Effects of monopolar and bipolar electrode configurations on surface EMG spike analysis

Effects of monopolar and bipolar electrode configurations on surface EMG spike analysis

Medical Engineering & Physics 33 (2011) 1079–1085 Contents lists available at ScienceDirect Medical Engineering & Physics journal homepage: www.else...

607KB Sizes 0 Downloads 30 Views

Medical Engineering & Physics 33 (2011) 1079–1085

Contents lists available at ScienceDirect

Medical Engineering & Physics journal homepage: www.elsevier.com/locate/medengphy

Effects of monopolar and bipolar electrode configurations on surface EMG spike analysis David A. Gabriel ∗ Department of Kinesiology, Brock University, 500 Glenridge Avenue, St. Catharines, ON, Canada L2S 3A1

a r t i c l e

i n f o

Article history: Received 3 March 2010 Received in revised form 12 April 2011 Accepted 23 April 2011 Keywords: Interference pattern analysis Surface electromyography Biceps brachii Muscle activity Isometric contractions

a b s t r a c t This study compared the effects of monopolar and bipolar electrode configurations on interference pattern analysis of the surface electromyographic (sEMG). Twenty-four college-aged male participants performed isometric actions of the elbow flexors at 40, 60, 80, and 100 percent of maximal voluntary contraction (MVC). Separate (Ag/AgCl) electrodes were used for both configurations. There were five measures associated with “spike shape” analysis: mean spike amplitude (MSA), mean spike frequency (MSF), mean spike slope (MSS), mean spike duration (MSD) and mean number of peaks per spike (MNPPS). A load-cell and wrist-cuff assembly was used to record isometric elbow flexion forces. Both electrode configurations resulted in the same trends force changes in spike shape measures across force levels: there was a linear increase in MSA, MSS, and a quadratic decrease in MSF and the MNPPS (p’s < 0.05). The MSD underwent a quadratic increase (p < 0.05). The spike shape measures had greater mean magnitudes and exhibited greater rates of changes across force levels for the monopolar electrode configuration (p’s < 0.05). The monopolar electrode configuration was therefore more sensitive to changes in muscle activity with increases in isometric force. This is an important consideration because the rate at which muscle electrical activity develops into a full interference pattern is an important qualitative and quantitative diagnostic measure. © 2011 IPEM. Published by Elsevier Ltd. All rights reserved.

1. Introduction Alterations in the recruitment and firing rate of motor units (MU) can change the amplitude and frequency content of the surface electromyographic (sEMG) signal in similar ways, so it is difficult to distinguish between the two [1]. Increased firing rates and synchronization also have a similar impact on the amplitude and frequency content of the sEMG signal [1,2]. Spike analysis was developed to detect changes in the sEMG signal associated with the different patterns of muscle activation [3]. Spike analysis is based on the observation that the superposition between motor unit action potentials (MUAPs) results in an interference pattern that is quantitatively different for each type of MU activity pattern. This has been observed for both indwelling [4,5,9] and surface [6–8,10] recordings. Five measures that describe the average shape of individual spikes within the interference pattern were synthesized from the literature [4–8]. These five measures are: mean spike amplitude (MSA), mean spike frequency (MSF), mean spike slope (MSS), mean spike duration (MSD) and mean number of peaks per spike (MSPPS).

∗ Tel.: +1 905 688 5550x4362; fax: +1 905 688 8364. E-mail address: [email protected]

Graphically, a spike is composed of a single upward and downward deflection that is greater than the 95 percent confidence interval for baseline activity (Fig. 1). A peak is defined as a pair of upward and downward deflections within a spike that do not together constitute a discrete spike. A pattern classification table was then constructed based on observations in the literature for how each measure changes according to alterations in MU activity patterns [4–8]. Note, that the pattern of change in the five measures is distinctly different for each MU activity pattern (Table 1). It is also important to emphasize that Table 1 does not imply that the three motor unit activity patterns occur in isolation of each other. Table 1 takes advantage of the observation that different motor unit activity patterns “dominate” the force gradation process over different intervals of percent MVC when contracting the muscle from 0 to 100% MVC. The type of motor unit activity pattern that dominates the force gradation process over a particular interval depends on the recruitment range of the muscle. For example, the first dorsal interosseus (FDI) recruits MUs up to approximately 50% MVC while the biceps brachii (BB) recruits MUs up to approximately 80% MVC. In both cases, the lower threshold MUs start rate-coding while higher threshold MUs are progressively recruited but MU recruitment still dominates the force gradation process until the end of the recruitment range: 50% MVC for the FDI and 80% MVC for the BB. At higher

1350-4533/$ – see front matter © 2011 IPEM. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.medengphy.2011.04.016

1080

D.A. Gabriel / Medical Engineering & Physics 33 (2011) 1079–1085 Table 2 The means (M) and standard deviations (SD) for subject’s (N = 24) physical characteristics.

Fig. 1. Surface electromyographic (sEMG) interference pattern to illustrate the difference between a spike and a peak. A spike is composed of a single upward and downward deflection that are greater than the 95 percent confidence interval for baseline activity (shaded area). There are six spikes and spike number two has one peak (circled).

levels of force where MU recruitment is diminished, rate coding then begins to play a greater role [11–13]. The same holds true for synchronization. Synchronization is greater at higher levels of force where recruitment is greatly reduced [14]. However, the situation for neuromuscular patients is very different. Disruptions to the neuromuscular system result in distinct adaptations in motor unit activity patterns to compensate for motor unit (neuropathic) versus muscle fiber (myopathic) loss. Muscle fiber loss decreases the twitch force for each MU so myopathic patients must rely to a greater degree on enhanced recruitment earlier in the force gradation process. In contrast, neuropathic patients relay more heavily upon synchronizations to coordinate the remaining MUs in the force gradation process [15–19]. The spike shape measures have been demonstrated to correlate highly with traditional time and frequency analyses [20], and exhibit deterministic changes with increases in force [3]. Spike shape analysis has recently shown promise in the clinical domain. Calder et al. [10] demonstrated that spike shape analysis was able to discriminate between normal controls, individuals with nonspecific arm pain, and those at risk for non-specific arm pain. The three groups differed with respect to the rates of change in specific sEMG spike shape measures with increases in contraction force from 20 to 80% of maximum. Further, the changes in the sEMG interference pattern were consistent with the predictions for myopathic disorders. These findings were supported by an earlier quantitative indwelling EMG study on the same participants [9]. There are a number of technical factors related to recording the sEMG signal that can affect the sensitivity and specificity to the different motor unit behaviors, and spike shape analysis is not immune from these considerations. The most obvious and relevant of these is the electrode configuration. Action potential shape is altered as it propagates along the muscle fibers towards the electrode, and then it is spatially filtered by the bipolar detection system [21–23]. The differential recording process decreases Table 1 Predicted changes in the surface electromyographic (sEMG) spike parameters: mean spike amplitude (MSA), mean spike frequency (MSF), mean spike duration (MSD), mean spike slope (MSS), and mean number of peaks per spike (MNPPS) for alterations in MU recruitment. Motor unit firing pattern

Increased firing frequency Increased recruitment Increased synchronization

Five sEMG spike measures MSA

MSF

↑ ↑

↑ ↑ ↓

MSS

MSD

MNPPS

↑ ↑

↓ ↓ ↑

↑ ↓

Measures

M ± SD

Age (years) Height (cm) Mass (kg) Skin-fold (mm) Forearm length (mm)

24 174 89 7 29

± ± ± ± ±

3.4 10 12 4 3

both the amplitude and rate of change in the signal, as well as introduces additional phases into the recorded potentials. The magnitude of these effects also depends on the interelectrode distance [24]. However, a bipolar electrode configuration has the advantage of reducing volume conducted potentials because the differential amplifier suppresses common signal components from far away bioelectric sources, e.g. deep MUs. This is not the case for a monopolar electrode configuration which is more susceptible to common mode signals which decrease the signal-to-noise ratio and make it more difficult to detect changes in muscle activity [25–29]. The purpose of this paper is therefore to determine the impact of monopolar versus bipolar recordings on the sensitivity to change in sEMG spike shape measures across force levels. This question is important because clinicians use “absolute” EMG amplitude and rates of change in EMG measures to categorize neuromuscular patients. It is hypothesized the measures obtained by monopolar recording will exhibit greater rates of change across force levels. This was assessed using an analysis of variance with an orthogonal polynomial breakdown for means across force levels. If for example, MSA exhibits a significant linear increase for both monopolar and bipolar electrode configurations, the interaction term can test if linear trend component is significantly different for the two electrode configurations. 2. Methods 2.1. Participants Twenty-four male participants reported twice to the laboratory. The first session was to familiarize participants with the experimental set-up and the demands of the task. Each participant then read and signed the informed consent document accordance with the university research ethics guidelines. All subjects were free of upper body disabilities and were all right-handed. Their physical characteristics are presented in Table 2. 2.2. Experimental design Height, weight, and skin-fold thickness of the biceps brachii (BB) at the location of the surface electrode were obtained at the beginning of the second session. Participants performed step isometric contractions because they allow for a longer data analysis window and therefore less variability in the calculated sEMG measures [30]. The selected percentages of maximal voluntary contraction (40, 60, 80, and 100 percent MVC) have already been demonstrated to result in significant changes in the time and frequency measures of sEMG activity consistent with numerous other studies [3]. Maximal isometric strength of the elbow flexors was first obtained by the following procedures. There were three 5-s contractions with a 3-min rest interval. A target was presented as a horizontal line on an oscilloscope (Hitachi, VC-6525) that was placed in front of the subject. The target was a load cell (JR3 Inc., Woodland, CA) voltage that represented 110 percent of the mean peak value of the previous three contractions. If participants were able to reach the target line during the fourth trial, the MVC value was updated. The mean of the 3 trials was used to identify the mean

D.A. Gabriel / Medical Engineering & Physics 33 (2011) 1079–1085

maximal contraction from which the load cell (JR3 Inc., Woodland, CA) voltages at 40, 60, 80, and 100 percent of MVC were identified. There were 3 trials at each percentage of MVC. Each contraction was 5-s in duration with a 3-min rest interval. There are 24 permutations for the order of presentation of the experimental conditions. Presentation of the experimental conditions was therefore balanced across the 24 participants to prevent any systematic bias due to order or potential fatigue effects. There were three contractions at 100 percent MVC immediately before and after the experiment to test for the presence of fatigue. The contractions had the same work-to-rest ratio as those during the experiment. 2.3. Apparatus and testing position All testing took place within an electrically isolated room. Participants sat in a testing chair and placed their arm in a jigg that was designed to isolate the action of the elbow flexors in an isometric contraction. Both the shoulder and elbow of the right arm were flexed at 90◦ in the sagittal plane. This testing position was maintained with the back of the elbow on a support so that the wrist was mid-way between pronation and supination. A cuff was attached to a load cell (JR3 Inc., Woodland, CA) was fastened around the wrist, just below the styloid process. The testing chair included aviator-style cross-straps at the shoulders and waist to secure the participant and minimize extraneous movement. An oscilloscope (Hitachi, VC-6525) was placed in front of the subject to provide a visual feedback to the subject about the required force level. Error bars of ±2.5% of the required force were constructed around the given force level. 2.4. Voluntary force and sEMG recording The general recording area of participants’ right arm was shaved, abraded with NuPrep® , and then cleansed with alcohol to reduce skin–electrode impedance during the recording. A meter (Grass EZM5, Astro-Med Inc., West Warwick, RI) was used to ensure that skin–electrode impedance was below 10 k. A constant-current (maximum 150 mA) source was used to find the motor point using the lowest possible voltage. The cathode and anode electrodes were connected in series with an isolation unit (Grass Telefactor SUI8, Astro-Med, Inc., West Warwick, RI) and a stimulator (Grass Telefactor S88, Astro-Med Inc., West Warwick, RI) to deliver a square-wave pulse, 1 ms in duration at a rate of 10 pps [31]. A self-adhesive anode (Pals Plus, 5.0 cm, Axelgaard, Fallbrook, CA) was secured on the back of the arm (triceps) while a stainless-steel cathode probe (3 mm) was used to systematically explore the skin surface of the biceps brachii until a barely perceptible muscle twitch is observed beneath the skin. Once identified, the motor point was mark and the voltage was increased to observe the twitching muscle fibers so that these locations could be marked and the electrodes placed in line along the same “twitching” muscle fibers. This is the same procedure used previously for recording muscle fiber conduction velocity for which electrode placement on the same fibers is critical [32]. A monopolar configuration was achieved by placing a Ag/AgCl recording electrode (Grass F-E9, Astro-Med Inc., West Warwick, RI) over the electrically identified motor point of the biceps brachii. Immediately next to the electrode on the motor point, the two Ag/AgCl electrodes for the bipolar configuration were place side-by-side so the rims of the plastic casings of all three electrodes were touching to form an array. The reference electrode for the monopolar configuration was placed on the biceps tendon [31]. A 5 cm common ground electrode (CF5000, Axelgaard Manufacturing CO., LTD, Fallbrook, CA) was placed onto the clavicle. The sEMG signals were band-pass filtered (10–1000 Hz) and amplified (Grass P511, Astro-Med Inc., West Warwick, RI) to maxi-

1081

Table 3 The means (M) and standard deviations (SD) for maximal isometric elbow flexion torque, root-mean-square (RMS) amplitude and mean power frequency (MF) for the biceps brachii surface electromyographic (sEMG) signal, skin temperature, and skinelectrode input impedance measured immediately before and after the experiment. Measures

M ± SD Pre-test

Torque (N m) RMS (mV) MF (Hz) Temperature (C) Impedance (k) *

64 1.24 77 31.7 6.32

± ± ± ± ±

29 0.56 14 0.75 2.46

Post-test 62 1.15 79 32.8 5.63

± ± ± ± ±

29 0.56 15 0.97* 2.37

Significant at the p < 0.01.

mize the resolution of the 16 bit analogue-to-digital converted (NI PCI-6052E, National Instruments, Austin, TX). All signals were then sampled at 2048 Hz using a computer-based data acquisition system (DASYLab, DASYTEC National Instruments, Amherst, NH). The digitized signals were stored on a Celeron PC for off-line processing (Dell, Round Rock, TX). 2.5. Data reduction and analysis The computer algorithms for calculating the five sEMG spike shape measures may be found in Gabriel et al. [3]. The following five spike shape variables were calculated (MATLAB, The MathWorks Inc., Natick, MA) for monopolar and bipolar activity within a 1-s window (N = 2048 points) centered in the first half of the contraction: mean spike amplitude (MSA), mean spike frequency (MSF), mean spike slope (MSS), mean spike duration (MSD) and mean number of peaks per spike (MSPPS). The location of this window provided the first stable region of force while avoiding the overshoot that often occurs at submaximal levels. The mean of the three trials was used for statistical analysis. A two-factor (percent MVC × electrode configuration) repeated measures analysis of variance (ANOVA) was used to determine the differences between monopolar and bipolar recordings. Post hoc testing to compare means was accomplished using Tukey’s Honestly Significant Difference test. Orthogonal polynomials for the main and interaction effects were used to identify statistically defined trend components in the means. If a trend component was statistically significant, it was deemed non-trivial only if it accounted more than 15 percent of the total variance. The percent variance accounted for corresponded to a medium effect, and it represented a balance between expected trends in the data and statistical power [33]. The statistical procedures were performed in SYSTAT (SPSS Inc., Chicago, IL) with alpha set at the 0.05 probability level. 3. Results 3.1. Methodological controls Three minutes rest intervals and a balanced experimental design were employed to minimize the effects of fatigue upon the criterion measures. Maximal isometric elbow flexion torque was obtained immediately before and after the experimental contractions to determine if fatigue was present. Traditional time and frequency measures were also calculated for bipolar sEMG activity to provide a basis of comparison to the literature (Table 3). There were no significant differences between the pre- versus postexperimental measures of torque and sEMG activity. Thus, it can be concluded that fatigue did not influence the observed results. The same was true for skin–electrode impedance. There was however a significant increase in skin temperature ( 1.1 ◦ C, p < 0.05) but

1082

D.A. Gabriel / Medical Engineering & Physics 33 (2011) 1079–1085

Fig. 3. Mean spike amplitude means and standard deviations (vertical bars) for the monopolar (circles) and bipolar (squares) electrode configurations.

Fig. 2. Representative monopolar (thick line) and bipolar (thin line) surface electromyographic (sEMG) activity (top panel). The power spectra associated with the monopolar (shaded) and bipolar (un-shaded) sEMG recordings (bottom panel).

no practical significance can be placed upon this slight alteration [34]. 3.2. Spike shape measures Representative monopolar and bipolar recordings for the same time period are presented in Fig. 2(top). The raw recordings illustrate the higher frequency content associated with bipolar recordings. The monopolar recordings (thick line) superimposed upon the bipolar recordings (thin line) have fewer fluctuations over time. The difference in frequency content between the two electrode configurations was evident in their respective power spectra presented below the raw sEMG traces (Fig. 2(bottom)). There were significant main effects and interaction terms for MSA and MSS while MSF, MSD, and MNPPS had only significant main effects (p’s < 0.05). Post hoc testing for MSA obtained by both electrode configurations revealed that each force level was significantly (p’s < 0.01) different from the next as the means increased from 40 to 100 percent MVC. As a result, there was a significant (p < 0.01) linear increase in MSA for both electrode configurations across force levels as illustrated in Fig. 3. However, the monopolar MSA underwent a greater rate of increase (p < 0.01). Mean spike frequency obtained by both electrode configurations exhibited no significant (p’s > 0.01) post hoc differences from 40 to 80 percent MVC, but the 100 percent MVC was significantly (p < 0.01) less than three previous submaximal force levels (see Fig. 4). This pattern of means resulted in a significant (p < 0.01) quadratic decrease in MSF for both electrode configurations. Mean spike duration had and inverse relationship with MSF. That is, there was no significant (p’s > 0.01) post hoc differences from 40 to 80 percent MVC, but the 100 percent MVC condition was significantly (p < 0.01) greater than three previous submaximal force levels. This was true for both electrode configurations (see Fig. 5). As might be expected, the pattern of means also resulted in a significant

Fig. 4. Mean spike frequency means and standard deviations (vertical bars) for the monopolar (circles) and bipolar (squares) electrode configurations.

(p < 0.01) quadratic increase in MSD for both electrode configurations. Post hoc testing for MSS obtained by both electrode configurations revealed that each force level was significantly (p’s < 0.01) different from the next as the means increased from 40 to 100 percent MVC. Fig. 6 shows that both electrode configurations resulted in a significant (p < 0.01) linear increase in MSS across levels of percent MVC. Furthermore, the monopolar MSS increased at a greater

Fig. 5. Mean spike duration means and standard deviations (vertical bars) for the monopolar (circles) and bipolar (squares) electrode configurations.

D.A. Gabriel / Medical Engineering & Physics 33 (2011) 1079–1085

Fig. 6. Mean spike slope means and standard deviations (vertical bars) for the monopolar (circles) and bipolar (squares) electrode configurations.

rate than bipolar MSS across levels of percent MVC (p < 0.01). The MNPPS obtained by both electrode configurations exhibited a decrease from 40 to 80 percent MVC so that the post hoc difference between each force level was significant (p’s < 0.01). There was however a plateau (p > 0.01) between 80 and 100 percent of MVC. The result was a significant (p < 0.01) quadratic trend component (see Fig. 7). 4. Discussion The purpose of this study was to determine if monopolar versus bipolar recordings have a differential effect upon the sEMG spike shape measures. The experimental manipulation involved step contractions at 40, 60, 80, and 100 percent of MVC because it has been previously demonstrated to result in deterministic changes in the five sEMG spike shape measures [3]. The sEMG spike shape measures for both electrode configurations were demonstrated to change in a predictable manner. There were no significant differences between electrode configurations with respect to the pattern of change in means across levels of percent MVC. That, is both electrode configurations resulted in the same type of significant differences between means from 40 to 100 percent MVC. There were however obvious magnitude differences. The following discussion will focus on practical considerations on the impact of monopolar versus bipolar recordings upon sEMG spike shape analysis. Changes in sEMG spike shape will then be discussed with reference to the predictions outline in Table 1.

Fig. 7. Mean number of peaks per spike means and standard deviations (vertical bars) for the monopolar (circles) and bipolar (squares) electrode configurations.

1083

Preliminary analysis of the pre- versus post-experimental measures confirmed the absence of fatigue-related effects upon sEMG activity. Traditional time and frequency measures were calculated for the bipolar sEMG activity during the 100 percent MVC condition to offer a basis of comparison to the literature. The maximal isometric elbow flexion torque observed in this study (64 ± 29 N m) was slightly greater in magnitude than that (56.9 ± 7.8 N m) reported by Beck et al. [35]. However, three out the ten subjects in the study by Beck et al. [35] were females. The root-mean-square (RMS) amplitude values for biceps brachii sEMG were, however, nearly identical. The RMS sEMG amplitude at 100 percent of MVC was 1.24 ± 0.56 mV while Beck et al. [35] observed 1.23 ± 0.22 mV. The mean power frequency (MF) at 100 percent of MVC was also similar. The MF observed in the current work was 77 ± 14 Hz, which was only slightly lower than the 84.9 ± 6.4 Hz reported by Beck et al. [35]. It is reasonable to assume that the same inter-electrode distance (2 cm) and similar experimental protocol between the two studies facilitated comparable results. The individual sEMG spike shape measures were also similar in magnitude to other studies in which a bipolar configuration was used for isometric contractions [3,10]. 4.1. Electrode configuration Volume conducted potentials is a consideration for all sEMG analysis techniques and is never completely eliminated. A bipolar electrode configuration has the advantage of reducing cross-talk and non-propagating waveforms through the common mode rejection function of the associated amplifiers. This is not the case for a monopolar electrode configuration which is more susceptible to common mode signals which decrease the signal-to-noise ratio and make it more difficult to detect changes in muscle electrical activity. However, placing the recording electrode directly over the electrically identified motor point ensures that the signal generator is significantly larger than volume conducted potentials. It is also important to choose a motor-point that is away from a muscle border; this minimizes activity from adjacent muscles and the impact of muscle-tendon end-effects [25–29]. Electrode configuration had no impact on the pattern of change in the means across levels of percent MVC but it did affect mean magnitudes and rates of change, which are critically important to the sensitivity of the spike shape measures. Calder et al. [10] demonstrated that the rate of change in the spike shape measures across levels of percent MVC was able to discriminate between normal controls, individuals with non-specific arm pain, and those at risk for non-specific arm pain. The rate of change in sEMG spike shape measures was important because individuals with nonspecific arm pain exhibited myopathic changes [10]. Myopathic patients recruit more MUs with higher firing rates and faster conduction velocities earlier in the force gradation process than in healthy individuals. This is because the lower number of muscle fibers per motor unit decreases the twitch force of each motor unit [36]. A higher firing rate at any given level of effort is also required to compensate for the reduction in twitch force associated with the loss of muscle fibers [4,15–19]. The same type of reasoning applies to neuropathic conditions. Motor neuron loss leads to a higher degree of motor unit potential summation and MU synchronization since collateral sprouting increases the fiber density within the existing MUs [37–39]. Spike shape measures indicating synchronization would therefore achieve higher values at any given force, and they would increase at a faster rate across levels of percent MVC [4,15–19]. Electrode configuration had a counter-intuitive impact upon the MNPPS measure. That is, the MNPPS was greater for the monopolar versus bipolar recordings, despite the fact that the frequency content for bipolar recordings was greater [23]. The difference between

1084

D.A. Gabriel / Medical Engineering & Physics 33 (2011) 1079–1085

the two electrode configurations is most likely due to a subtle interaction between the signal-to-ratio and the MNPPS detection algorithm. The computer algorithm first detects the 95 percent confidence interval for baseline noise. Peaks are then identified as upwards and downwards deflection that do not cross the isoelectric line, but are greater in amplitude than baseline noise. The reduction in MSA associated with bipolar recordings [24,40] decreased the signal-to-nose ratio to a degree that fewer peaks were identified above baseline noise. Interestingly, Beck et al. [40] failed to observe a higher frequency content for sEMG obtained with bipolar versus monopolar recordings. The main methodological difference is that the current study placed the active electrode for the monopolar configuration over the electrically identified motor point and tested within an electrically isolated room. Beck et al. [40] place the monopolar electrode away from the motor point, then used notch filtering at 60 Hz and its harmonics to remove the effects of electromagnetic noise.

as part of a multivariate statistical model to predict myopathic versus neuropathic disorders and/or disease stage [52,54]. Unfortunately, there was no linkage between any specific measure and underlying pattern of motor unit activity. The use of the multiple measures is a step in the right direction but it is also important to link each measure with a specific pattern of activity to better understand underlying processes of neuromuscular disorders. This is a key and critical missing component of current sEMG processing methods [55]. Spike shape analysis assumes that the summations of a population motor unit action potentials for one type of motor unit behavior results in an interference pattern that is distinctly different from one generated by another type of behavior. Gross analysis of the sEMG signal using recurrence quantification has demonstrated that this is assumption is valid [56]. Unfortunately, recurrence quantification is only a gross pattern analysis method that evaluates the degree to which a signal is deterministic, and not sufficient for evaluating the variety of potential underling physiological mechanisms [57].

4.2. Spike shape analysis 5. Conclusion Ramp and step isometric contractions of the elbow flexors at increasing force levels have been used to study changes in the amplitude and frequency content of the sEMG signal. Some studies have demonstrated that the MF plateaus [41,42], the majority of others have demonstrated that it decreases between 80 and 100 percent MVC [35,43–48]. The current study observed the same findings as Beck et al. [40]. That is, the MF remained stable across force levels until approximately 80 percent of MVC and then it decreased. There is evidence that the magnitude of the increase in MSF across force levels is dependent on inter-electrode distance [35,44] and amount of subcutaneous tissue between the source and the electrode [49,50]. Regardless, spectral compression at the highest force levels has been repeatedly demonstrated for the biceps brachii [35,50], tibialis anterior [51], and the vastus lateralis [40]. There are two possible explanations for spectral compression at 100 percent of MVC. First, motor unit recruitment dominates force generation below 80 percent MVC in the biceps brachii. Ratecoding is then responsible for an increase in force beyond that point [11]. As firing rates within the motor unit pool starts to converge towards their maximum at 100 percent of MVC [13], there is an increased probability of temporal overlap between motor unit action potentials [1]. The second possibility is motor unit synchronization [2]. Gabriel and Kamen [50] recently conducted a modeling and simulation study of these two potential mechanisms. High firing rates and synchronization altered the root-mean-square (RMS) amplitude and mean power frequency (MF) of the simulated sEMG signal in similar ways. However, only synchronization could result in spectral compression to the same degree as observed in the experimental data. In support, Fling et al. [14] reported an increased prevalence of motor unit synchronization in the biceps brachii at 80 percent of MVC compared to 30 percent of MVC, using indwelling needle recordings. While high firing rates and synchronization can result in similar changes to traditional time and frequency measures of the sEMG signal, the two motor unit behaviours produce different alterations in spike shape measures. The observed changes in spike shape measures are consistent with those that indicate synchronization (Table 1). Thus, the advantage of using multiple measures of the sEMG signal conjointly is that no two patterns of motor unit activity produce the same changes in sEMG spike measures. Spike shape analysis builds on previous attempts to translate interference pattern analyses that have been applied to indwelling muscle activity to the sEMG signal [52,53]. While earlier methods have had some degree of diagnostic success, the approach was quite different from what is presented in this paper. Other techniques have used of multiple descriptors of the interference pattern

The two electrode configurations resulted in the same alterations in muscle activity, but the bipolar configuration result in lower mean magnitudes for the sEMG spike shape measures. However, the monopolar configuration with the active electrode placed over the motor point was demonstrated to be more sensitive to changes in muscle activity with increases in force. This is an important consideration because the rate at which muscle electrical activity develops into a full interference pattern is an important qualitative and quantitative diagnostic measure [58,59]. The monopolar configuration is therefore the preferred recording method if the examination room is electrically isolated. Furthermore, voluntary muscle activity can then be assessed following evoked potentials in the same muscle without additional set-up. Conflicts of interest The author must disclose any financial and personal relationships with other people or organizations that could inappropriately influence (bias) their work. Acknowledgement The work was supported by the Natural Sciences and Engineering Research Council of Canada. References [1] Fuglevand AJ, Winter DA, Patla AE. Models of recruitment and rate coding organization in motor-unit pools. J Neurophysiol 1993;70:2470–88. [2] Yao W, Fuglevand AJ, Enoka RM. Motor-unit synchronization increases EMG amplitude and decreases force steadiness of simulated contractions. J Neurophysiol 2000;83:441–52. [3] Gabriel DA, Lester SM, Lenhardt SA, Cambridge EDJ. Analysis of surface EMG spike shape across different levels of isometric force. J Neurosci Methods 2007;159:142–52. [4] Magora A, Gonen B. Computer analysis of the shape of spikes from the electromyographic interference pattern. Electromyography 1970;10:261–71. [5] Fusfeld RD. Analysis of electromyographic signals by measurement of wave duration. Electroencephalogr Clin Neurophysiol 1971;30:337–44. [6] Komi PV, Viitasalo JHT. Signal characteristics of EMG at different levels of muscle tension. Acta Physiol Scand 1976;96:267–76. [7] Vittasalo JHT, Komi PV. Signal characteristics of EMG with special reference to reproducibility of measurements. Acta Physiol Scand 1975;93:531–9. [8] Vittasalo JHT, Komi PV. Signal characteristics of EMG during fatigue. Eur J Appl Physiol 1977;37:111–21. [9] Calder KM, Stashuk DW, McLean L. Motor unit potential morphology differences in individuals with non-specific arm pain and lateral epicondylitis. J NeuroEng Rehabil 2008;5:34. [10] Calder KM, Gabriel DA, McLean L. Differences in EMG spike shape between individuals with non-specific arm pain. J Neurosci Methods 2009;178:148–56.

D.A. Gabriel / Medical Engineering & Physics 33 (2011) 1079–1085 [11] Kukulka CG, Clamann HP. Comparisons of the recruitment and discharge properties of motor units in human brachial biceps an adductor pollicis during isomeric contraction. Brain Res 1981;219:45–55. [12] DeLuca CJ, LeFever RS, McCue MP, Xenakis AP. Behaviour of motor units in different muscles during linearly varying contraction. J Physiol 1982;329:113–28. [13] Erim Z, De Luca CJ, Mineo K, Aoki T. Rank-ordered regulation of motor units. Muscle Nerve 1996;19:563–73. [14] Fling BW, Christie A, Kamen G. Motor unit synchronization in FDI and biceps brachii muscles of strength-trained males. J Electromyogr Kinesiol 2009;19:800–9. [15] Latash ML. Spectral analysis of the electromyogram (EMG) in spinal cord trauma patients. I. Different types of the EMG and corresponding spectra. Electromyogr Clin Neurophysiol 1988;28:319–27. [16] Latash ML. Spectral analysis of the electromyogram (EMG) in spinal cord trauma patients. II. Motor unit and interference EMG power spectra. Electromyogr Clin Neurophysiol 1988;28:329–34. [17] Magora A, Gonen B. Clinical evaluation of the analysis of the shape of electromyographic spike. Electromyography 1972;12:255–66. [18] Magora A, Gonen B. Amplitude of spike in the interpretation of electromyographic recordings. Electromyography 1975;15:377–96. [19] Muro M, Nagata A, Murakami K, Moritani T. Surface EMG power spectral analysis of neuromuscular disorders during isometric contractions. Am J Phys Med 1982;61:244–54. [20] Gabriel DA, Basford JR, An K-N. Assessing fatigue with electromyographic spike parameters. IEEE Eng Med Biol 2001;20:90–6. [21] Bromberg MB, Spiegelberg T. The influence of active electrode placement on CMAP amplitude. Electroencephalogr Clin Neurophysiol 1997;105:385–9. [22] Kleinpenning PH, Gootzen THJM, Van Oosterom A, Stegeman DF. The equivalent source description representing the extinction of an action potential at a muscle fiber ending. Math Biosci 1990;101:41–61. [23] Zipp P. Effect of electrode parameters on the bandwidth of the surface e.m.g. power-density spectrum. Med Biol Eng Comput 1978;16:537–41. [24] Tucker KJ, Türker KS. Triceps surae stretch and voluntary contraction alters maximal M-wave magnitude. J Electromyogr Kinesiol 2007;17:203–11. [25] De Luca CJ, Merletti R. Surface myoelectric cross-talk among muscle of the leg. Electroencephalogr Clin Neurophsyiol 1988;69:568–75. [26] Dimitrova NA, Dimitrov GV, Nikitin OA. Neither high-pass filtering nor mathematical differentiation of the EMG signals can considerably reduce cross-talk. Electromyogr Kinesiol 2002;13:235–46. [27] Mogk JPM, Keir PJ. Crosstalk in surface electromyography of the proximal forearm. J Electromyogr Kinesiol 2003;13:63–71. [28] Solomonow M, Baratta R, Zhou B, Lu Y, Zhu M, Aciero S. Surface and wire EMG crosstalk in neighbouring muscle. Electromyogr Kinesiol 1994;4: 131–42. [29] Winter DA, Fuglevand AJ, Archer SE. Crosstalk in surface electromyography: theoretical and practical estimates. Electromyogr Kinesiol 1994;4:15–26. [30] Farina D, Merletti R. Comparison of algorithms for estimation of EMG variable during voluntary isometric contractions. J Electromyogr Kinesiol 2000;10:337–49. [31] Calder K, Hall LA, Lester SM, Inglis GI, Gabriel DA. Reliability of the biceps brachii m-wave. J NeuroEng Rehabil 2005;2:33. [32] Zwarts MJ, Arendt-Nielsen L. The influence of force and circulation on average muscle fiber conduction velocity during local muscle fatigue. Eur J Appl Physiol 1988;58:278–83. [33] Cohen J. Statistical power analysis for the behavioural sciences. 2nd ed. Hillsdale, NJ: Lawrence Erlbaum Associates; 1988. [34] Rutkove SB, Pseudofacitiation:. A temperature-sensitive phenomenon. Muscle Nerve 2000;23:115–8. [35] Beck TW, Housh TJ, Johnson GO, Weir JP, Cramer JT, Coburn JW, et al. The effects of interelectrode distance on electromyographic amplitude and mean power frequency during isokinetic and isometric muscle actions of the biceps brachii. J Electromyogr Kinesiol 2005;15:482–95. [36] Beardwell A. The spatial organization of motor units and the origin of different types of potential. Ann Phys Med 1967;9:139–57.

1085

[37] Erminio F, Buchthal F, Rosenfalck P. Motor unit territory and muscle fiber concentration in paresis due to peripheral nerve injury and anterior horn cell involvement. Neurol 1959;9:657–71. [38] McComas AJ, Sica RE, Campbell MJ, Upton AR. Functional compensation in partially denervated muscles. J Neurol Neurosurg Psych 1971;34:453–60. [39] Stålberg E. Electrophysiological studies of reinnervation in ALS. Adv Neurol 1982;36:47–59. [40] Beck TW, Housh TJ, Cramer JT, Malek MH, Mielke M, Hendrix R, et al. A comparison of monopolar and bipolar recording techniques for examining the pattern of responses for electromyographic amplitude and mean power frequency versus isometric torque for the vastus lateralis muscle. J Neurosci Methods 2007;166:159–67. [41] Dupont L, Gamet D, Pérot C. Motor unit recruitment and EMG power spectra during ramp contractions of bifunctional muscle. J Electromyogr Kinesiol 2000;10:217–24. [42] Moritani T, Muro M. Motor unit activity and surface electromyogram power spectrum during increasing force of contraction. Eur J Appl Physiol 1987;56:260–5. [43] Bilodeau M, Arsenault AB, Gravel D, Bourbonnais D. Time and frequency analysis of EMG signals of homologous elbow flexors and extensors. Med Biol Eng Comput 1992;30:640–4. [44] Gerdle B, Eriksson N-E, Brundin L. The behaviour of the mean power frequency of the surface electromyogram in biceps brachii with increasing force and during fatigue. With special regard to electrode distance. Electromyogr Clin Neurophysiol 1990;30:483–9. [45] Hagberg M, Ericson B-E. Myoelectric power spectrum dependence on muscular contraction level of elbow flexors. Eur J App Physiol 1982;48:147–56. [46] Ollivier K, Portero P, Maisetti O, Hogrel JY. Repeatability of surface EMG parameters at various isometric contraction levels and during fatigue using bipolar and Laplacian electrode configurations. J Electromyogr Kinesiol 2005;15:466–73. [47] Sanchez JH, Solomonow M, Baratta RV, D’Ambrosia R. Control strategies of the elbow antagonist muscle pair during two types of increasing isometric contractions. J Electromyogr Kinesiol 1993;3:33–40. [48] Sbriccoli P, Bazzucchi I, Responi A, Bernardi M, De Vito G, Felici F. Amplitude and spectral characteristics of biceps brachii sEMG depend upon speed of isometric force generation. J Electromyogr Kinesiol 2003;13:139–47. [49] Farina D, Fosci M, Merletti R. Motor unit recruitment strategies investigated by surface EMG variable. J Appl Physiol 2002;92:235–47. [50] Gabriel DA, Kamen G. Experimental and modeling investigation of spectral compression of biceps brachii SEMG activity with increasing levels of force. J Electromyogr Kinesiol 2009;19:437–48. [51] Lenhardt S, McIntosh K, Gabriel D. The surface EMG–force relationship during isometric dorsiflexion in males and females. Electromyogr Clin Neurophysiol 2009;49:227–34. [52] Toulouse P, Carrault G, Le Rumeur GE, Coatrieux JL. Surface electromyogram automated analysis and Guillain–Barré Syndrome follow up. Electromyogr Clin Neurophysiol 1992;32:51–62. [53] Preece AW, Wimalaratna HSK, Green JL, Churchill E, Morgan HM. Non-invasive quantitative EMG. Electromyogr Clin Neurophysiol 1994;34:81–6. [54] Priez A, Duchene J, Goubel F. Duchenne muscular dystrophy quantification: a multivariate analysis of surface EMG. Med Biol Eng Comput 1992;30:283–91. [55] Zwarts MJ, Drost G, Stegeman DF. Recent progress in the diagnostic use of surface EMG for neurological diseases. J Electromyogr Kinesiol 2000;10:287–91. [56] Fattorini L, Felici F, Flligoi GC, Traballesi M, Farina D. Influence of high motor unit synchronization levels on non-linear and spectral variables of the surface EMG. J Neurosci Methods 2005;143:133–9. [57] Farina D, Fattorini L, Felici F, Filligoi G. Nonlinear surface EMG analysis to detect changes for motor unit conduction velocity a synchronization. J Appl Physiol 2002;93:1753–63. [58] Nandedkar SD, Sanders DB. Measurement of the amplitude of the EMG envelope. Muscle Nerve 1990;13:933–8. [59] Nandedkar SD, Sanders DB, Stålberg EV. Automatic analysis of the electromyographic interference pattern. Part II. Findings in control and in some neuromuscular diseases. Muscle Nerve 1986;9:491–500.