| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Department of Clinical Neurophysiology, Institute of Neurology, University Medical Centre, 6500 HB Nijmegen, The Netherlands
Submitted 19 July 2002 ; accepted in final form 21 May 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONREFERENCES |
|---|
tibialis anterior muscle; size principle; surface electromyography; peak velocity; muscle fatigue; maximum voluntary contraction
If the recruitment of motor units is as orderly as suggested by our laboratory's NMR spectroscopy studies (18, 19), also a clear-cut pattern in the MFCV development of the TA could be expected. Namely, an MFCV decrease in the first part of a sustained 30% MVC exercise, followed by an increase of MFCV, and probably followed by a decrease again toward the end of exercise.
The MFCV consists of the contributions of different (motor unit) action potentials in the sEMG. Lange et al. (25) showed that the spread in MFCV followed a normal (Gaussian) distribution in the m. biceps brachii at different contraction levels (0-100% MVC) of short duration (1.5 s). Fatigue of motor units and a shift in motor unit recruitment from type I units at the start of exercise to type II units toward the end should be reflected in a spread of MFCV. We also developed a method to calculate the conduction velocities of action potentials in the bipolar sEMG to add information about the composition of the MFCV during long lasting exercises.
For proper analyses, the correct placement of surface electrodes is very important (17). Therefore, we applied a multichannel sEMG method, which enables the choice of the best quality signals post hoc.
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
Exercise. Subjects had a supine position on the investigation table with the left leg slightly bent and supported with a vacuum pillow (Fig. 1A). The angle between the left lower leg and the left foot was 90°, and the foot was secured to a foot plate by using Velcro straps. Subjects were supplied with visual feedback of the force and were verbally encouraged during exercise. The ankle ergometer system used was home built and was applicable both for dorsal and plantar ankle flexion. The force signal was digitally stored, together with an externally generated time code signal, at a sampling rate of 100 Hz.
|
After offset correction for the force, two MVC measurements were done, from which the maximum was used to calculate the force level (Table 1). MVC measurements and exercise were separated by at least 10 min. The subject had to maintain the desired force level within a bandwidth of ±3% MVC. When he or she could no longer meet this condition, despite encouragement, exercise was stopped. The exercise consisted of an isometric dorsal ankle flexion until fatigue at 30% MVC in all subjects and, on the basis of their results at this MVC level, also at 40% MVC in subjects 3 and 5. Measurements were separated by 2-14 days. sEMG. For skin surface recordings, 31 gold-coated electrodes, with a diameter of 1.52 mm, were used [type "serrated contact"; Farnell (7)]. The electrodes were fixed into an electrode holder with interelectrode distances of 6 mm in a 4 x 8 configuration (Fig. 1B), with one empty corner. First, the skin was shaved, and electrode cream was rubbed into the skin. After the superfluous cream was carefully removed from the skin surface, the holder was placed with the four columns of seven to eight electrodes parallel to the muscle fiber direction on the distal part of the TA, where the muscle fibers are situated parallel to the skin surface. Additionally, a 10-mm-diameter gold-coated conventional EEG electrode filled with electrode gel was attached to the knee as a common reference, which allowed so-called monopolar recording from all 31 "active" electrodes. A ground electrode was attached to the ankle.
The monopolar signals were amplified, band-pass filtered (3-800 Hz), and simultaneously analog-to-digital converted (16 bits with a resolution of 0.5 µV/bit at a rate of 4,000 samples · s-1 · channel-1) by using a multichannel amplifier system (Mark 6, Biosemi, Amsterdam, The Netherlands). A second gold-coated EEG electrode placed on the upper part of the TA acted as a sense electrode, which allowed the use of the "driven right leg" principle for common mode rejection improvement (30). Data were stored for off-line analysis on the hard disk of a 180-MHz Pentium personal computer. To synchronize the data with the force recording afterward, the externally generated time code signal was also registered on one of the amplifier's channels.
Time series analysis of sEMG signal. Bipolar signals were derived by subtracting two monopolar signals from electrodes at consecutive positions along the four columns (fiber direction). The column with the overall highest amplitudes and the shortest durations of the individual peaks was chosen for further analysis. The bipolar sEMG signals of this column were baseline-corrected and divided in successive epochs of 2.048 s. For each of these epochs, the following parameters were calculated: the root mean square (RMS) voltage (as a measure of the mean amplitude), the median frequency (fmed), and the MFCV using the phase-difference method (see MFCV and PV: phase difference method below). The MFCV calculations were based on two consecutive bipolar signal pairs along the column (over a distance of 6 mm in the fiber direction). Signal pairs showing a maximum cross-correlation coefficient of <0.9 were excluded from the MFCV calculations. MFCV values tend to increase when they are observed close to the motor endplate or close to the muscle tendon transition (17). Therefore, the pair of consecutive bipolar signals with the overall lowest MFCV values was selected for presentation and also for the subsequent analysis of velocities of individual peaks.
Velocity analysis of individual peaks. The peak velocity (PV) method calculates the conduction velocities of action potentials in the sEMG signal, which gives additional information about the composition of the MFCV. For each of the before-mentioned 2-s epochs, the peaks were selected, and then the associated delay for each peak to travel between the two adjacent electrodes (a known distance apart) was estimated.
Peak selection. The selected pair of two consecutive signals was given the polarity whereby the largest (depolarization) action potentials were oriented upward. These action potentials were called peaks. Each peak of >30 µV was detected. This threshold value was based on the noise level of the bipolar sEMG signal in rest preceeding exercise. The amplitude and timing of each peak were stored. Around the maximum of each peak, a square window of 16 ms was placed. Such a window was large enough to contain the whole peak, unless peaks were substantially broadened. In that case, the window was widened to 32 ms. The fmed value of the associated 2-s signal segment was used as an objective criterion for window width. The window was doubled if fmed was <40 Hz, because the frequency content of the sEMG signal decreases as the peak width broadens. In this study, fmed was only used for this purpose. If the distance between two consecutive peaks was smaller than half of the window size, the largest peak would be counted twice, and if a small peak was located on the flank of another peak, the flank would be selected as maximum instead of the peak. Therefore, double-counted peaks and peaks with their maximum values appearing close (3 samples) to the time border of a peak window were excluded.
Velocity calculation. First, each of the 16- or 32-ms peak signals was weighted with a Hanning window. The MFCV previously determined for the 2-s signal segment was used as a first estimate of the delay between two corresponding peaks detected at the two adjacent electrode pairs. The peaks were prealigned according to this MFCV-based delay. For a further estimate of the delay, the phase-difference method was used (see below).
MFCV and PV: phase difference method. The method is based on a
property of the Fourier transform: assuming a constant time delay between two
identical signals, the phase difference (
) between the
Fourier-transformed signals will be proportional to the frequency. This
property can be used to align waveforms
(29) and/or to determine the
time delay with, in principle, an unlimited temporal resolution. The steps for
this procedure were as follows.
For the two baseline-corrected bipolar signal segments (detected at the two
adjacent electrode pairs), the cross correlation was calculated. The shift at
which the maximum cross correlation occurs is a rough estimate in sample point
resolution of the delay. The prealignment of the signals with this delay
avoids >2
phase differences in the next step. Both signals, now roughly
aligned, were Fourier transformed. This results in a description of the data
in phase and power as functions of frequency. We assumed that the signals are
identical except for a delay T. We calculated this delay by fitting a straight
line through the phase difference between both signals as function of
frequency 
( = T · ). Frequencies at which almost
no power is measured overcontribute to this phase-difference function.
Therefore, the power distributions were used as a weighting factor in the
linear regression. At each frequency value, the smallest of the two power
functions was taken. The least-square error method was used as a fitting
procedure. The resultant delay used to calculate MFCV or PV was the sum of the
(initial) time-domain delay (limited by the sampling resolution) and the
finely estimated delay (determined from the phase-based method).
MFCV was calculated by using 2.048 s of two bipolar sEMG signals from the adjacent electrode pairs. PV values were estimated by using two prealigned filtered bipolar sEMG signals of 16 or 32 ms from adjacent electrode pairs (containing roughly 1 peak).
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
sEMG. From the left column of Fig. 2, it can be concluded that all five subjects show two different periods of MFCV decrease in the course of the exercise both for 30 and 40% MVC (Fig. 2, A-G). After an initial decrease, MFCV starts to fluctuate. The mean trend of MFCV then increases. This increase is accompanied or followed by an increase of amplitude and by gradually increasing fluctuations in amplitude (Fig. 2, H-N). The MFCV then starts to decrease again toward the end of the exercise, although sometimes a final increase in the MFCV profile during the last minute of exercise can be observed (Fig. 2, A and G). Only subject 5 (Fig. 2, M and N) does not fully follow the described sEMG amplitude profile (initially almost constant and increased in the second half of the exercise). The pattern of MFCV development is most clear in subjects 1, 2, and 4 at 30% MVC, and in subjects 3 and 5 at 40% MVC (Fig. 2, A, B, D, E, and G). At 30% MVC, the period of fluctuating increase of MFCV is longer in subjects 3 and 5 (Fig. 2, C and F). The experiments were therefore repeated in these subjects at a slightly higher force level.
|
Two representative examples will be presented more extensively: the exercise at 30% MVC performed by subject 1 (Fig. 2, A and H) and the exercise at 40% MVC performed by subject 5 (Fig. 2, G and N). All other cases were analyzed in the same way. An exhaustive presentation of all cases would not add insight, however. For these experiments, segments of the raw bipolar signal (Fig. 3), amplitude, MFCV and force values together in one figure (Fig. 4), histograms of the PVs (Fig. 5), and PVs presented on one segment of the raw bipolar signal (Fig. 6) are shown.
|
|
|
|
Figure 3, A and B, shows segments of bipolar sEMG signal pairs of 500 ms at different times during the selected exercises. In Fig. 3A, larger peaks occur in the second half of the exercise. Visual inspection of the sEMG signals shows that these large peaks concern one or only a few motor units. Their large peak sizes contribute to an increased RMS amplitude toward the end of exercise. In Fig. 3B, the initial density of peaks and their relative amplitude is higher than in Fig. 3A. (Note the scaling differences.) The density of peaks decreases, and the peak amplitudes hardly change with time, which is the apparent reason for the relative high initial value and the limited dynamic range for the RMS curve in Fig. 2N.
Figure 4A shows the results for MFCV and amplitude in more detail for subject 1 at 30% MVC with the force curve added. MFCV fluctuates strongly during the transition from low to high MFCV values (between 190 and 245 s). The amplitude slowly increases but also shows large jumps during the same period. A number of simultaneous jumps of amplitude and MFCV are striking, e.g., at time (t) = 190, 216, 229, and 243 s (marked with dotted lines). From about t = 330 s, the decrease in MFCV levels off and changes into an increase until the end of exercise. The amplitude shows increased variability again during this period. The force recording remains remarkably flat during the whole period and tends to become slightly more variable toward the end of exercise.
Figure 4B shows the
more detailed results for subject 5 at 40% MVC. Again, transiently
higher values of MFCV correspond to simultaneous bumps in the amplitude
profile (marked with dotted lines at t = 1, 61, 102, 126, 147, 163,
181, and 194 s). The amplitude shows an unusual decrease during the first 2
min of exercise. This coincides with the decrease of peak density
(Fig. 3B). During the
transition from low to high MFCV values (between 
125 and 195 s), the RMS
data can be seen to decrease further and be transiently interrupted by high
peaks. The MFCV follows the same pattern: a further decrease interrupted by
substantial larger values. During the last half minute of exercise, a
transient irregular pattern of slightly higher MFCV values is observed. As in
subject 1, the amplitude shows large fluctuations during the latter
period. The fluctuations in the force seem to have a minimal correlation with
the sEMG variables (RMS and MFCV).
The PV histograms in Fig. 5, each representing selected periods of 20 s, are calculated to throw additional light on what determines the observed MFCV changes. The changes in PV distribution for both subjects demonstrate the shift of an initial population toward lower values without much change in the histogram shape (Fig. 5, Aa-Ab and Ba-Bb). A second PV population emerges with a slightly higher mean velocity than the mean at the beginning of exercise (Fig. 5, Ab-Ad and Bc-Be). The first histogram shifts further toward lower values, whereas the number of detected peaks belonging to this population decreases. At the same time, the number of peaks belonging to the second population increases (Fig. 5, Ad-Ag and Bd-Bg). When the first distribution has almost disappeared, the second one shifts toward lower values (Fig. 5, Ag-Ah and Bg-Bi). Toward the end of exercise, the histogram widens (Fig. 5, Ai-Aj and Bi-Bj).
The peak amplitudes (not shown) show a large variability. On average, the amplitudes of the peaks belonging to the first population are smaller than those belonging to the second population. But peaks with high velocities do not always have the largest amplitudes (see Fig. 6 for selected signal examples of 500 ms).
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Our laboratory's former studies (18, 19) show a striking increase of metabolic activity of (most probably) the type II fibers in the second half of a sustained isometric exercise at 30% MVC, whereas the metabolic activity of these fibers was small in the first half. It was explained by an extra recruitment of type II motor units toward the end of exercise. On the basis of these former experiments, the explanation of our present results becomes more explicit: it is most likely that the first population can be ascribed to type I motor units and the latter to type II motor units. In that line of reasoning, the MFCV is mainly determined by contributions of type I motor units during the first part of the exercise and mainly by contributions from type II motor units during the second part of exercise despite the fact that type II motor units are relatively scarce in the TA muscle (±27%; Ref. 20). The transition periods are characterized by an intermingled activation of both types, as illustrated by Fig. 6.
Although a straightforward explanation has been given here, the possibility of obtaining nearly distinct functional motor unit populations is a peculiar finding. Even if a fiber-type dichotomy is often used in global discussions, histochemical and structural investigations often reveal more gradual transitions (e.g., Ref. 8). This cannot be ascribed to a more discrete division of fiber types, especially in the TA, since a continuous distribution of motor unit properties was also found in this muscle (35). Therefore, the cause of the appearance of distinct motor unit populations has to be sought in the central nervous drive behavior, meaning a discontinuous recruitment of motor unit populations, which may be specific for this type of exercise and for the TA muscle (19). The PV method seems to be a powerful method to give insight into the composition of the MFCV, which is a mean value of the velocities of different action potentials in the sEMG. However, with the present method, we do not decompose the different action potentials into the contribution of specific motor units. Furthermore, the PV method is fire-frequency biased: motor units with high firing rates will more often be detected than those with low firing rates. It is not easy to estimate which motor units are over-represented in the PV distribution; the threshold frequency of type II motor units is higher, but the firing rate of already recruited type I motor units may be increased (10).
A remarkable PV result is the decrease of the amount of detected peaks in the course of the exercise. This goes together with an emptier visual impression of the sEMG signals. For the latter part of the exercise, this is easiest to explain. Synchronization of the motor units leads to the occurrence of large very broadened action potentials and a decrease of the amount of peaks (37). For the intermediate part, it might be related to a fatigue-induced decrease of firing rates for almost exhausted motor units (5) (especially visible in subject 5) and probably a derecuitment of the first population of motor units.
The amplitude of the sEMG signal is influenced by many factors. A first comment to make is about the basic underlying sources of the sEMG signal, namely the transmembrane action potentials of the muscle fibers. The properties of these action potentials, mainly their conduction velocity and duration, are important factors that influence the sEMG amplitude (26, 34). This makes interpretation of sEMG amplitudes solely in terms of motor unit recruitment and firing patterns rather risky under fatiguing circumstances. A constant or even decreasing amplitude accompanied with a decreasing MFCV (which happens often) can fully be ascribed to the influence of fatigue on the muscle fiber action potentials of already activated motor units. An amplitude increase accompanied with a constant MFCV (which happens sometimes) might be ascribable to recruitment of motor units with higher MFCV compensating the fatigued motor units with lowered velocities. An increase in amplitude with a decrease in MFCV (which happens toward the end of exercise) can be ascribed to synchronization of motor units (37).
But the sEMG amplitude is also influenced by the motor unit firing frequencies; this can be seen nicely in subject 5 [Figs. 3B and 4B (6-126 s)], where a decrease of peak density and not of peak amplitude is mainly responsible for the RMS decrease. Furthermore, the recruitment of a new motor unit with a large amplitude may increase the RMS substantially, as can be seen in subject 1 at 30% MVC [Figs. 3A and 4A (205-255 s)]. As was already indicated, the last part of these two exercises is characterized by a scattered distribution of peak amplitudes and velocities and irregularly occurring large broad peaks (Fig. 3). The latter feature in particular points to an increased short-term synchronization of firings of different motor units. The irregular shapes of the peaks apparently lead to "unpredictable" peak amplitudes and velocities and thus, among other things, to irregular amplitude values toward the end of exercise (Fig. 4).
The transient variations of MFCV, which coincide with transient variations of amplitude (not always vice versa), can be ascribed to the temporal (de)recruitment of large motor units with higher MFCV, possibly in connection with the influence of action potential velocity on the motor unit potential amplitudes mentioned before.
Other factors that influence the MFCV are firing rate changes (32) and probably muscle swelling (36). The decrease of peak density (firing rate) in subject 5 may also contribute to the decrease of MFCV during the first part of the exercise. Muscle swelling, because of water accumulation, seems to appear during the course of an isometric exercise at a low exertion level (9) and will potentially increase MFCV. This will probably also affect MFCV in our subjects. However, it apparently does not compensate for the decrease of MFCV during large parts of the exercise for any of our subjects.
Cross talk is always difficult to exclude or to estimate. We measured at the distal side of the TA, where fibers are parallel to the skin surface, but where unfortunately, the cross-sectional area of the TA is less. At this position along the TA, the extensors of the toes (m. extensor hallucis longus and m. extensor digitorum longus), located at the lateral side of the TA, together have about the same cross-sectional area as the TA. To minimize the chance for cross talk, we placed the electrodes at the medial side of the TA (the most medial column of electrodes is placed along the tibia bone). Furthermore, we asked the subjects to concentrate on the contraction of the TA and to relax their toes as much as possible.
In two of our subjects (3 and 5) we repeated the exercise at the higher force level of 40% MVC to get comparable results as in subjects 1, 2, and 4. It is noteworthy that, for those subjects at 40% MVC, the absolute forces are in the same range as the absolute forces of subjects 1 and 2 at their relative 30% MVC. Barnes (4) showed that the intramuscular circulatory occlusion depends on absolute forces, regardless of maximum strength. Recruitment strategies of type I and type II motor units also appear to depend on absolute forces. Unfortunately, subject 4, who showed the same pattern at 51 N (30% MVC), which is close to half the absolute force of the others, breaks this hypothesized relationship.
Krogh-Lund and Jørgensen (22-24) reported a leveling off of the decrease or an increase in MFCV simultaneously with an enlarged amplitude increase during a period of sustained exercises at 15, 25, and 30% MVC. They measured sEMG from the m. triceps brachii (25% MVC), from the m. biceps brachii (15% MVC), and from the m. brachioradialis (30% MVC). They ascribe their results to the recruitment of fresh large motor units increasing the amplitude and the MFCV and, therefore, compensating for the fatiguing of already recruited motor units, which tend to decrease the MFCV. Although, in their data, the same tendency was found as in our results, the illustration of the underlying processes appears much clearer from our data in the TA. Apart from a more detailed method of analysis, the specific muscle we studied may be a factor of importance. The mean percentage of type I fibers differs between the above muscles. It is typically 42% in the m. biceps brachii, 40% in the m. brachioradialis, 33% in the m. triceps brachii, and much higher, namely 73%, in the TA (20). The lower fraction of type I muscle fibers in the other muscles is likely to call for an earlier recruitment of type II motor units to maintain the desired force. This could be a possible explanation for the more gradual involvement of motor units with higher initial MFCV in the studies of Krogh-Lund and Jørgensen. The choice of the TA muscle was based on the clear-cut results obtained in metabolic variables studied by 31P NMR spectroscopy (18, 19).
Other than the percentage of type I fibers in a specific muscle, the occurrence of a clear alternating pattern of MFCV decrease and increase, as found in this study, depends on the chosen contraction level. At too low a level, only type I motor units will be recruited; at too high a level, motor units of both types will be already recruited in the beginning of the exercise. The extent to which a clear separation in time between the recruitment of type I and type II motor units, respectively, can be found in other muscles remains an interesting question.
In conclusion, we revealed a decrease, increase, and subsequent decrease in MFCV during sustained isometric fatiguing contractions of the TA at 30-40% MVC. PV analysis shows two populations of PVs that occur simultaneously halfway through the exercise. It is most likely that the pattern of MFCV can be ascribed to the fatiguing of already recruited predominantly type I motor units, followed by the recruitment of fresh, predominantly type II, motor units, which similarly show signs of fatigue toward the end of exercise.
|
FOOTNOTES |
|---|
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  I.-S. Hwang, Z.-R. Yang, C.-T. Huang, and M.-C. Guo Reorganization of multidigit physiological tremors after repetitive contractions of a single finger J Appl Physiol, March 1, 2009; 106(3): 966 - 974. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. S. Karlsson, K. Roeleveld, C. Gronlund, A. Holtermann, and N. Ostlund Signal processing of the surface electromyogram to gain insight into neuromuscular physiology Phil Trans R Soc A, January 28, 2009; 367(1887): 337 - 356. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. M. Enoka, P. Bawa, J. M. Wakeling, D. A. Gabriel, G. Kamen, F. Meyer, D. F. Stegeman, T. van Wessel, M. J. Zwarts, A. Rainoldi, et al. Spectral properties of the surface EMG can characterize motor unit recruitment strategies. J Appl Physiol, November 1, 2008; 105(5): 1676 - 1677. [Full Text] [PDF]
|
|
|
|
 |
|
 V. J. Boucher Acoustic Correlates of Fatigue in Laryngeal Muscles: Findings for a Criterion-Based Prevention of Acquired Voice Pathologies J Speech Lang Hear Res, October 1, 2008; 51(5): 1161 - 1170. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Banzet, N. Koulmann, N. Simler, O. Birot, H. Sanchez, R. Chapot, A. Peinnequin, and X. Bigard Fibre-type specificity of interleukin-6 gene transcription during muscle contraction in rat: association with calcineurin activity J. Physiol., August 1, 2005; 566(3): 839 - 847. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Gazzoni, F. Camelia, and D. Farina Conduction Velocity of Quiescent Muscle Fibers Decreases During Sustained Contraction J Neurophysiol, July 1, 2005; 94(1): 387 - 394. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. Farina, M. Gazzoni, and F. Camelia Conduction velocity of low-threshold motor units during ischemic contractions performed with surface EMG feedback J Appl Physiol, April 1, 2005; 98(4): 1487 - 1494. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. Farina, L. Arendt-Nielsen, and T. Graven-Nielsen Spike-triggered average torque and muscle fiber conduction velocity of low-threshold motor units following submaximal endurance contractions J Appl Physiol, April 1, 2005; 98(4): 1495 - 1502. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
