra during stepwise and linearly increasing contractions in the triceps brachii and the anconeus. They reported an increase in the median frequency up to 20% MVC during stepwise contractions and an increase up to 80% MVC during linearly increasing contractions, and a decrease thereafter. Their report suggests that there is a different control strategy for stepwise and linearly increasing contractions. The numbers obtained in the present study differ from those in Bilodeau’s work. However, in a previous study conducted in this laboratory using intramuscular electrodes, Zoutman (23) reported an increase in the median frequency of the biceps in stepwise flexion up to 50% MVC, and either a slow decrease or constant median frequency thereafter. Zoutman
CONTROL
STRATEGIES
0 lO20304060607030eOl00 Force (%)
Force (%)
A
MFR (Hz)
I
36. 30. 76. 70. 66. 60. 60’
* ’ ’ ’ ’ ’ ’ ’ 0 10 20304060607030Wl00
Force (%)
C FIG. 3. Mean and level for the median wise contractions. tagonist (b), triceps (d) are shown.
’
*’
660 lO203040608070600010 Force (%)
D values versus force standard deviation _ frequencies of subjects performing stepThe biceps as agonist (a), biceps as anas agonist (c), and triceps as antagonist
OF ANTAGONIST
37
MUSCLES
muscles when acting as antagonists during sustained constant force flexion and extension. They monitored the median frequencies of the power density spectra from the biceps and triceps muscles, while maintaining a contraction force of 60% MVC. Zhou et al. observed that during both flexors and extensors acting as antagonists had median frequencies higher than the agonist’s median frequencies, and that these values decreased while the contraction was maintained. In the present study, during both stepwise and linearly increasing extension, the antagonist’s median frequencies were higher than the agonist’s median frequencies. During stepwise and linearly increasing flexion, however, the agonist’s median frequencies were slightly higher than the antagonist’s median frequencies. These differences observed between Zhou’s work (22) and the present study could be a result of the factors affecting the detection of the conduction velocity. The fact that the results are different does not suggest that either experiment ,&AFR
(Hz)
,rOwR
n-31
III1
(Hz)
I
n-40 130.
cm-
120. 30.
110. loO*
70.
00.
used a procedure similar to the one used in the experiment reported here. Further studies with a fixed procedure for determining MVC are necessary to better understand this behavior and the cause for these differences. Prior studies (4,23) suggest that a muscle uses different control strategies for stepwise and linearly increasing contractions, as mentioned previously. The difference in the control strategies suggested by the results of the present report may explain the differences in control strategies reported for the same muscle in several studies from different groups (57,10,11). Most of these studies do not specify the type of contractions performed and most likely the differences in protocol for each experiment are a key factor for these differences. It is essential to study muscles under different types of contractions using the same experimental procedures in order to obtain a better understanding of the causes for the differences in control strategies. Zhou et al. (22) studied the median frequency and mean absolute value (MAV) changes of the elbow
60. 60,
30. .
.
.
.
.
.
.
,(
70’
.
0 10 2030406060703080100
.
.
.
.
.
.
,
.
.,
.
.
J
0 10 2030406~60703080100
Force (a;)
Force (a;)
B
A MFR (Hz) 86
.ndO s-10
0 10 203040606070308010(
36’
’
’
-
.
*
A
*
0 102030406080703090lCNJ
Force (%)
Force (%I
D
C
FIG. 4. Mean and standard deviation values versus force level for the median frequencies of subjects performing linearly increasing contractions. The biceps as agonist (a), biceps as antagonist (b), triceps as agonist (c), and triceps as antagonist (d) are shown.
J Electrom_vogr Kinesiol.
Vol. 3. No. I. I!293
38
J. H. SANCHEZ
has wrong results. In addition, the populations of subjects used for each experiment could also have differences among themselves. For example, the EMG recorded with surface electrodes is subjected to several factors that may induce a low-pass filtering effect. Skin thickness which is greater for men than women, fatty tissue between skin and muscle, electrode diameter, and interelectrode distance are some of these factors. Another issue which may affect the value of the MF is the exact location of the electrodes over the muscle group, and their individual predominant fiber type composition. Although most of the EMG was recorded from the biceps, some signal came from the brachialis and coracobrachialis. Each of these muscles may have slightly or largely different predominant fiber types, and the exact location of the electrodes over the flexors may emphasize or attenuate the EMG of fast twitch fibers, for example, to yield somewhat higher or lower MF. This, however, explains variabilities in MF values from subject to subject and from study to study, but it does not impact on the fact that the overall control strategy changed according to the specific function performed by the muscle, e.g., stepwise or linearly increasing force contractions. In the present study the median frequency of the biceps in extension was found to decrease with increasing force level for both linearly increasing and stepwise contractions. This behavior suggests that derecruitment of motor units takes place as the net extension force is increased. In contrast, the median frequency of the triceps in flexion was found to increase with force level on both linearly increasing and stepwise contractions. These results suggest that the,triceps and the biceps as antagonists have completely different control strategies. Solomonow et al. (20) studied the normalized MAV of the EMG from the biceps and triceps muscles as both agonists and antagonists in a stepwise increasing contraction. They observed that the MAV from each muscle, acting as either agonist or antagonist, increases with force level. They suggested that the increase in activity with force level observed in the antagonists, stabilized the joint and augmented the contact area of the joint, thus reducing stress concentrations on the articular surfaces (2,18,20). The electrical activity (MAV) in a muscle increases as a result of two factors: an increase in the number of active motor units and an increase in the average firing rate of the active motor units. The
J Electmnyogr
Kinesiol,
Vol. 3. No. 1, 1993
ET AL. MAV is not a simple summation of the effect of firing rate or recruitment, but some complicated type of summation. The median frequency changes when motor units are recruited or derecruited, and stays relatively unaffected whether the firing rate increases or decreases (19). According to Solomonow’s observations (20), the MAV increases with force in both the biceps and triceps acting as antagonists. No distinction can be made as to which of the two processes for increasing force (if not both) is taking place. However, since the median frequency reflects recruitment and derecruitment of motor units, combining these two parameters (median frequency and MAV), assessment of the individual contributions of both recruitment and rate coding (change in firing rate) can be made. Based on this argument, some suggestions can be made on the observed behavior of the biceps and triceps as antagonists. The median frequency for the biceps in extension was noted to decrease with increasing force level, whereas the MAV was noted to increase with force level (20). The median frequency and the MAV increased with force level for the triceps in flexion, as in the case of the biceps and triceps each acting as agonist. In this study, it seems that the flexor in extension mostly uses increase in rate coding to generate the necessary stabilizing force while derecruiting motor units, whereas the triceps in flexion uses both rate coding and recruitment to generate the stabilizing force. Comparing the control strategy of the biceps in extension and triceps in flexion for linearly increasing and stepwise contractions is interesting. In both muscles, the median frequency patterns changed with contraction type. Derecruitment occurred in the biceps throughout the whole force range for linearly increasing extension, and down to 60% MVC in stepwise extension. In the triceps in flexion, recruitment occurred up to 70% MVC during linearly increasing contractions, and up to 90% MVC during stepwise contractions. According to these results, it seems that both the biceps in extension and the triceps in flexion use different strategies when performing stepwise and linearly increasing contractions. The biceps in extension seems to rely more on rate coding than recruitment to generate the necessary force for stabilizing the joint, as stated previously. During linearly increasing extension, the biceps derecruits motor units slowly throughout the force range to achieve a smoother extension, thus showing a longer derecruitment as compared with stepwise extension. To
CONTROL
STRATEGIES
generate the necessary force, it is possible that for the higher force levels, it relies on increasing the firing rate of those motor units that remain active. In the case of the triceps in flexion, it appears to rely on both rate coding and recruitment to generate the stabilizing force, as previously suggested. When performing stepwise flexion, recruitment takes place up to 90% MVC, suggesting that the motor units are recruited slowly, as compared with linearly increasing flexion, where recruitment occurs up to 70% MVC. This behavior represents a reversal of the results obtained for the agonists in linearly increasing and stepwise contractions, where motor units were recruited up to higher force levels during linearly increasing contractions than during stepwise contractions. These differences can not be explained with the available literature, and more work is necessary to understand the behavior of the antagonists during linearly increasing and stepwise contractions. CONCLUSIONS From the data and its analysis, the following conclusions can be made: (a) Differences exist in the control strategies of muscles performing as agonists during stepwise and linearly increasing contractions. The biceps in stepwise flexion recruited motor units up to 50% MVC and in linearly increasing contraction up to 70% MVC. Similarly, triceps motor units were recruited up to 50% MVC in stepwise contraction, and up to 70% MVC in linearly increasing contraction. For the higher force levels, the muscles probably depended on rate coding to generate extra force. (b) The MAV reflects both change in firing rate and recruitment. The median frequency is mostly affected by motor unit recruitment. Using both the median frequency and the MAV, an assessment can be made of the role of rate coding in contributing to the control strategy. (c) The biceps in extension derecruits motor units as force level increases. (d) The triceps in flexion recruits more motor units as force level increases. (e) There are differences in the control strategy of the biceps in extension during stepwise and linearly increasing force contractions. Derecruitment is slower during linearly increasing extension than during stepwise extension. (f) There are differences in the control strategy of the triceps in flexion during stepwise and linearly increasing contractions. Recruitment is slower during stepwise flexion than during linearly increasing flexion.
OF ANTAGONIST
39
MUSCLES
Acknowledgment: This work was supported by Grants BCS 9207007 and BCS 9006639 from the National Science Foundation.
1. Arendt-Nielsen L, Forster A, Mills KR: EMG power spectral shift and muscle fibre conduction velocity during human muscle fatigue. J Physiol 353:54, 1984. 2. Baratta R, Solomonow M, Zhou BH, Letson D. Chuinard R, D’Ambrosia R: Muscular coactivation. The role of the antagonist musculature in maintaining knee stability. Am J Sports Med 16: 113-122, 1988. 3. Bigland-Ritchie B, Donovan EF. Roussos CS: Conduction velocity and EMG power spectrum changes in fatigue of sustained maximal efforts. J Appl PhysioL 51: 1300-1305, 1981. 4. Bilodeau M, Arsenault AB, Gravel D, Bourbonnais D, Kemp F: Comparison of EMG power spectra while performing stepwise and ramp contractions. In: Electromyogruphical Kinesiology. Elsevier, Amsterdam, p. 31-34. 1991. 5. Clamann HP: Activity of single motor units during isometric tension. Neurolopv 20:25&260. 1970. 6. DeLuca Cl, LeFever RS. McCue MP, Xenakis AP: Behaviour of human motor units in different muscles during linearly varying contractions. J Physiol 329:113-128, 1982. 7. Gydikov A, Kosarov D: Some features of different motor units in human biceps brachii. Pfliigers Arch 347:75-88, 1974. 8. Hagberg M, Ericson BE: Myoelectric power spectrum dependence on muscular contraction level of elbow flexors. Eur J Appl Physiol48:147-156, 1982. 9. Kadefors R, Kaiser E, Petersen I: Dynamic spectrum analysis of myopotentials with special reference to muscle fatigue. Electromyography 8:39-74, 1968. 10. Kanosue K, Yoshida M, Akazawa K, Fuji K: The number of active motor units and their firing rates in voluntary contractions of the human brachiahs muscle. Jpn J Physiol29:427443, 1979.
11. Kukulka CG. Clamann HP: Comparison of the recruitment and discharge properties of motor units in human brachial biceps and adductor pollicis during isometric contractions. Brain Res 219:45-55. 1981. 12. Kwatney E, Thomas DH, Kwatny HG: An application of signal processing techniques to the study of myoelectric signals. IEEE Trans Biomed Eng 17:303-313. 1970. 13. Lindstrom LR, Magnusson R, Petersen I: Muscular fatigue and action potential conduction velocity changes studied with frequency analysis of EMG signals. EIectromyography 4:341-353.
1970.
14. Lindstrom L, Kadefors R, Petersen I: An electromyographic index for localized muscle fatigue. J Appl Physiol 43:7_5@754. 1977.
15. Mills KS: Power spectral analysis of electromyogram and compound muscle action potential during fatigue and recovery. .I Physiol 362:4Ol-t09, 1982. 16. Petrofsky SJ, Lind AR: Frequency analysis of the surface electromyogram during sustained isometric contractions. Eur J Appl Physiol43:173-182,
1980.
17. Solomonow M: External control of the neuromuscular system. IEEE Trans Biomed Eng BME 31~752-763. 1984. 18. Solomonow M, Baratta R, Zhou BH, D’Ambrosia R: Electromyogram coactivation patterns of the elbow antagonist muscles during slow isokinetic movement. Exp Neurol 100: 470-477. 1988. 19. Solomonow M, Baten C, Smit J. Baratta R, Hermens H,
J Electromyogr
Kinesiol. Vol. 3. No. 1. 199.7
40
J. H. SANCHEZ
D’Ambrosia R, Shoji H: Electromyogram power spectra frequencies associated with motor unit recruitment strategies. J Appl Physio/68:1177-1185, 1990. 20. Solomonow M, Guzzi A, Baratta R, Shoji H, D’Ambrosia R: EMG-force model of the elbow antagonist muscle pair. The effect of joint position, gravity and recruitment. Am J Phys Med 65~223-244, 1986. 21. Stulen FB, DeLuca CJ: Frequency parameters of the myoelectric signal as a measure of muscle conduction velocity. IEEE
Trans Biomed Eng BME 28:515-523,
J Ekctromyogr Kinesiol, Vol. 3, No. 1, 1993
1981.
ET AL. 22. Zhou B, Ding S, Liu P, Wu X: Median frequency changes of
elbow antagonist force contraction. tional Conference
muscle pair during sustained constant In: Proceedings 12th Annual Internaof IEEE Eng Med Biol Sot pp. 2217-2218,
1990. 23. Zoutman AE: Analysis of the surface EMG during voluntary contraction. The relationship between the median frequency of the power density spectrum of the EMG and the contraction level [Dissertation]. Enschede, The Netherlands, Univ. of Twente, 1989.
Kinesiology
Ltd
Control Strategies of the Elbow Antagonist Muscle Pair During Two Types of Increasing Isometric Contractions J. H. Sanchez, M. Solomonow, Bioengineering
Laboratory,
Department
R. V. Baratta, and R. D’Ambrosia
of Orthopaedic
New Orleans, Louisiana,
Surgery,
Louisiana State University,
U.S.A.
Summary: The purpose of this study was to determine if differences exist between the control strategies of the elbow flexor and extensor muscles performing stepwise and linearly increasing isometric contractions, and to determine their control strategy when active as antagonists to each other. The electromyogram (EMG) from the biceps brachii and triceps brachii were recorded during stepwise and linearly increasing contractions in flexion and extension. The power density spectrum of the EMG was determined and the median frequency (MF) for each spectrum was calculated for assessment of changes in the average conduction velocity, which reflects motor unit recrnitment and derecruitment, and thereby the control strategy of the muscle. The results suggest that differences exist in the control strategies employed by a single muscle during stepwise and linearly increasing contractions. Furthermore, the antagonist triceps recruits motor units up to a higher force level during stepwise contractions than during linearly increasing contractions. The antagonist biceps derecruits motor units up to a higher force level during linearly increasing contractions than during stepwise contractions. Key Words: Electromyogram-Coactivation-Motor units-Recruitment.
In order to improve existing functional neuromuscular stimulation (FNS) systems, it is important to understand the firing rate and motor unit recruitment control strategies of muscles and their antagonist pairs during different types of contractions. Establishing relationships between the antagonist pair control strategies for the contraction type would allow for more efficient stimulation of the muscles, which represents less energy expenditure and longer muscle endurance while maintaining joint stability. The overall result of the implementation of such a relationship in an FNS system would be refined limb control (17). Accepted Address Solomonow University LA 70112.
Lindstrom et al. (13) developed a mathematical model for the power density spectrum that relates the spectrum to the conduction velocity of action potentials along the muscle fibers. Many investigators have noted a shift in the power density spectrum toward the low frequencies during sustained contraction (3,8,9,12-16,21). Lindstrom et al. (13) demonstrated that the shift in the power density spectrum represents a decrease in the average conduction velocity of the active muscle fibers. ArendtNielsen et al. (1) and Bigland-Ritchie et al. (3), among others, provided experimental evidence of the correlation between conduction velocity and the spectral parameters of the myoelectric signal. Stulen et al. (21) determined that the median frequency of the power density spectrum had a linear relationship to the conduction velocity, and that it was more immune to noise than other spectral pa-
November 19, 1992. correspondence and reprint requests to Prof. M. at Department of Orthopaedics, Louisiana State Medical Center, 2025 Gravier Street, New Orleans, U.S.A.
33
J. H. SANCHEZ ET AL.
34
rameters. Using different control strategies to stimulate cat muscles, Solomonow et al. (17,19) studied the spectrum median frequency, observing that the median frequency was affected only by recruitment, and not by changes in the firing rate (19). Bilodeau et al. (4) tracked the median and the mean frequencies of the spectrum obtained from the triceps brachii and the anconeus during voluntary contractions in an attempt to determine if the control strategies of the muscles change during stepwise and linearly increasing contractions. During stepwise contractions, the spectral parameters increased with force up to 20% of the maximal force level, and then either leveled off or decreased for the higher force levels. During linearly increasing contractions, the spectral parameters increased up to 80% maximal voluntary contraction (MVC) and then leveled off or decreased for the higher force levels. Bilodeau et al. (4) concluded that the type of contraction performed (stepwise or linearly increasing) has a major role in the behavior of the median frequency patterns, and thus, the control strategy of the muscles. Zhou et al. (22) studied the median frequency patterns as a function of time for the antagonist muscle pair of the elbow while maintaining a contraction force of 60% MVC. They studied the biceps and triceps brachii in flexion and extension and observed that in a given direction of contraction (flexion or extension), the antagonist muscle had higher median frequency values than the agonist. The objectives of this study are to confirm that a single muscle, acting as agonist, can change its motor unit recruitment strategy during stepwise and linearly increasing contraction; to determine the control strategies it employs when acting as an antagonist; and to delineate any differences in the strategies when the muscle acts as an antagonist in the two types of contractions mentioned above. METHODS AND PROCEDURES
ric flexion and extension. In order to allow conditions to be as close to normal as possible, no restraints were provided for the shoulder or elbow. The wrist was maintained in a neutral position (half way between pronation and supination). Instrumentation A Lebow bidirectional load cell was connected through a metal rod to a cuff around the forearm. A coupling amplifier was used to include the cell in a Wheatstone bridge configuration and to calibrate offset voltages and gain. Two silver/silver chloride pregelled surface electrodes of 1.5 cm in diameter, with a center to center interelectrode distance of 2.5 cm, were applied over the belly of the biceps brachii. A second pair of electrodes was placed over the belly of the triceps brachii. A fifth electrode was attached to unrelated tissue in the same arm, serving as ground. The myoelectric signal from each muscle was detected by a differential amplifier with a gain of up to 200,000, a common mode rejection ratio (CMRR) of 100 dB, and then band-passed in a range of l& 550 Hz. The signal from the load cell and the myoelectric signals from the biceps and triceps muscles were then digitized using a data acquisition card and stored in an IBM 386 computer at a sampling rate of 1,024 Hz for further processing. Two B & K 1477 dual trace oscilloscopes were used, one to display the raw myoelectric signals from each muscle in each channel and one to display the force exerted by the subject and the “target” force to be matched by the subject. A HewlettPackard 3964A FM tape recorder was used to provide the ramp function to be tracked by the subject during the linearly increasing contractions. The ramp function simulated a linearly increasing contraction from 0 to 100% MVC in 3 s.
Subjects
Protocol
Eleven men and one woman were used as subjects in this study, ranging in age from 19 to 46 years. The subjects were seated in a specially designed chair with their elbows flexed to 90” and their upper arms parallel to their sides. A plastic cuff was attached to their forearms, and the cuff was rigidly connected to a bidirectional load cell by a metal rod, which in turn was attached to the chair. This arrangenient allowed the subject to perform isomet-
The subjects were seated on the chair, and their forearms were strapped to the load cell as described previously. The area on the skin where electrodes were to be placed was thoroughly cleansed and slightly abraded with an alcohol pad in order to remove any skin oils, and thus reduce skin impedance. The experiment was divided into two sections, flexion and extension, and these two were further
J Electromyogr
Kinesiol. Vol. 3, No. I. 1993
CONTROL
STRATEGIES
divided into stepwise and linearly increasing contractions. The subjects were trained to obtain their true MVC first, and adjustments were made in the oscilloscope to range full scale from 0 to 100% MVC levels. Beginning the flexion part of the experiment, the MVC force during flexion was determined and eleven steps corresponding to 0 through 100% MVC force levels increasing in steps of 10% MVC were marked. During stepwise flexion the data for each of the eleven target forces was collected in a random order, in order to minimize the possibility of a systematic error introduced by fatigue. Once the subject’s force matched the target line, the force was maintained constant at that level for 2 s while the data were sampled. A 5-min rest period was strictly observed between consecutive trials. During linearly increasing flexion, the subject was to follow as close as possible a ramp signal displayed on the oscilloscope which corresponded to an increase in force from 0 to 100% MVC over 3 s, as described previously. Five linearly increasing trials were recorded for each subject. Five-min rest periods were observed between each trial. The same procedure was used for the stepwise and linearly increasing extension part of the experiment, after proper determination and calibration of the maximal voluntary contraction in extension. It was asserted that cross-talk between flexors and extensors did not exist by monitoring the EMG of one muscle plotted against the EMG of the other to show that constant phase was absent (18).
OF ANTAGONIST
MUSCLES
35
Analysis The stored myoelectric signals from the stepwise contractions were divided into four segments of 0.5 s. Each segment was multiplied by a Tukey window (10% taper) and zero padded by 5 12 points to obtain a resolution of 1 Hz. A fast Fourier transform (FFT) was applied, and an estimate of the power density spectrum was obtained. The median frequency (MF) was estimated by finding the frequency point which divided the power density spectrum into two parts of equal area. To process the myoelectric signals of linearly increased contraction, each trial was divided into overlapping segments of 128 d points. Each of the 128-point segments had 50% overlap with the previous segment. Each segment was multiplied by a Tukey window (10% taper) and zero padded by 128 points to obtain a resolution of 4 Hz. Such a short interval was necessary to minimize the nonstationary effect of the EMG caused by the changing force during the course of each epoch. Then, as in stepwise contractions, FFT was performed on each segment, followed by MF calculation. RESULTS Figure 1 shows a sample of the data obtained for a subject during a stepwise flexion trial at a constant force of 50% MVC maintained for 2 s. Figure 2 shows a sample trial obtained for a subject perform-
FIG. 1. Stepwise flexion at 50% maximal voluntary contraction (MVC). Force is displayed in the top section, and the biceps and triceps myoelectric signals are displayed in the middle and bottom sections. The force is expressed in percentage of MVC, and the myoelectric signals in millivolts. Note the scaling difference in millivolts on the EMG axis for the biceps and triceps.
J Electrom_vogr Kinaid.
Vol. 3. No. 1. 1993
J. H. SANCHEZ
36
ET AL.
100 80 66 !O 20 0 3.0
” ‘- II’
IHcnror EN6
ing a linearly increasing flexion. The force increases linearly from 0 to 100% MVC in 3 s. The mean and standard deviation values of the MF at each force level obtained from the pooled data of all the subjects while performing stepwise and linearly increasing contractions are shown in Figs. 3 and 4, respectively. A second order polynomial model for each datum set was developed, which related median frequency to force level. The peak of each model was determined by finding where the polynomial derivative equated to zero and was compared with the corresponding peak from the averaged data. The peaks obtained statistically were slightly higher than the ones obtained from the average of the data. The statistically obtained peaks could be higher because a continuous model was tit to discrete data. When the biceps was performing flexion in stepwise contractions the median frequency increased up to 50% MVC and then decreased at a slow rate, as observed in Fig. 3. When the biceps performed linearly increasing flexion, the median frequency increased up to 70% MVC and then decreased at a slow rate. When the triceps performed stepwise extension, the median frequency increased up to 50% MVC and then decreased at a slow rate. In linearly increasing extension, the triceps median frequency increased up to 70% MVC and then remained constant for the rest of the force range. When acting as antagonist during stepwise increasing extension (Fig. 3), the biceps exhibits a nearly exponential decrease in the MF up to 50% MVC, after which it levels off up to 100% MVC.
J Electromyogr
Kinesiol, Vol. 3, No. I, 1993
FIG. 2. An example of a trial with linearly increasing flexion used for analysis of the EMG power spectra. A straight line was drawn from 0 to 100% MVC to assess the quality of the linear increase of force.
The triceps, however, shows that, when acting as antagonist during elbow flexion, the MF increases up to 90% MVC, with a sharp drop thereafter. The antagonist activity during linearly increasing contraction is shown in Fig. 4. The biceps exhibits a short increase in MF from 0 to 20% MVC and a gradual decrease thereafter, for the complete force range up to 100% MVC. The triceps, as antagonist, exhibits a gradual increase in MF up to 70% of the force range, and a gradual decline thereafter. DISCUSSION The results suggest that there is a difference in the control strategy of a muscle performing two types of contractions. These results agree in general with those obtained by Bilodeau et al. (4) who studied the EMG power spectra during stepwise and linearly increasing contractions in the triceps brachii and the anconeus. They reported an increase in the median frequency up to 20% MVC during stepwise contractions and an increase up to 80% MVC during linearly increasing contractions, and a decrease thereafter. Their report suggests that there is a different control strategy for stepwise and linearly increasing contractions. The numbers obtained in the present study differ from those in Bilodeau’s work. However, in a previous study conducted in this laboratory using intramuscular electrodes, Zoutman (23) reported an increase in the median frequency of the biceps in stepwise flexion up to 50% MVC, and either a slow decrease or constant median frequency thereafter. Zoutman
CONTROL
STRATEGIES
0 lO20304060607030eOl00 Force (%)
Force (%)
A
MFR (Hz)
I
36. 30. 76. 70. 66. 60. 60’
* ’ ’ ’ ’ ’ ’ ’ 0 10 20304060607030Wl00
Force (%)
C FIG. 3. Mean and level for the median wise contractions. tagonist (b), triceps (d) are shown.
’
*’
660 lO203040608070600010 Force (%)
D values versus force standard deviation _ frequencies of subjects performing stepThe biceps as agonist (a), biceps as anas agonist (c), and triceps as antagonist
OF ANTAGONIST
37
MUSCLES
muscles when acting as antagonists during sustained constant force flexion and extension. They monitored the median frequencies of the power density spectra from the biceps and triceps muscles, while maintaining a contraction force of 60% MVC. Zhou et al. observed that during both flexors and extensors acting as antagonists had median frequencies higher than the agonist’s median frequencies, and that these values decreased while the contraction was maintained. In the present study, during both stepwise and linearly increasing extension, the antagonist’s median frequencies were higher than the agonist’s median frequencies. During stepwise and linearly increasing flexion, however, the agonist’s median frequencies were slightly higher than the antagonist’s median frequencies. These differences observed between Zhou’s work (22) and the present study could be a result of the factors affecting the detection of the conduction velocity. The fact that the results are different does not suggest that either experiment ,&AFR
(Hz)
,rOwR
n-31
III1
(Hz)
I
n-40 130.
cm-
120. 30.
110. loO*
70.
00.
used a procedure similar to the one used in the experiment reported here. Further studies with a fixed procedure for determining MVC are necessary to better understand this behavior and the cause for these differences. Prior studies (4,23) suggest that a muscle uses different control strategies for stepwise and linearly increasing contractions, as mentioned previously. The difference in the control strategies suggested by the results of the present report may explain the differences in control strategies reported for the same muscle in several studies from different groups (57,10,11). Most of these studies do not specify the type of contractions performed and most likely the differences in protocol for each experiment are a key factor for these differences. It is essential to study muscles under different types of contractions using the same experimental procedures in order to obtain a better understanding of the causes for the differences in control strategies. Zhou et al. (22) studied the median frequency and mean absolute value (MAV) changes of the elbow
60. 60,
30. .
.
.
.
.
.
.
,(
70’
.
0 10 2030406060703080100
.
.
.
.
.
.
,
.
.,
.
.
J
0 10 2030406~60703080100
Force (a;)
Force (a;)
B
A MFR (Hz) 86
.ndO s-10
0 10 203040606070308010(
36’
’
’
-
.
*
A
*
0 102030406080703090lCNJ
Force (%)
Force (%I
D
C
FIG. 4. Mean and standard deviation values versus force level for the median frequencies of subjects performing linearly increasing contractions. The biceps as agonist (a), biceps as antagonist (b), triceps as agonist (c), and triceps as antagonist (d) are shown.
J Electrom_vogr Kinesiol.
Vol. 3. No. I. I!293
38
J. H. SANCHEZ
has wrong results. In addition, the populations of subjects used for each experiment could also have differences among themselves. For example, the EMG recorded with surface electrodes is subjected to several factors that may induce a low-pass filtering effect. Skin thickness which is greater for men than women, fatty tissue between skin and muscle, electrode diameter, and interelectrode distance are some of these factors. Another issue which may affect the value of the MF is the exact location of the electrodes over the muscle group, and their individual predominant fiber type composition. Although most of the EMG was recorded from the biceps, some signal came from the brachialis and coracobrachialis. Each of these muscles may have slightly or largely different predominant fiber types, and the exact location of the electrodes over the flexors may emphasize or attenuate the EMG of fast twitch fibers, for example, to yield somewhat higher or lower MF. This, however, explains variabilities in MF values from subject to subject and from study to study, but it does not impact on the fact that the overall control strategy changed according to the specific function performed by the muscle, e.g., stepwise or linearly increasing force contractions. In the present study the median frequency of the biceps in extension was found to decrease with increasing force level for both linearly increasing and stepwise contractions. This behavior suggests that derecruitment of motor units takes place as the net extension force is increased. In contrast, the median frequency of the triceps in flexion was found to increase with force level on both linearly increasing and stepwise contractions. These results suggest that the,triceps and the biceps as antagonists have completely different control strategies. Solomonow et al. (20) studied the normalized MAV of the EMG from the biceps and triceps muscles as both agonists and antagonists in a stepwise increasing contraction. They observed that the MAV from each muscle, acting as either agonist or antagonist, increases with force level. They suggested that the increase in activity with force level observed in the antagonists, stabilized the joint and augmented the contact area of the joint, thus reducing stress concentrations on the articular surfaces (2,18,20). The electrical activity (MAV) in a muscle increases as a result of two factors: an increase in the number of active motor units and an increase in the average firing rate of the active motor units. The
J Electmnyogr
Kinesiol,
Vol. 3. No. 1, 1993
ET AL. MAV is not a simple summation of the effect of firing rate or recruitment, but some complicated type of summation. The median frequency changes when motor units are recruited or derecruited, and stays relatively unaffected whether the firing rate increases or decreases (19). According to Solomonow’s observations (20), the MAV increases with force in both the biceps and triceps acting as antagonists. No distinction can be made as to which of the two processes for increasing force (if not both) is taking place. However, since the median frequency reflects recruitment and derecruitment of motor units, combining these two parameters (median frequency and MAV), assessment of the individual contributions of both recruitment and rate coding (change in firing rate) can be made. Based on this argument, some suggestions can be made on the observed behavior of the biceps and triceps as antagonists. The median frequency for the biceps in extension was noted to decrease with increasing force level, whereas the MAV was noted to increase with force level (20). The median frequency and the MAV increased with force level for the triceps in flexion, as in the case of the biceps and triceps each acting as agonist. In this study, it seems that the flexor in extension mostly uses increase in rate coding to generate the necessary stabilizing force while derecruiting motor units, whereas the triceps in flexion uses both rate coding and recruitment to generate the stabilizing force. Comparing the control strategy of the biceps in extension and triceps in flexion for linearly increasing and stepwise contractions is interesting. In both muscles, the median frequency patterns changed with contraction type. Derecruitment occurred in the biceps throughout the whole force range for linearly increasing extension, and down to 60% MVC in stepwise extension. In the triceps in flexion, recruitment occurred up to 70% MVC during linearly increasing contractions, and up to 90% MVC during stepwise contractions. According to these results, it seems that both the biceps in extension and the triceps in flexion use different strategies when performing stepwise and linearly increasing contractions. The biceps in extension seems to rely more on rate coding than recruitment to generate the necessary force for stabilizing the joint, as stated previously. During linearly increasing extension, the biceps derecruits motor units slowly throughout the force range to achieve a smoother extension, thus showing a longer derecruitment as compared with stepwise extension. To
CONTROL
STRATEGIES
generate the necessary force, it is possible that for the higher force levels, it relies on increasing the firing rate of those motor units that remain active. In the case of the triceps in flexion, it appears to rely on both rate coding and recruitment to generate the stabilizing force, as previously suggested. When performing stepwise flexion, recruitment takes place up to 90% MVC, suggesting that the motor units are recruited slowly, as compared with linearly increasing flexion, where recruitment occurs up to 70% MVC. This behavior represents a reversal of the results obtained for the agonists in linearly increasing and stepwise contractions, where motor units were recruited up to higher force levels during linearly increasing contractions than during stepwise contractions. These differences can not be explained with the available literature, and more work is necessary to understand the behavior of the antagonists during linearly increasing and stepwise contractions. CONCLUSIONS From the data and its analysis, the following conclusions can be made: (a) Differences exist in the control strategies of muscles performing as agonists during stepwise and linearly increasing contractions. The biceps in stepwise flexion recruited motor units up to 50% MVC and in linearly increasing contraction up to 70% MVC. Similarly, triceps motor units were recruited up to 50% MVC in stepwise contraction, and up to 70% MVC in linearly increasing contraction. For the higher force levels, the muscles probably depended on rate coding to generate extra force. (b) The MAV reflects both change in firing rate and recruitment. The median frequency is mostly affected by motor unit recruitment. Using both the median frequency and the MAV, an assessment can be made of the role of rate coding in contributing to the control strategy. (c) The biceps in extension derecruits motor units as force level increases. (d) The triceps in flexion recruits more motor units as force level increases. (e) There are differences in the control strategy of the biceps in extension during stepwise and linearly increasing force contractions. Derecruitment is slower during linearly increasing extension than during stepwise extension. (f) There are differences in the control strategy of the triceps in flexion during stepwise and linearly increasing contractions. Recruitment is slower during stepwise flexion than during linearly increasing flexion.
OF ANTAGONIST
39
MUSCLES
Acknowledgment: This work was supported by Grants BCS 9207007 and BCS 9006639 from the National Science Foundation.
1. Arendt-Nielsen L, Forster A, Mills KR: EMG power spectral shift and muscle fibre conduction velocity during human muscle fatigue. J Physiol 353:54, 1984. 2. Baratta R, Solomonow M, Zhou BH, Letson D. Chuinard R, D’Ambrosia R: Muscular coactivation. The role of the antagonist musculature in maintaining knee stability. Am J Sports Med 16: 113-122, 1988. 3. Bigland-Ritchie B, Donovan EF. Roussos CS: Conduction velocity and EMG power spectrum changes in fatigue of sustained maximal efforts. J Appl PhysioL 51: 1300-1305, 1981. 4. Bilodeau M, Arsenault AB, Gravel D, Bourbonnais D, Kemp F: Comparison of EMG power spectra while performing stepwise and ramp contractions. In: Electromyogruphical Kinesiology. Elsevier, Amsterdam, p. 31-34. 1991. 5. Clamann HP: Activity of single motor units during isometric tension. Neurolopv 20:25&260. 1970. 6. DeLuca Cl, LeFever RS. McCue MP, Xenakis AP: Behaviour of human motor units in different muscles during linearly varying contractions. J Physiol 329:113-128, 1982. 7. Gydikov A, Kosarov D: Some features of different motor units in human biceps brachii. Pfliigers Arch 347:75-88, 1974. 8. Hagberg M, Ericson BE: Myoelectric power spectrum dependence on muscular contraction level of elbow flexors. Eur J Appl Physiol48:147-156, 1982. 9. Kadefors R, Kaiser E, Petersen I: Dynamic spectrum analysis of myopotentials with special reference to muscle fatigue. Electromyography 8:39-74, 1968. 10. Kanosue K, Yoshida M, Akazawa K, Fuji K: The number of active motor units and their firing rates in voluntary contractions of the human brachiahs muscle. Jpn J Physiol29:427443, 1979.
11. Kukulka CG. Clamann HP: Comparison of the recruitment and discharge properties of motor units in human brachial biceps and adductor pollicis during isometric contractions. Brain Res 219:45-55. 1981. 12. Kwatney E, Thomas DH, Kwatny HG: An application of signal processing techniques to the study of myoelectric signals. IEEE Trans Biomed Eng 17:303-313. 1970. 13. Lindstrom LR, Magnusson R, Petersen I: Muscular fatigue and action potential conduction velocity changes studied with frequency analysis of EMG signals. EIectromyography 4:341-353.
1970.
14. Lindstrom L, Kadefors R, Petersen I: An electromyographic index for localized muscle fatigue. J Appl Physiol 43:7_5@754. 1977.
15. Mills KS: Power spectral analysis of electromyogram and compound muscle action potential during fatigue and recovery. .I Physiol 362:4Ol-t09, 1982. 16. Petrofsky SJ, Lind AR: Frequency analysis of the surface electromyogram during sustained isometric contractions. Eur J Appl Physiol43:173-182,
1980.
17. Solomonow M: External control of the neuromuscular system. IEEE Trans Biomed Eng BME 31~752-763. 1984. 18. Solomonow M, Baratta R, Zhou BH, D’Ambrosia R: Electromyogram coactivation patterns of the elbow antagonist muscles during slow isokinetic movement. Exp Neurol 100: 470-477. 1988. 19. Solomonow M, Baten C, Smit J. Baratta R, Hermens H,
J Electromyogr
Kinesiol. Vol. 3. No. 1. 199.7
40
J. H. SANCHEZ
D’Ambrosia R, Shoji H: Electromyogram power spectra frequencies associated with motor unit recruitment strategies. J Appl Physio/68:1177-1185, 1990. 20. Solomonow M, Guzzi A, Baratta R, Shoji H, D’Ambrosia R: EMG-force model of the elbow antagonist muscle pair. The effect of joint position, gravity and recruitment. Am J Phys Med 65~223-244, 1986. 21. Stulen FB, DeLuca CJ: Frequency parameters of the myoelectric signal as a measure of muscle conduction velocity. IEEE
Trans Biomed Eng BME 28:515-523,
J Ekctromyogr Kinesiol, Vol. 3, No. 1, 1993
1981.
ET AL. 22. Zhou B, Ding S, Liu P, Wu X: Median frequency changes of
elbow antagonist force contraction. tional Conference
muscle pair during sustained constant In: Proceedings 12th Annual Internaof IEEE Eng Med Biol Sot pp. 2217-2218,
1990. 23. Zoutman AE: Analysis of the surface EMG during voluntary contraction. The relationship between the median frequency of the power density spectrum of the EMG and the contraction level [Dissertation]. Enschede, The Netherlands, Univ. of Twente, 1989.