Fatigue-related decrease in Piper rhythm frequency of the abductor pollicis brevis muscle during isometric contractions

Journal of Electromyography and Kinesiology 21 (2011) 190–195

Contents lists available at ScienceDirect

Journal of Electromyography and Kinesiology journal homepage: www.elsevier.com/locate/jelekin

Fatigue-related decrease in Piper rhythm frequency of the abductor pollicis brevis muscle during isometric contractions Vinzenz von Tscharner a,⇑, Marina Barandun b, Lisa M. Stirling a a b

Human Performance Laboratory, University of Calgary, Calgary, Alberta, Canada Department of Plastic and Reconstructive Surgery, University Hospital Basel, Switzerland

a r t i c l e

i n f o

Article history: Received 31 March 2010 Received in revised form 8 August 2010 Accepted 12 October 2010

Keywords: Hand EMG Wavelet analysis Beta band Gamma band

a b s t r a c t The purpose of this study was to analyze how the frequency of the Piper rhythm of the abductor pollicis brevis muscle (APB) and thus of the rhythmic synchronization of motor units changes with fatigue. Fourteen subjects participated in the study. The EMG signals were measured during maximum voluntary contractions, and a mimicked motor unit action potential was used to simulate an EMG signal containing no rhythmicity. The simulated EMG was used as a reference. The Piper rhythm was extracted from the high frequency power (170–271 Hz) of the wavelet transformed real and simulated EMG data using the difference of the autocorrelation functions of the power. The study shows that the Piper rhythm of the APB muscle, its pacing frequency and pacing amplitude can be extracted from the EMG signal recorded during a fatiguing task. One can conclude that the pacing frequencies observed in various hands covered the whole frequency range of the Piper band which includes the beta and the gamma band frequencies observed in brain activity (17–60 Hz). While the pacing frequency decreased with fatigue the pacing amplitude did not change significantly. The Piper rhythm is a result of a changing central drive and its measurement thus allows observing changes of central drive to the muscle. The ability to better resolve the Piper rhythm in the EMG without using the coherence with the brain activity opens the possibility to study the behavior of central control in the peripheral signal. Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction Fatigue is a condition of the human body that results from exercising and affects different parts of the body. At the level of the muscle fiber, fatigue was defined as a ‘‘less than anticipated contractile response for a given level of stimulation’’ (MacIntosh and Rassier, 2002). Fatigue involves peripheral and central aspects governing the contractile response. The peripheral aspects involved in fatigue relate to changes that occur locally at the muscles and result in the hampered execution of the descending central commands. We previously reported on the decay of muscle fiber conduction velocity (MFCV) and mean frequency of the EMG signal during peripheral fatigue of the abductor pollicis brevis muscle (APB), a member of the thenar muscles (Barandun et al., 2009). In the APB muscle the force generated by maximum voluntary contraction (MVC) decreased at a relative decay rate of ( 2.1% per second) while the MFCV decay rate was ( 1.5% per second) and the mean frequency decay rate was much larger ( 4% per second) (Barandun et al., 2009). Generally, peripheral fatigue results in a decreased ⇑ Corresponding author. Address: Human Performance Laboratory, University of Calgary, 2500 University Drive, Calgary, Alberta, Canada T2N 1N4. E-mail address: [email protected] (V. von Tscharner). 1050-6411/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.jelekin.2010.10.005

MFCV (Buchthal et al., 1955; Stalberg, 1966) and was measured by different techniques (Arendt-Nielsen and Zwarts, 1989). In turn, the lower MFCV results in a shift of the power spectrum of an electromyogram (EMG) to lower frequencies (Arendt-Nielsen and Mills, 1985; Arendt-Nielsen et al., 1989; Mills, 1982; Petrofsky and Lind, 1980; Sadoyama and Miyano, 1981). The mean and the median frequency were often used as a measure of this shift. However, the shape of the power spectrum of the EMG is also critically dependent on the shape of the motor unit action potentials (MUAPs) which also change with fatigue. Thus there is no linear relationship between the decay of mean frequency and MFCV during fatigue (Brody et al., 1991; Broman et al., 1985; Merletti and Lo Conte, 1995; Sadoyama et al., 1983; Zwarts et al., 1987). The central aspects of fatigue relate to the central nervous system that generates and sends descending control signals to the muscles. With increasing fatigue the amplitude of the EMG signal increases, indicating that more and different types of motor units get activated and some of them may become synchronized (Merletti et al., 1990). The rhythmic occurrence of synchronized motor units which create rhythmic bursts (30–60 Hz) in the EMG signal were first documented by Piper (Piper, 1907) and were therefore called Piper rhythm. Motor unit synchronization is defined as a higher occurrence of simultaneous discharge of action potentials from different motor units than expected by chance

V. von Tscharner et al. / Journal of Electromyography and Kinesiology 21 (2011) 190–195

(Holtermann et al., 2009). It is generally assumed that synchronization refers to a percentage of MUAPs that arrives at approximately the same time (Hermens et al., 1992; Farina et al., 2002). The synchronization can, but does not necessarily lower the mean frequency of a power spectrum (Hermens et al., 1992). This shortterm synchronization which exists at all unit firing frequencies increases with fatigue (Mesin et al., 2009) and is believed to be due to common motoneuron inputs (Freund, 1983). There is also a rhythmic synchronization whereby some external rhythmicity synchronizes the MUAPs at a certain driving frequency (McAuley et al., 1997) causing the MUAPs cluster at well-defined time points. The Piper rhythm has been related to the brain activity (Brown et al., 1998; Brown, 2000). The coherence between the brain and the muscle activity is known to change with fatigue, however, whether it increases or decreases is still controversial (Tecchio et al., 2006; Yang et al., 2009). Thus there are at least two aspects of the central control that have to be considered, one is the short-term synchronization and the other is the rhythmic synchronization resulting from a common drive causing rhythmicity in the muscles activation. Both may change with fatigue. It is generally difficult to discriminate between peripheral and central fatigue because the EMG variables are affected by both. It seems that the fractal dimension and the synchronization index are most sensitive to the central fatigue whereas MFCV is most sensitive to peripheral fatigue (Mesin et al., 2009; Holtermann et al., 2009). To further understand how fatigue affects the control of muscles, additional aspects of central fatigue, especially the rhythmic synchronization should be studied. We have previously shown that the Piper rhythm can be isolated from the power extracted by a wavelet transform applied to an EMG signal of the APB muscle (von Tscharner et al., in press). The rhythms were defined as repetitive (non random) bursts of MUAPs creating oscillations of the power of the EMG signal. The purpose of this study was to analyze how the frequency of the Piper rhythm of the APB muscle and thus of the rhythmic synchronization of motor units changes with fatigue. The frequency of the Piper rhythm reflects the central command of muscle activation and a change in frequency with fatigue would therefore reflect one aspect of central fatigue. The study was further motivated by the fact that if we are able to quantify rhythmicity, a variable that reflects parts of the central control, other factors which affect the Piper rhythm could be studied. 2. Methods 2.1. The subjects The study was approved by the Conjoint Health Research Ethics Board of the University of Calgary. Informed consent was obtained from fourteen healthy, righthanded subjects (7 females and 7 males, average age 43 years) who participated in this study. 2.2. Experimental setup Details of the experimental setup for the right and the left hand were reported previously (Barandun et al., 2009), and the essential parts for this study are summarized below. The skin covering the APB muscle was washed with water and soap, lightly abraded and cleaned with alcohol. A linear array of five Ag-electrodes (inter electrode distance 6 mm; diameter 2 mm) placed above the muscle belly parallel to the muscle fibers allowed EMG signals to be recorded from four adjacent electrode pairs by a purpose built amplifier with active ground (Human Performance Laboratory, University of Calgary, Canada, bandwidth of 10–700 Hz, amplification 500–1000 times, sampling rate 10 kHz, bandwidth

191

10–700 Hz). Two adjacent electrode pairs (3 electrodes) were selected if the two bipolar EMG recordings showed a clear correlation, thus indicating that they were placed between the innervation zone and the muscle tendon interface. The EMG signal from the pair of electrodes that was closer to the innervation zone was used for the analysis because at this location the dispersion of the timing of the motor unit action potentials caused by the various conduction velocities is minimal. A support and fixation for the arm and hand was built to place the hand in an intrinsic plus position for the measurement of force and EMG. To obtain an isolated abduction of the APB while pushing down on a force transducer, the thumb was positioned in 30° abduction. The force transducer (LC101-25, Omega Engineering, Inc., Stamford, CT, USA) was placed below the distal insertion point of the APB muscle. Simultaneous force and EMG measurements were displayed on a screen for visual feedback and recorded for maximal voluntary contraction on a laptop computer equipped with a purpose written recording software using Matlab programming language. Force and EMG measurements from a hand placed in an intrinsic plus position were displayed on a screen for visual feedback and recorded during maximal voluntary contraction (MVC). Subjects performed six trials per hand each lasting between 5 and 6 s with a rest interval of 2 min in between. From the measurement two periods of 1.64 s were selected for the analysis (subdividable in 4 sequences of 4096 points), one starting 0.3 s and the other 3.7 s after maximal voluntary contraction was reached. The duration (1.64 s) was short enough to consider the signals as stationary. The first period contained data from the non-fatigued condition. The EMG from the second period was measured shortly before the force could not anymore be sustained and revealed a significantly lower mean frequency of the power spectrum (Barandun et al., 2009) and thus represented a fatigued condition. 2.3. Estimation of a mimicked MUAP and simulation of an EMG The EMG is often explained as superposition of multiple MUAPs that occur at random instances (interference EMG). A simulated EMG is computed by convolving a modeled MUAP with a pulse train reflecting the random instances (Hermens et al., 1992). The power spectra of a modeled MUAP and the resulting simulated EMG are identical in shape, however, the power spectrum of the simulated EMG contains additional noise, contributed by the random process of the superposition of the MUAPs (von Tscharner, 2010). In this and in our previous study (von Tscharner et al., in press), we reversed this procedure to compute a mimicked MUAP from a recorded EMG. The power spectrum from the raw EMG signal was computed using the sequential Fourier transform (using 32 sequences of 512 points each) (Rosenberg et al., 1989). An inverse Fourier transform of its square root multiplied by i yielded a symmetric, mimicked MUAP. A new, simulated EMG was then computed by convolving the mimicked MUAP with a randomly distributed pulse train of 2000 pulses. Its amplitude was adjusted to recover the energy of the raw EMG signal. If there was any rhythm in the raw EMG this rhythm will be eliminated in the simulated EMG. However, the general characteristics of the raw EMG will be closely reproduced and the power spectra of the measured and simulated EMGs will be identical (von Tscharner, 2010). Three sets of simulated data (differing by the random pulse train) were generated for each set of real data and were used as reference signals containing no neurally generated rhythmicity. 2.4. Intensities of the EMG signal extracted by the wavelet transform A set of non-linearly scaled, slightly modified Cauchy type wavelets were used to extract the EMG intensity from the EMG

V. von Tscharner et al. / Journal of Electromyography and Kinesiology 21 (2011) 190–195

2.5. Pacing frequency extracted by autocorrelation The intensities extracted by the wavelet transform were divided into 102.4 ms (1024 points) segments and the autocorrelation (ac) of a trial was the mean across these segments. The ac representing the fatigued and non-fatigued conditions were computed for each trial. It is known that random signals may in certain cases appear as if they contain rhythms. It was therefore necessary to include the ac of the simulated data as reference signals containing no neurally generated rhythmicity. From the data of one trial three sets of simulated data (differing by the random pulse train) were generated. Thus for one hand in the fatigued or non-fatigued condition there were EMG and ac data for 6 measured trials and 18 simulations. For each hand and condition one ‘‘measured averaged ac’’ was computed from the 6 trials, and three ‘‘simulated averaged ac’’ were computed by averaging sets of six of the 18 simulated ac. The next step consisted in computing the pacing frequency and the pacing amplitude. To remove the effects of any randomly generated rhythmicity, a ‘‘net measured ac’’ was obtained by taking the ‘‘measured averaged ac’’ of the 6 measured signals for a hand and subtracting the ‘‘simulated averaged ac’’ of 6 simulated signals. The times of occurrence of the first three extremes (T1, T2 and T3) in the ‘‘net measured ac’’ were determined together with their amplitudes (A1, A2, and A3). The mean of the time from 0 to T2 and from T1 to T3 represents the period DT of the oscillations. The pacing frequency of the oscillations in the ‘‘net measured ac’’ was 1/DT. The pacing amplitude of the oscillations was ((A2–A1) + (A2–A3))/2. Pacing frequency and pacing amplitude were computed for all hands in the non-fatigued and fatigued condition. As mentioned above, random signals may in certain cases appear as if they contain rhythms. A threshold for the pacing amplitude was therefore selected to represent the maximal net ac that would occur if the raw data contained only random rhythmicity. The threshold was computed using the simulated ac. A ‘‘net simulated ac’’ was computed similarly to the ‘‘net measured ac’’, whereby the actual measurement was replaced by a simulated one. To be specific, the average of the three ‘‘simulated averaged ac’’ signals were subtracted from a new ‘‘simulated averaged ac’’, comprising the average of a new set of 6 randomly selected simulated signals from the original 18. The ‘‘net simulated ac’’ has no distinct oscillatory pattern yielding A1, A2 and A3 but it has a maximum and a minimum. The difference between the maximum and the minimum of the ‘‘net simulated ac’’ within the range of 2 * DT represented the threshold for the pacing amplitude. Rhythmicity in the EMG signal was deemed to be present if the pacing amplitude extracted from the ‘‘net measured ac’’ was above this threshold. Only data from subjects where the threshold was reached for both hands and the non-fatigued and fatigued condition were used for further analysis. This allows comparing right and left hands of those subjects where there was no doubt about the presence of rhythmicity in the EMG signals of non-fatigued and fatigued conditions.

2.6. Statistics The hypothesis, that the signum of the paired differences between the pacing frequency of the non-fatigued and the fatigued state was purely random was tested by a binomial test. If the hypothesis was falsified at the 95% level of confidence then presence of a systematic change was established. The hypotheses that the differences between the pacing frequencies and between the pacing amplitudes of the fatigued and non-fatigued conditions were not different from 0 was tested by a one sided paired student’s t-test (p < 0.05). To test whether the pacing frequencies of the left and the right hands were related, and whether the pacing amplitudes correlated with the pacing frequencies, a second order linear regression analysis was used. The regression line and the correlations with their statistical errors were computed (Bevington, 1969).

3. Results A period of 0.3 s of the rectified raw EMG signal is shown in Fig. 1a. The signal was drawn from a representative hand showing moderate rhythmicity. The rhythmicity can be seen by the dark line representing the power extracted by wavelets 7–9. However, it is known that random signals may in certain cases also appear as if there were rhythms. It was therefore necessary to use the simulated data as a reference. The simulated EMG shown in Fig. 1b had the same power spectrum as the measured signal. The dark line does not show the distinct rhythmicity, however, the fluctuations had similar oscillatory characteristics. In many cases the presence of the rhythm can readily be seen in the raw data, however, in some cases visual inspection was not sufficient to determine whether the intensities really reflected rhythms or were an effect of the random superposition of MUAPs. The rhythms were extracted using the ac of the intensity extracted by wavelets 7–9. The ‘‘measured averaged ac’’ of 6 trials recorded from a non-fatigued representative hand and the ‘‘simulated averaged ac’’ of 6 simulated trials were displayed in Fig. 2 (solid and dashed lines starting at 1). The solid line reveals an initial decay followed by an oscillatory pattern reflecting the rhythmicity. The dashed line representing the ‘‘simulated averaged ac’’ showed a similar initial decay, however, the oscillatory pattern less pronounced, has a smaller and irregular amplitude fluctuation. The

(a) Amplitude

signal (von Tscharner, 2000; Barandun et al., 2009). The wavelets were characterized by their center frequencies (cf: 7, 19, 38, 62, 92, 128, 170, 218, 271, 331, 395, 466, 542 Hz). The wavelet transform yields the power of the EMG signal at each time point subdivided into the frequency bands covered by each wavelet. In this study the power recovered by the wavelets with the center frequencies 170–271 Hz were used to measure the EMG intensity representing the presence of the EMG signal. The low frequencies were eliminated because the long time resolutions of the corresponding wavelets may mask shorter rhythms. The higher frequencies were eliminated because they recover increasing amounts of power corresponding to noise in the signal. From each hand and for each trial, the intensities (power) were computed for the EMGs and the simulated EMGs.

0

0.1

0.2

0.3

0.2

0.3

Time s

(b) Amplitude

192

0

0.1 Time s

Fig. 1. (a) The rectified raw EMG signal of a typical hand is shown for a time period of 0.3 s. Superimposed is the line representing the intensity extracted by wavelets 7–9. (b) The simulated EMG of for the same hand is shown for the same time period.

193

V. von Tscharner et al. / Journal of Electromyography and Kinesiology 21 (2011) 190–195

1

Pacing frequency (Hz)

60

Autocorrelation

0.5

0

T2

0

T3

50

40

30

T1 -0.5 0

0.05 Time s

0.1

Fig. 2. ‘‘measured averaged ac’’ representing the averaged ac of 6 trials (solid line starting at 1); ‘‘simulated averaged ac’’ representing the averaged ac of 6 simulated trials (dashed line starting at 1); ‘‘net measured ac’’ representing the difference between the ‘‘measured averaged ac’’ and the ‘‘simulated averaged ac’’ (lower solid line starting at 0). The first three extremes are labeled T1, T2, and T3.

lowest trace in Fig. 2 shows the ‘‘net measured ac’’, the difference between the ‘‘measured averaged ac’’ and the ‘‘simulated averaged ac’’. The first three extremes (labeled T1, T2, and T3) were used to compute the pacing frequency and the pacing amplitude. The pacing amplitude in this example was 0.3 and was larger than the threshold amplitude of 0.08. Ten out of the 14 subjects had valid EMG signals for both hands with pacing amplitudes larger than the computed thresholds. The average of all thresholds of the non-fatigued and fatigued conditions was 0.09. Thus there were EMG data available for 20 hands. The pacing frequencies of the left hands were plotted against the ones of the right hands in Fig. 3. There was no detectable correlation between the pacing frequencies of the left and the right hands (r = 0.067; p = 0.78). The mean of the absolute differences of the pacing frequencies of the left and right, fatigued and non-fatigued hands was 9.0 ± 6.5 Hz (mean ± standard deviation). The process of getting fatigued is therefore not starting or ending at the same pacing frequency in the left or right hand. However, the pacing amplitudes of the left and right hand seemed trend wise correlated (slope = 0.94; r = 0.4; p = 0.1) but the correlation did not yield the

Pacing frequency left hands (Hz)

60

20 0

10 Index of sorted hands

20

Fig. 4. Pacing frequency of the EMG measured for the non-fatigued condition (dots) and the fatigued condition (diamonds) of the hands. The hands were sorted according to the pacing frequency of the non-fatigued condition.

95% confidence level required for rejecting the hypothesis of being no different from 0. The pacing frequency obtained for the non-fatigued and fatigued situations are displayed in Fig. 4, in the order of increasing frequency of the non-fatigued condition. The pacing frequencies of the non-fatigued and the fatigued hands covered a range from about 25 to 60 Hz. The pacing frequencies were lower in the fatigued conditions in 16 out of the 20 hands. The hypothesis that the differences in pacing frequencies were equally distributed between positive and negative values was falsified by a binomial test (p = 1.3
10 3). On average the pacing frequency dropped by 5.1 ± 1.5 Hz (mean ± standard error) when the muscle was in the fatigued state. A paired one sided t-test indicated that this decay was highly significant (p = 1.5
10 3). Thus the finding that the pacing frequencies of the fatigued hands were generally about 12% smaller than of the non-fatigued ones, as seen in Fig. 4, was statistically supported. In contrast to the pacing frequency, the pacing amplitude averaged over all hands, rose by about 0.036 ± 0.024 (mean ± standard error) in the fatigued condition. This increase was statistically not significantly different from 0 at the 95% level of confidence (p = 0.075). Both, the pacing frequencies of the non-fatigued and fatigued EMG signals were correlated to the pacing amplitude (Fig. 5). The correlation was 0.45 and was significantly different from zero (p = 3.4
10 3). This indicates that the pacing frequency was generally lower for hands which showed a pronounced rhythmicity.

4. Discussion

40

20

0 0

20 40 Pacing frequency right hands (Hz)

60

Fig. 3. Pacing frequencies of the EMG of the left and right hands. Dots: non-fatigued condition; diamonds: fatigued condition.

The results of the non-fatigued and fatigued hands showed distinct rhythmicity in the intensity of the EMG signal. The pacing frequency covered a range from about 25 to 60 Hz and was about 12% lower in the fatigued condition which corresponds to the decay rate of the MFCV (Barandun et al., 2009). The range of the pacing frequencies was the same as the frequency band obtained earlier in a spectral analysis of all hands (von Tscharner et al., in press). In contrast to the spectral analysis, the analysis based on the ac allowed the pacing frequency to be measured in the EMG signals of each hand. This was a much more difficult task and required that a threshold was used to asses whether the rhythmicity was detectable. It is possible that rhythmicity could still be present in those situations where the pacing amplitude was smaller than the threshold but could not be detected by our method. It could be that factors not related to fatigue caused the rhythm to fluctuate. If so,

194

V. von Tscharner et al. / Journal of Electromyography and Kinesiology 21 (2011) 190–195

0.8

Pacing amplitude

0.6

0.4

0.2

0.0

Threshold

-0.2 20

30 40 Pacing frequency (Hz)

50

60

Fig. 5. Pacing amplitude versus pacing frequency for the non-fatigued hands (dots) and the fatigued hands (diamonds). The horizontal line indicates the threshold of the pacing amplitude of the condition yielding the lowest pacing amplitude. The line with a slope of 0.023 Hz 1 was obtained by a second order linear regression.

the ‘‘net measured ac’’ which were observed by averaging across all trials for each hand and condition would have been dampened or would have disappeared. Furthermore the rhythm must have returned to the one of the non-fatigued condition between the trials. Synchronization of MUAPs generating rhythmicity was observed in a majority of hands and is therefore not an exceptional feature. Our results support the view that temporal pattern coding, synchronization and rhythmicity represent an integral part of central nervous system information which seems to form the basis for muscle activation (Farmer, 1999). It is known that the ABP is often affected by muscle atrophy associated with carpal tunnel syndrome (CTS) (Kulick et al., 1986; MacDermid and Wessel, 2004; Rainoldi et al., 2008). Future work may show whether CTS changes the rhythmicity. The frequency range covered by the pacing frequency is equal to the one reported for the Piper rhythm (Brown, 2000; Piper, 1907). Various studies indicated that the Piper rhythm most likely results from the rhythmic drive to human muscles in the beta band. Examples of coherence between brain activity and the rectified EMG signals indicate its relation to the central drive. The rhythms seem to originate from the primary motor cortex (Brown et al., 1998) and alternative pathways have been discussed with respect to the first dorsal interosseous muscle (McAuley et al.,1997). This pathway of activation was supported for a variety of muscles (Conway et al., 1995; Salenius et al., 1996; Salenius et al., 1997). For the APB muscle, the corticomuscular interaction in this frequency range has been well established (Mima and Hallett, 1999; Mima et al., 2000). Because the Piper rhythm is linked to the brain activity, its decrease in frequency with fatigue (Fig. 4) is, most likely, a consequence of central fatigue. During a fatiguing exercise a decay of coherence between the brain and muscle activity of the elbowflexion muscles was reported (Yang et al., 2009). Others observed an opposite effect, the coherence of the muscle and the brain activity in the beta band was increased after a fatiguing effort (Tecchio et al., 2006). The synchronization descriptor used by Holtermann et al. (2009) increased with fatigue together with tremor. The tremor may be related to the Piper rhythm. In the present study the amplitude of the Piper rhythm was sufficiently large in both, the non-fatigued and the fatigued condition to measure its frequency, however, the pacing amplitude difference induced by fatigue was not sufficient to support a strong dependence on fatigue. Furthermore, the mean of the absolute differences of the pacing frequen-

cies of the left and right hands, fatigued and non-fatigued, was larger than the drop in the pacing frequency caused by fatigue which was similar in the left and right hands. It seems as if the APB muscle of the right and left muscles are paced by a frequency in the range of the beta band but not by the same frequency. Consequently the beta band must be viewed as a broad band of frequencies whereas the individual pacing of the muscle occurs in a more narrow frequency band. Our previous results from the analysis in the frequency domain indicated that the band representing the rhythms in the EMG signals was not very narrow and thus confirm this view (von Tscharner et al., in press). There were some limitations to this study. The first 3 extreme of the ‘‘net measured ac’’ were used to measure the pacing frequency and pacing amplitude. If the intervals between the rhythmic bursts of the EMG signal rhythms were perfectly constant, one would expect more oscillations in the ‘‘net measured ac’’. Thus the rhythm seems to get out of phase after a short while, resulting in a leveling of the ‘‘net measured ac’’ as the time shift is increased. In the analysis done in the frequency domain (von Tscharner et al., in press) the pacing frequency could be extracted without using a simulated EMG signal. The extraction of the pacing frequency in individual hands could not anymore be done without the simulated EMG data. In fact, one would not know whether the oscillations seen in the ‘‘net measured ac’’ trace (Fig. 2) reflect rhythms or random fluctuations. The drawback of using simulated EMG data was its additional contribution of noise, a process that unfortunately was unavoidable. Our primary assumption was that one characteristic pacing frequency would be typical for the Piper rhythm. Previous work showed that both, the beta and a gamma band had to be considered (von Tscharner et al., in press). Although the pacing frequency in most hands is more likely associated with beta band activity, we cannot exclude that the higher pacing frequencies were present. The ‘‘net measured ac’’ may show the pacing frequency of the beta band modulated by the pacing of the gamma band. Both the beta and the gamma bands contribute rhythms seen in the EMG intensity. This additional power was not sufficient to create an obvious interference pattern in the observed ‘‘net measured ac’’. However, because there may be different superimposed rhythms one has to consider using an autoregressive analysis in future studies to discriminate multiple rhythms. It is known that rhythmicity can also be produced by stochastic processes. Therefore the dashed line of the ‘‘simulated averaged ac’’ shown in Fig. 2 does not approach the base line in a monotone increasing way. However, similar statistical fluctuation occurs in measured data and represents a major limitation in the detection of the Piper rhythm. In this study subjects were asked to apply maximal voluntary contraction because it was expected that most of the motor units were activated and synchronization would be strongly expressed. In future studies the question of synchronization at sub-maximal or dynamic contractions has to be studied. In this case one might find a task specific firing rate activating a subpopulation of motor units. Currently it remains an open question whether the rhythm will still be visible. The study shows that the Piper rhythm of the APB muscle, its pacing frequency and pacing amplitude can be extracted from the EMG signal recorded during a fatiguing task. One can conclude that the pacing frequencies observed in various hands covered the whole frequency range of the Piper band which includes the beta and the gamma band frequencies typically observed in brain activity. While the pacing frequency decays with fatigue the pacing amplitude does not change significantly. The Piper rhythm is a result of a changing central drive and its measurement thus allows observing changes of central drive to the muscle. The ability to better resolve the Piper rhythm in the EMG without using the

V. von Tscharner et al. / Journal of Electromyography and Kinesiology 21 (2011) 190–195

coherence with the brain activity opens the possibility to study the behavior of central control in the peripheral signal. This is not limited to fatigue but will be of special value when studying EMG signals obtained during movements. References Arendt-Nielsen L, Mills KR. The relationship between mean power frequency of EMG spectrum and muscle fibre velocity. Electroencephalogr Clin Neurophysiol 1985;60:130–4. Arendt-Nielsen L, Zwarts M. Measurement of muscle fiber conduction velocity in humans: techniques and applications. J Clin Neurophysiol 1989;6:173–90. Arendt-Nielsen L, Mills KR, Forster A. Changes in muscle fiber conduction velocity, mean power frequency, and mean EMG voltage during prolonged submaximal contractions. Muscle Nerve 1989;12:493–7. Barandun M, von Tscharner V, Meuli-Simmen C, Bowen V, Valderrabano V. Frequency and conduction velocity analysis of the abductor pollicis brevis muscle during early fatigue. J Electromyogr Kinesiol 2009;9(1):65–74. Bevington PR. Data reduction and error analysis for the physical sciences. New York: McGraw-Hill; 1969. Brody LR, Pollock MT, Roy SH, De Luca CJ, Celli B. PH-induced effects on median frequency and conduction velocity of the myoelectric signal. J Appl Physiol 1991;71:1878–85. Broman H, Bilotto G, De Luca CJ. Myoelectric signal conduction velocity and spectral parameters: influence of force and time. J Appl Physiol 1985;58:1428–37. Brown P. Cortical drives to human muscle: the Piper and related rhythms. Prog Neurobiol 2000;60(1):97–108. Brown P, Salenius S, Rothwell JC, Hari R. Cortical correlate of the Piper rhythm in humans. J Neurophysiol 1998;80(6):2911–7. Buchthal F, Guld C, Rosenfalck P. Propagation velocity in electrically activated muscle fibres in man. Acta Physiol Scand 1955;34:75–89. Conway BA, Halliday DM, Farmer SF, Shahani U, Maas P, Weir AI, et al. Synchronization between motor cortex and spinal motoneuronal pool during the performance of a maintained motor task in man. J Physiol 1995;489:917–24. Farina D, Fattorini L, Felici F, Filligoi G. Nonlinear surface EMG analysis to detect changes of motor unit conduction velocity and synchronization. J Appl Physiol 2002;93:1753–63. Farmer SF. Pulsatile central nervous control of human movement. J Physiol 1999;15:517. Freund HJ. Motor unit and muscle activity in voluntary motor control. Physiol Rev 1983;63(2):387–436. Hermens HJ, v Bruggen TA, Baten CT, Rutten WL, Boom HB. The median frequency of the surface EMG power spectrum in relation to motor unit firing and action potential properties. J Electromyogr Kinesiol 1992;2(1):15–25. Holtermann A, Grönlund C, Karlsson JS, Roeleveld K. Motor unit synchronization during fatigue: described with a novel sEMG method based on large motor unit samples. J Electromyogr Kinesiol 2009;19(2):232–41. Kulick MI, Gordillo G, Javidi T, Kilgore Jr ES, Newmayer III WL. Longterm analysis of patients having surgical treatment for carpal tunnel syndrome. J Hand Surg [Am] 1986;11:59–66. MacDermid JC, Wessel J. Clinical diagnosis of carpal tunnel syndrome: a systematic review. J Hand Ther 2004;17:309–19. MacIntosh BR, Rassier DE. What is fatigue? Can J Appl Physiol 2002;27(1):42–55. McAuley JH, Rothwell JC, Marsden CD. Frequency peaks of tremor, muscle vibration and electromyographic activity at 10 Hz, 20 Hz and 40 Hz during human finger muscle contraction may reflect rhythmicities of central neural firing. Exp Brain Res 1997;114:525–41. Merletti R, Lo Conte LR. Advances in processing of surface myoelectric signals: Part 1. Med Biol Eng Comput 1995;33:362–72. Merletti R, Knaflitz M, De Luca CJ. Myoelectric manifestations of fatigue in voluntary and electrically elicited contractions. J Appl Physiol 1990;69(5):1810–20. Mesin L, Cescon C, Gazzoni M, Merletti R, Rainoldi A. A bi-dimensional index for the selective assessment of myoelectric manifestations of peripheral and central muscle fatigue. J Electromyogr Kinesiol 2009;9(5):851–63. Mills KR. Power spectral analysis of electromyogram and compound muscle action potential during muscle fatigue and recovery. J Physiol 1982;326:401–9. Mima T, Hallett M. Corticomuscular coherence. a review. J Clin Neurophysiol 1999;16(6):501–11. Mima T, Steger J, Schulman AE, Gerloff C, Hallett M. Electroencephalographic measurement of motor cortex control of muscle activity in humans. Clin Neurophysiol 2000;111(2):326–37. Petrofsky JS, Lind AR. Frequency analysis of the surface electromyogram during sustained isometric contractions. Eur J Appl Physiol Occup Physiol 1980;43:173–82. Piper H. Ueber den willkuerlichen Muskeltetanus. Pfluegers Gesamte Physiol Menschen Tiere 1907;119:301–38. Rainoldi A, Gazzoni M, Casale R. Surface EMG signal alterations in Carpal Tunnel syndrome: a pilot study. Eur J Appl Physiol 2008;103(2):233–42. Rosenberg JR, Amjad AM, Breeze P, Brillinger DR, Halliday DM. The Fourier approach to the identification of functional coupling between neuronal spike trains. Prog Biophys Mol Biol 1989;53(1):1–31. Sadoyama T, Miyano H. Frequency analysis of surface EMG to evaluation of muscle fatigue. Eur J Appl Physiol Occup Physiol 1981;47:239–46.

195

Sadoyama T, Masuda T, Miyano H. Relationships between muscle fibre conduction velocity and frequency parameters of surface EMG during sustained contraction. Eur J Appl Physiol 1983;51:247–56. Salenius S, Salmelin R, Neuper C, Pfurtscheller G, Hari R. Human cortical 40 Hz rhythm is closely related to EMG rhythmicity. Neurosci Lett 1996;213(2):75–8. Salenius S, Portin K, Kajola M, Salmelin R, Hari R. Cortical control of human motoneuron firing during isometric contraction. J Neurophysiol 1997;77(6):3401–5. Stalberg E. Propagation velocity in human muscle fibers in situ. Acta Physiol Scand Suppl 1966;287:1–112. Tecchio F, Porcaro C, Zappasodi F, Pesenti A, Ercolani M, Rossini PM. Cortical shortterm fatigue effects assessed via rhythmic brain–muscle coherence. Exp Brain Res 2006;174(1):144–51. von Tscharner V. Intensity analysis in time-frequency space of surface myoelectric signals by wavelets of specified resolution. J Electromyogr Kinesiol 2000;10(6):433–45. von Tscharner V. Amplitude cancellations in surface EMG signals. J Electromyogr Kinesiol 2010;20(5):1021–22. von Tscharner V, Barandun M, Stirling LM. Piper rhythm of the electromyograms of the abductor pollicis brevis muscle during isometric contractions. J Electromyogr Kinesiol 2011;21(1):184–9. Yang Q, Fang Y, Sun CK, Siemionow V, Ranganathan VK, Khoshknabi D, et al. Weakening of functional corticomuscular coupling during muscle fatigue. Brain Res 2009;1250:101–12. Zwarts MJ, van Weerden TW, Haenen HT. Relationship between average muscle fibre conduction velocity and EMG power spectra during isometric contraction, recovery and applied ischemia. Eur J Appl Physiol Occup Physiol 1987;56:212–6.

Vinzenz von Tscharner was born in Switzerland, 1947. He received his diploma in applied physics and mathematics in 1974 and his PhD degree in biophysics at the University of Basel, Switzerland. He was a post doctorate fellow at Oxford University, Dept. Biochemistry, England in 1978 and 1979, and a post doctorate fellow at Stanford University, California USA, Dept. Biochemistry in 1998. He returned to the Biocenter in Basel, 1981. He was then research affiliate at the Theodor Kocher Institute in Bern and specialized in signal transduction studying cellular responses related to cytokin binding. He became Adj. Assistant Professor (1997) and Adj. Associate Professor (2000) at the Human Performance Laboratory, University of Calgary. His main field of research is the signal propagation controlling movement patterns of humans. This involves biophysical/biomedical measurements and the analysis of sensory systems.

Marina Barandun graduated from Medical school at the University of Zurich, Switzerland in 2004. In 2005, she obtained a visiting doctor research fellowship from the Department of Kinesiology, University of Calgary, Canada. During her residency, she completed two years of common trunk in General Surgery at the Triemli Hospital in Zurich, Switzerland, followed by one year of Hand Surgery at the Kantonsspital Liestal, Switzerland. Currently, she’s a resident at the Department of Plastic, Reconstructive, Aesthetic and Hand Surgery at the University Hospital of Basel, Switzerland.

Lisa Stirling (née Guevremont) received her B.A.Sc. degree (with honours) in electrical engineering from the University of Toronto, Canada, in 2002. She completed her PhD in medical sciences (biomedical engineering) at the University of Alberta, Canada, in 2007. Her graduate research focused on the development of control algorithms and functional electrical stimulation protocols for restoring standing and stepping after spinal cord injury. She is currently a post-doctoral fellow at the Human Performance Laboratory, University of Calgary, where she is pursuing her interests in the neural control of movement and the application of engineering approaches to the fields of rehabilitation and movement science.