| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
POINT-COUNTERPOINT COMMENTS
University of Colorado at Boulder
MOTOR UNIT RECRUITMENT THRESHOLD
TO THE EDITOR: Motor unit activity can be characterized either with direct physiological measurements or with indirect estimates that are based on the association of another parameter with the physiology. Key physiological properties include contractile characteristics, discharge rate modulation, and recruitment threshold. Classic indirect assessments of motor unit physiology include quantifying the level of myosin ATPase activity and determining the myosin heavy chain composition of muscle fibers, both of which are used as indexes of contractile speed.
Von Tscharner and colleagues (5) propose an alternative indirect index of motor unit activity, one based on a spectral analysis of the surface EMG. According to this scheme, a wavelet approach can be used to identify the response time of muscle twitches in the interference EMG and thereby infer the recruitment patterns of motor units (3). On the basis of the rationale that rapid contractions should involve the preferential recruitment of faster motor units, they report exceptions to the pattern of motor unit recruitment predicted by the size principle (6). This rationale is flawed, however, because it ignores the decrease in recruitment threshold that occurs with an increase in contraction speed (1; see Fig. 1 in Ref. 2), which renders a strategy to bypass lower threshold motor units unnecessary. Indeed, low-threshold motor units are recruited during rapid contractions (4). In the absence of a convincing rationale for a change in recruitment order, the purported recruitment patterns of motor units determined with the wavelet analysis require a more rigorous validation.
REFERENCES
Professor
Simon Fraser University
RECRUITMENT ORDER OF MOTOR UNITS CANNOT BE DETERMINED FROM SURFACE EMG
TO THE EDITOR: Von Tscharner and Nigg (5) separate motor units by their force-generating capacities and task specificity. It is true that each motor unit has a specific force-generating capacity; but task specificity of each motor unit is not a defined physiological parameter. Task-specific recruitment of groups of motor units does occur in many muscles; however, each task group comprises slow and fast motor units and orderly recruitment occurs within each task group (2, 4). No one has shown the existence of a task group that is composed exclusively of one motor unit type.
Surface EMG records low-pass filtered version of muscle-fiber action potentials; the amount of filtering of each action potential depends on the distance of the muscle fiber from the recording electrodes and on the total electrical activity generated by other active muscle fibers. It is well known that the muscle fibers of different motor units within a muscle have a complex distribution and that there is a wide range of shapes of motor unit territories and muscle fibers of different types interdigitate in a normal muscle (1). Due to such complexities, it is difficult to assign a "filtering" weight to the action potential of each muscle fiber of each motor unit recorded by surface EMG electrodes. As a result, one has to agree with the compelling arguments made by Farina (3) against using frequency composition of surface EMG to identify motor unit types during different movements.
REFERENCES
Simon Fraser University
SPECTRAL PROPERTIES OF THE SURFACE EMG CAN CHARACTERIZE MOTOR UNIT RECRUITMENT STRATEGIES
TO THE EDITOR: A critical issue in the Counterpoint argument is the assumption that changes in spectral properties of the sEMG are associated with changes in average muscle fiber conduction velocity (1). The Point argument explains that spectral properties and conduction velocity are independent, and being able to break free from this assumption actually provides an "opportunity" for understanding the recruitment process (4). This issue can be resolved by providing physiological evidence that spectral properties of the EMG are associated with motor unit recruitment, a point raised by the Counterpoint argument (1). Such evidence has been demonstrated using both histochemical and contractile measures of fiber type (2–3, 5): these findings were from fine-wire EMG studies and it is appropriate to consider whether they would also apply to surface EMG signals. A recent simulation study (6), that incorporated distributed end-plate zones (4), volume conductor effects, and continuously distributed conduction velocities (1), found associations between the spectral properties of sEMG and motor unit recruitment strategies: changes in spectral properties were associated with the level of activation when motor units were recruited in an orderly fashion (single-task manner); however, when recurrent inhibition changed the recruitment patterns (as could happen for multi-task actions) then more pronounced changes in sEMG spectra were observed.
The evidence demonstrates that sEMG certainly can characterize MU recruitment strategies and active fiber types. However, it is likely that this will not necessarily be the case for all physiological situations and restricted recruitment tasks.
REFERENCES
Associate Professor
Brock University
THE MOST CONSISTENT FINDING IN SEMG IS INCONSISTENT FINDINGS
TO THE EDITOR: Our paper was cited by Farina (1) to support the lack of relationship between surface electromyographic (sEMG) spectral variables and motor unit behavior. We reported no significant increase in mean power frequency (MF) with increases in isometric elbow flexion force from 40 to 80 percent of maximal voluntary contraction (MVC) (2). However, it is possible to find other studies wherein the increase in MF is moderate but others demonstrate a large increase in MF across force levels. For example, Sbriccoli et al. (4) demonstrated a large increase in median power frequency (MDF) and muscle fiber conduction velocity during ramp increases in isometric elbow flexion force. It has been suggested that MDF increase reflects the recruitment of higher threshold motor units. It also generally accepted that firing rate characteristics dominate the lower end of the frequency spectrum between 10 and 40 Hz (3). Since the sEMG bandwidth trails off appreciably by 200 Hz, the low frequency band between 10 and 40 Hz represents a significant proportion of the total power. In the absence of changes in conduction velocity associated with fatigue, we showed that changes in firing rate characteristics can impact the MNF to an appreciable degree, causing spectral compression. Thus our experimental and modeling work is in agreement with von Tscharner and Nigg (6).
REFERENCES
Associate Professor
University of Colorado at Boulder
IT IS NOT A WAVELET ANALYSIS
TO THE EDITOR: von Tscharner and Nigg (6) argue that the orderly recruitment of motor units can be revealed from the intensity pattern generated by a so-called "wavelet" analysis. However, the "wavelet" analysis of Ref. 5, which possesses none of the properties of a wavelet analysis (2), cannot resolve the subtle changes in the EMG recordings that could separate the different types of motor units. Aside from the common, but incorrect, assumption that there exist distinct types of motor units (1, 4), the interpretation of the "wavelet" coefficients defined in (4) is obfuscated by several factors. First, the "wavelet" transform has a poor time resolution. Because the wavelets are computed by taking the inverse Fourier transform of a series of overlapping bumps, the resulting functions have an infinite extent and are imprecise in the time domain. Second, the frequency resolution is poor because any frequency in the range [0:50] Hz is covered by many bumps, and an oscillatory signal, such as the surface EMG, has its energy spread across many different "wavelet" coefficients. This well-established limitation of wavelet analysis (3) is aggravated here because of the ad hoc design of the wavelets. Third, the mathematical interpretation of the so-called "intensity in time space" is completely unclear. As a result, von Tscharner and colleagues cannot assess the statistical significance of the "wavelet intensity" and have resorted to anecdotal visual inspection of the data as validation of the approach.
REFERENCES
Professor of Applied Electrophysiology
Institute for Fundamental and Clinical Human Movement Sciences
Amsterdam, Nijmegen
NEURAL DRIVE STRATEGIES FROM SHORT-TERM SPECTRAL CHANGES IN SURFACE EMG
TO THE EDITOR: Von Tscharner and Nigg (6) claim that EMG spectral changes directly reflect (de-)recruitment of distinct motor unit populations and of distinct fiber types. Farina (2) rightly tackles this narrow interpretation, although it is not farfetched. Van der Laarse et al. (5) showed a strong inverse relationship between muscle fiber oxygen consumption and fiber diameter. Both parameters vary over a 100-fold range in nature's species. But regrettably for the argument, human muscle fiber-type properties are largely overlapping (2, 3). We could show recruitment of fiber type populations only in a precisely tuned protocol (3).
Then how to explain the results by Von Tscharner and coworkers? Referenced work (6, e.g. Refs. 17, 19), show very low mean frequencies (<20–40 Hz), coinciding with the lowest EMG amplitudes. An obvious explanation for these results is cross-talk from distant agonists or antagonists. It is remarkable that this notion did not show up in this Point:Counterpoint discussion. But then, if cross-talk was or will be ruled out, which neural drive mechanisms could be involved? We argued that animal results should be taken with care, but a functionally relevant topographical shift in recruited fiber population during rat walking is described (e.g., Ref. 4). In the large human muscles this could be reflected in spectral shifts. And with respect to motor unit firing properties, especially the indirect influence of the firing interval (1, 2) should be taken seriously. Firing synchronization, also mentioned (2) would mean a frequency decrease with increasing force demands and EMG amplitude, contrary to what is reported.
REFERENCES
Professor and Director of the Motor Science Research Center
SUISM, Università degli Studi di Torino, Italy
TO THE EDITOR: A huge number of studies tried to correlate EMG variable estimates with muscle fiber-type composition. No consensus can be found in the literature about the power of spectral properties in such a task. Spectral properties of EMG signals are influenced by an incredible high number of confounding factors (1); moreover they represent not a physiological variable directly related to CNS, as happens in the case of muscle fiber conduction velocity, that is however influenced by many other confounding factors. Some works found such correlation associated to muscle biopsies (2, 3), others missed such a finding for mean frequency and found it for conduction velocity (4) or vice versa in case/control studies (5). My opinion is that the proposed Point (6) sounds too far from the recent developments in methodology for recording and analyzing EMG signal. The complexity of the system under study, in fact, requires extreme caution in extracting correlations which sometime can be due to randomness and/or to specific experimental setups. Findings about this issue are quite always obtained in protocols aimed to study fatigue. Differences in EMG manifestations of fatigue are actually physiologically mediated by differences in fiber type composition, but not exclusively. Thus, even if "any changes in the EMG signals during a contraction task can be considered an expression of myoelectric fatigue," it is not always possible to solve the inverse problem associating to a modification of a single variable (i.e., spectral properties) the value of a set of parameter (fiber-type composition).
REFERENCES
Professor
Bulgarian Academy of Sciences
Centre of Biomedical Engineering
THE RESULTS OF "DIRECT EXPERIMENTS" SHOULD BE ACCEPTED WITH CAUTION
TO THE EDITOR: To support their Point, Von Tscharner and Nigg (5) used the results of animal experiments, in which an original elegant method of orderly electrical stimulation of nerve fibers of different diameter was applied (4). However, the results of such "direct experiments" should be accepted with caution. The linear increase in median frequency of power spectrum (PS) of intramuscularly detected M-responses was related to corresponding increasing of muscle fiber propagation velocity (MFPV) in larger motor units (MU). As the nerve stimuli were applied a few centimeters from the muscle, the delay of M-responses used for estimating the MFPV was affected considerably by propagation velocity along the corresponding nerve fibers, whose diameters differed. As a few nerve fibers were stimulated simultaneously in such experiments, desynchronization of individual motor unit potentials (MUPs) in M-responses was larger under stimulation of slower nerve fibers. The effect of desynchronization on EMG signals is more considerable under intramuscular detection (2) when duration of MUP is shorter than under surface detection. The larger MUP desynchronization under stimulation of slower nerve fibers artificially induces a stronger shift of PS of their M-response toward lower frequencies. Thus the PS shift in "direct experiments" has to be affected by differences in velocities along nerve fibers. The effect depends on distance of the stimulating electrode from the motor point.
I support arguments represented by Farina (3) in the Counterpoint. However, one should pay special attention to the effect on PS of intracellular action potential that can change considerably especially under fatigue (1).
REFERENCES
Professor
Bulgarian Academy of Sciences
Centre of Biomedical Engineering
SPECTRAL PROPERTIES OF THE SURFACE EMG CAN NOT PROVIDE RELIABLE INFORMATION ABOUT RECRUITMENT STATE
TO THE EDITOR: The potential produced by motor units (MUs) can be considered (2, 3) as a convolution of the input signal (IS) and corresponding transfer (T) function. IS reflects intracellular action potential (IAP). T function reflects (2) relation between recording electrode position, MU anatomy, volume conductor properties, and muscle fiber propagation velocity (MFPV). In accordance with Borel's convolution theorem, Fourier's transform of convolution is a product of Fourier's transform of both functions.
The widely spread opinions on linear relationship between MFPV and the position of the power spectrum (mean frequency) (6) and on MFPV as mean determinant of the frequency content (4) are misleading. There is a linear relationship between MFPV and spectral characteristics of T function, but the property of one term cannot be accepted as a property of the product (3).
Reducing MFPV through muscle cooling or muscle fatigue, it was found (1) that change in MFPV required to generate equal spectral shift toward lower frequencies was 
10 times greater under reduction of muscle temperature than under fatigue. Irrespective of similar effects of fatigue and muscle temperature reduction on IAP spike, such drastic difference was due to considerable decrease of the IAP after-potential (5) that neutralized the effect of MFPV reduction.
The fast-twitch fibers are fatigable. Thus, in addition to arguments noted in (4), the increased after-potentials typical for fatigue induced, for example, by long-distance running (6), should neutralize the effect of higher MFPV on the EMG spectral characteristics and compromise information about recruitment strategies and muscle fiber type.
REFERENCES
Professor of Engineering of the Neuromuscular System
Politecnico di Torino, Italy
SPECTRAL VARIABLES DO NOT NECESSARILY REFLECT HISTOLOGICAL TYPES OF RECRUITED MOTOR UNITS
TO THE EDITOR: Mean power frequency (MNF) and muscle fiber conduction velocity (CV) are related exclusively when a stable pool of MUs decrease their CV with NO other change. Otherwise, MNF changes no longer reflect only changes of CV but other phenomena as well. CV is related to fiber diameter, which is not necessarily associated to fiber type.
Consider case 1: some small (type I) superficial MUs are active. A few new larger and deeper MUs (type II) are then recruited. The MNF of the surface EMG likely decreases because of the greater depth of the newcomers. Shall we then conclude that type I MUs have been recruited?
Consider case 2: some deep MUs (of either type) are active. New small and superficial (type I) MUs are then recruited. The MNF of the surface EMG increases because of the small tissue filtering affecting the newcomers. Shall we then conclude that type II MUs have been recruited?
Although extreme, cases 1 and 2 are not unrealistic. Changes produced by other factors, such as synchronization or MUAP shape changes, may not be distinguishable from those produced in case 1 or 2. Of course recruitment of superficial type II MUs would increase MNF but many other phenomena would produce the same result. An increase of MNF does NOT necessarily imply recruitment of type II MUs and EMG spectral features are not reflecting percentage of active type I or type II fibers or MUs unless the role played by other factors in determining such features is known.
Academic
The University of Queensland
MOTOR UNIT RECRUITMENT BEHAVIOR IS NOT READILY IDENTIFIABLE FROM EMG SPECTRAL CHANGES
TO THE EDITOR: von Tscharner and Nigg (4) lay foundation to their argument that spectral properties of surface EMG do provide information on motor unit recruitment and muscle fiber type, on the basis of experimental results from running studies (5, 6). They relate time-related changes in frequency components of the surface EMG signal to the recruitment strategy of low- and high-threshold motor neurons during the different phases of the gait cycle.
Farina (1) rightly argues that the link between the spectrum and motor unit behavior is flawed, as no direct measurements were made at times in the stride cycle to identify the recruitment behavior of motor units contributing to the surface signal. A dilemma exists in the fact that in experiments where an orderly recruitment of motor units is expected, such as during ramp contractions (2) or submaximal fatiguing contractions (3), the spectral properties of the surface EMG signal appear to rarely reflect the underlying changes in motor neuron recruitment and rate coding, but are more influenced by some of the confounding factors pointed out by Farina (1). I do not believe it is acceptable to discount many of these concerns on the basis of task specificity and therefore side with the persuasive arguments made by Farina (1).
REFERENCES
Lecturer
School of Medical Sciences
The University of New South Wales
MOTOR UNIT TASK GROUPS
TO THE EDITOR: Von Tscharner and colleagues (6) have used wavelet analyses of the interference electromyogram to infer that fast motor units are selectively recruited, for example, in certain phases of gait. However, they also state that the wavelet analysis is not sufficiently sensitive to detect (or reject) differences during tightly controlled experimental tasks that have confirmed size-ordered recruitment of motor units (4). This is a conundrum; single motor unit recordings from humans are not yet possible during the types of motor tasks for which wavelet analyses have been used to infer selective recruitment of fast motor units.
As noted in other letters, the technique does not demonstrate a departure from size-ordered recruitment of motor units, but it does prompt the question of whether a task group can comprise predominantly fast motor units (5). The constituent motor units for a task group need not conform to particular anatomical boundaries (3) and there is recent evidence that neural drive to motor units may be determined by the mechanical action of muscle fibers (1). However, the indirect means of measuring motor unit type (2) and the limited utility and reliability of the wavelet technique to detect changes in recruitment patterns for some tasks and muscles (6) raise reasonable doubts about whether increased high frequencies in the electromyogram actually correspond to activation of fast motor units (2). To investigate the confounding influence of electrode location relative to active muscle fibers, proponents of the technique might consider using multiple recording electrodes in different positions on a muscle.
REFERENCES
Associate Professor
The University of Toledo
SPECTRAL PROPERTIES OF THE SURFACE EMG DO NOT CHARACTERIZE MOTOR UNIT RECRUITMENT AND RATE CODING
TO THE EDITOR: The surface EMG has been used effectively to qualify and quantify skeletal muscle recruitment patterns during a variety of single- and multi-joint tasks. The tendency to conclusively link the spectral profile of the surface EMG to motor unit activation (temporal and spatial patterns) is attractive as it implies the presence of a non-invasive cause -and-effect measurement technique. On the basis of the available data in the scientific literature, the argument outlined by Farina (3) is not just simply compelling, but is factual. Although our own previous investigations demonstrated synergistic muscle differences in the contraction-intensity-dependent change in the EMG median frequency (4, 5), it is improbable, at best, to derive specific motor unit recruitment patterns from this spectral measure. This fact is additionally exemplified by weak correlations that were demonstrated between the surface-, and needle-detected MUAP during submaximal contractions (1). It is also well understood that technical factors alone, such as skinfold thickness (2), have a significant influence on shaping the surface EMG interference pattern. Notwithstanding the valid suggestion by von Tscharner and Nigg (6) that muscle fiber type will contribute to the shaping the surface EMG signal, it is unlikely that the commonly used median (center) frequency spectral index can be used to distinguish motor unit activation subtleties.
REFERENCES
This article has been cited by other articles:
|
|
 |
|
  G. Layec, A.él. Bringard, Y. Le Fur, C. Vilmen, J.-P. Micallef, S.ép. Perrey, P. J. Cozzone, and D. Bendahan Effects of a prior high-intensity knee-extension exercise on muscle recruitment and energy cost: a combined local and global investigation in humans Exp Physiol, June 1, 2009; 94(6): 704 - 719. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 V. von Tscharner and B. M. Nigg Last Word on Point:Counterpoint: Spectral properties of the surface EMG can characterize/do not provide information about motor unit recruitment strategies and muscle fiber type J Appl Physiol, November 1, 2008; 105(5): 1682 - 1682. [Full Text] [PDF]
|
|
|
|
|
|
D. Farina Last Word on Point:Counterpoint: Spectral properties of the surface EMG can characterize/do not provide information about motor unit recruitment and muscle fiber type J Appl Physiol, November 1, 2008; 105(5): 1683 - 1683. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
