Modeling of surface myoelectric signals--Part II: Model-based signal interpretation

Does the frequency content of the surface mechanomyographic signal reflect motor unit firing rates? A brief review

The purpose of this review is to examine the literature that has investigated the potential relationship between mechanomyographic (MMG) frequency and motor unit firing rates. Several different experimental designs/methodologies have been used to address this issue, including: repetitive electrical stimulation, voluntary muscle actions in muscles with different fiber type compositions, fatiguing and non-fatiguing isometric or dynamic muscle actions, and voluntary muscle actions in young versus elderly subjects and healthy individuals versus subjects with a neuromuscular disease(s). Generally speaking, the results from these investigations have suggested that MMG frequency is related to the rate of motor unit activation and the contractile properties (contraction and relaxation times) of the muscle fibers. Other studies, however, have reported that MMG mean power frequency (MPF) does not always follow the expected pattern of firing rate modulation (e.g. motor unit firing rates generally increase with torque during isometric muscle actions, but MMG MPF may remain stable or even decrease). In addition, there are several factors that may affect the frequency content of the MMG signal during a voluntary muscle action (i.e. muscle stiffness, intramuscular fluid pressure, etc.), independent of changes in motor unit firing rates. Despite the potential influences of these factors, most of the evidence has suggested that the frequency domain of the MMG signal contains some information regarding motor unit firing rates. It is likely, however, that this information is qualitative, rather than quantitative in nature, and reflects the global motor unit firing rate, rather than the firing rates of a particular group of motor units.

A Model for Surface EMG Generation in Volume Conductors With Spherical Inhomogeneities

Most models for surface electromyography (EMG) signal generation are based on the assumption of space-invariance of the system in the direction of source propagation. This assumption implies the same shape of the potential distribution generated by a source in any location along the propagation direction. In practice, the surface EMG generation system is not space invariant and, therefore, the surface signal detected along the direction of the muscle fibers may significantly change shape along the propagation path. An important class of nonspace invariant systems is that of volume conductors inhomogeneous in the direction of source propagation. In this paper, we focused on inhomogeneities introduced by the presence of spheres of different conductivities with respect to the tissue where they are located. This effect may prove helpful to model the presence of glands, vessels, or local changes in the conductivity of a tissue. We present an approximate analytical solution that accounts for an arbitrary number of spheres in an arbitrary complex volume conductor. As a representative example, we propose the solution for a planar layered volume conductor, comprised of fat and muscle layers with spherical inhomogeneities inside the fat layer. The limitations of the approximations introduced are discussed. The model is computationally fast and constitutes an advanced means for the analysis and interpretation of surface EMG signal features.

A simulation model of the surface EMG signal for analysis of muscle activity during the gait cycle

This work describes a model able to synthetize the surface EMG (electromyography) signal acquired from tibialis anterior and gastrocnemious medialis muscles during walking of asymptomatic adult subjects. The model assumes a muscle structure where the volume conductor is represented by multiple layers of anisotropic media. This model originates from analysis of the single fiber action potential characterized by the conduction velocity. The surface EMG of voluntary contraction is calculated by gathering motor unit action potentials estimated by the summation of all activities of muscle fibers assumed to have a uniformly parallel distribution. The parameters related to the gait cycle, such as onset and cessation timings of muscle activation, amplitude of muscle contraction, periods and sequences of motor units' recruitment, are included in the model presented. In addition, the relative positions of the electrodes during gait can also be specified in order to adapt the simulation to the different acquisition settings.

Volume Conduction in an Anatomically Based Surface EMG Model

A finite-element model to simulate surface electromyography (EMG) in a realistic human upper arm is presented. The model is used to explore the effect of limb geometry on surface-detected muscle fiber action potentials. The model was based on magnetic resonance images of the subject's upper arm and includes both resistive and capacitive material properties. To validate the model geometry, experimental and simulated potentials were compared at different electrode sites during the application of a subthreshold sinusoidal current source to the skin surface. Of the material properties examined, the closest approximation to the experimental data yielded a mean root-mean-square (rms) error of the normalized surface potential of 18% or 27%, depending on the site of the applied source. Surface-detected action potentials simulated using the realistic volume conductor model and an idealized cylindrical model based on the same limb geometry were then compared. Variation in the simulated limb geometry had a considerable effect on action potential shape. However, the rate of decay of the action potential amplitude with increasing distance from the fiber was similar in both models. Inclusion of capacitive material properties resulted in temporal low-pass filtering of the surface action potentials. This effect was most pronounced in the end-effect components of action potentials detected at locations far from the active fiber. It is concluded that accurate modeling of the limb geometry, asymmetry, tissue capacitance and fiber curvature is important when the specific action potential shapes are of interest. However, if the objective is to examine more qualitative features of the surface EMG signal, then an idealized volume conductor model with appropriate tissue thicknesses provides a close approximation.

Oscillations in the power spectra of motor unit signals caused by refractoriness variations

The refractory period of a motor unit is an important mechanism that regulates the motor unit firing, and its variation has been found in many physiological cases. In this study, a new observation that an increase in the motor unit refractoriness results in an enhancement of oscillations, or ripple effects, in the motor unit output power density spectra (PDS) has been identified and studied. The effects of the refractoriness variation on the PDS of motor unit firing were investigated on three levels: theoretical modeling, simulation and electromyographic (EMG) experimentation on human subjects. Both theoretical modeling and simulation showed the enhanced oscillations, ripple effects, in MUAPT PDS, given the increase in the refractoriness. It was also found that the extent of the increment in output PDS oscillation could be related to the motor unit size and the mean firing rate of the stimulation. A needle EMG experiment on biceps brachii muscles of five healthy human subjects was carried out during isometric contraction at 20% maximum voluntary contraction (MVC) for 20 s with a fatigue effort proceeded by MVC. The increased oscillations in the PDS of the real MUAPTs were observed with the rising of the motor unit refractoriness due to fatigue. The study gives new information for EMG spectra interpretation, and also provides a potential method for accessing neuromuscular transmission failure (NTF) due to fatigue during voluntary contraction.

Simulation of Surface EMG Signals Generated by Muscle Tissues With Inhomogeneity Due to Fiber Pinnation

Surface electromyographic (EMG) signal modeling has important applications in the interpretation of experimental EMG data. Most models of surface EMG generation considered volume conductors homogeneous in the direction of propagation of the action potentials. However, this may not be the case in practice due to local tissue inhomogeneities or to the fact that there may be groups of muscle fibers with different orientations. This study addresses the issue of analytically describing surface EMG signals generated by bi-pinnate muscles, i.e., muscles which have two groups of fibers with two orientations. The approach will also be adapted to the case of a muscle with fibers inclined in the depth direction. Such muscle anatomies are inhomogeneous in the direction of propagation of the action potentials with the consequence that the system can not be described as space invariant in the direction of source propagation. In these conditions, the potentials detected at the skin surface do not travel without shape changes. This determines numerical issues in the implementation of the model which are addressed in this work. The study provides the solution of the nonhomogenous, anisotropic problem, proposes an implementation of the results in complete surface EMG generation models (including finite-length fibers), and shows representative results of the application of the models proposed.

Multiple-electrode nerve cuffs for low-velocity and velocity-selective neural recording

In the paper, a method using multiple-electrode nerve cuffs is presented that enables electroneurographic signals (ENG) to be recorded selectively by action potential velocity. The theory uses a one-dimensional model of the electrodes in the cuff. Using this model, the transfer function for a single tripole is derived, and it is shown that more than one tripole signal can be recorded from within a cuff. When many tripole signals are available and are temporally aligned by artificial delays and summed, there is a significant increase in the amplitude of the recorded action potential, depending on the cuff length and the action potential velocity, with the greatest gain occurring for low velocities. For example, a cuff was considered that was constrained by surgical considerations to 30 mm between the end electrodes. For action potentials with a velocity of 120 m s(-1), it was shown that, as the number of tripoles increased from one, the peak energy spectral density of the recorded output increased by a factor of about 1.6 with three tripoles, whereas, for 20 m s(-1), the increase was about 19, with ten tripoles. The time delays and summation act as a velocity-selective filter. With consideration of the energy spectral densities at frequencies where these are maximum (to give the best signal-to-noise ratio), the tuning curves are presented for these velocity-selective filters and show that useful velocity resolution is possible using this method. For a 30 mm cuff with nine tripoles, it is demonstrated that it is possible to resolve at least five distinct velocity bands in the range 20-120m s(-1).

Functional asymmetry of pelvic floor innervation and its role in the pathogenesis of fecal incontinence

While the regular and symmetric innervation of the pelvic floor has been regarded as "established" for many years, recent data indicate that asymmetry of innervation of the sphincters may exists and may contribute to the occurrence and severity of incontinence symptoms in case of pelvic floor trauma.

METHODS

A systematic review of published papers on asymmetry of sphincter innervation was performed including studies in healthy volunteers and patients with incontinence. 234 consecutive patients with fecal incontinence were investigated by means of side-separated mass surface EMG from the left and right side anal canal, these data were correlated to clinical and anamnestic findings.

RESULTS

The literature survey indicates that asymmetry of sphincter innervation exists in a subgroup of healthy male and female volunteers, and may be a risk factor to become incontinent in case of trauma. Patients with incontinence in whom asymmetry of sphincter innervation could be shown more frequently reported a history of pelvic floor trauma during childbirth. Childbirth per se but not the number of deliveries predicted sphincter asymmetry. Asymmetrically innervated sphincters show a compromised sphincter function in routine anorectal manometry.

CONCLUSION

Assessment of sphincter innervation asymmetry may be of value in clinical routine testing of patients with incontinence. However, a new technology is needed to replace mass surface EMG by multi-electrode arrays on a sphincter probe. This is one of the goals of the EU-sponsored research project OASIS.

The application of independent component analysis to the multi-channel surface electromyographic signals for separation of motor unit action potential trains: part II—modelling interpretation

The purpose of this article was to investigate whether or not FastICA can separate identical motor unit action potential trains (MUAPTs) of the 8-channel surface electromyographic (sEMG) signals constructed by an sEMG model into the independent components. Firstly, we have examined how much the increase of motor units (MUs) in the simulated sEMG signals influenced the performance on the separation of MUAPTs by kurtosis. The decreased trend of mean kurtosis on both sEMG signals and their independent components were observed as MUs were increased. These data suggested that the separation performance decayed when MUs were increased. Secondary, the differences between the independent components and the principal components have been also applied to the simulated sEMG signals with or without time delay between the sEMG channels. FastICA could successfully separate identical MUAPTs with no time delay but principal component analysis (PCA) could not do so. Against it, both FastICA and PCA could not separate MUAPTs with some time delay. In conclusion, our results suggested that FastICA could separate identical MUAPTs with no time delay into the independent components by FastICA, which might offer a new technique for the separation of interfered MUAP waveforms based on statistical properties of sEMG signal distributions.

The effects of bipolar electrode montage on conduction velocity estimation from the surface electromyogram

This study examines the influence of the bipolar electrode montage on conduction velocity (CV) estimation. Electrode montage refers to the combination of two parameters, the inter-electrode distance (IED), the distance between the two electrodes of a bipolar pair, and the inter-signal distance (ISD), the distance between two bipolar signals used to calculate CV. Data from the biceps brachii (BB) and tibialis anterior (TA) of healthy subjects are analysed. Two approaches are used for CV estimation. The first returns a single value per epoch. The second is based on finding velocity values from individual peaks in the signal and results in a peak velocity (PV) distribution being generated per epoch. It is concluded that CV estimation is significantly dependent on the choice of the (IED, ISD) electrode montage. The main results are that the electrode montage affects (1) the mean PV and CV estimates, (typically P < 0.001), (2) the degree of spatial variability, and (3) the width of the PV distributions. The combination of a small IED and large an ISD is recommended.

Surface electromyogram signal modelling

The paper reviews the fundamental components of stochastic and motor-unit-based models of the surface electromyogram (SEMG). Stochastic models used in ergonomics and kinesiology consider the SEMG to be a stochastic process whose amplitude is related to the level of muscle activation and whose power spectral density reflects muscle conduction velocity. Motor-unit-based models for describing the spatio-temporal distribution of individual motor-unit action potentials throughout the limb are quite robust, making it possible to extract precise information about motor-unit architecture from SEMG signals recorded by multi-electrode arrays. Motor-unit-based models have not yet been proven as successful, however, for extracting information about recruitment and firing rates throughout the full range of contraction. The relationship between SEMG and force during natural dynamic movements is much too complex to model in terms of single motor units.

A Surface EMG Generation Model With Multilayer Cylindrical Description of the Volume Conductor

We propose a model for surface electromyography (EMG) signal generation with cylindrical description of the volume conductor. The model is more general and complete with respect to previous approaches. The volume conductor is described as a multilayered cylinder in which the source can be located either along the longitudinal or the angular direction, in any of the layers. The source is represented as a spatio-temporal function which describes the generation, propagation, and extinction of the intracellular action potential at the end-plate, along the fiber, and at the tendons, respectively. The layers are anisotropic. The volume conductor effect is described as a two-dimensional spatial filtering. Electrodes of any shape or dimension are simulated, forming structures which are described as spatial filters. The analytical derivation which leads to the signal in the temporal domain is performed in the spatial and temporal frequency domains. Numerical issues related to the frequency-based approach are discussed. The descriptions of the volume conductor and of the source are applied to the cases of signal generation from a limb and a sphincter muscle. Representative simulations of both cases are provided. The resultant model is based on analytical derivations and constitutes a step forward in surface EMG signal modeling, including features not described in any other analytical approach.

Simulation analysis of the ability to estimate motor unit propagation velocity non-invasively by different two-channel methods and types of multi-electrodes

Ability to estimate motor unit propagation velocity correctly using different two-channel methods for delay estimation and different non-invasive spatial filters was analysed by simulation. It was established that longitudinal double difference electrodes could be not a better choice than simple bipolar parallel electrodes. Spatial filtration with a new multi-electrode (performing difference between signals detected by two transversal double difference electrodes positioned along the muscle fibres) promises to give the best estimate. Delay estimation between reference points is more preferable than that based on the cross-correlation technique, which is considerably sensitive to the fundamental properties of the muscle fibre extracellular fields. Preliminary averaging and approximation of the appropriate parts of the signals around chosen reference points could reduce the larger noise sensitivity and the effects of local tissue inhomogeneities as well as eliminate the sampling problem. A correct estimate of the propagation velocity could be impossible, even in the case of not very deep motor units (15 or 10 mm, depending on the spatial filter used) with relatively long (about 120 mm) muscle fibres. In the case of fibres with asymmetrical location of the end-plates in respect to the fibre ends, the propagation velocity estimates could be additionally biased above the longer semilength of the motor unit fibres.

Analysis and simulation of changes in EMG amplitude during high-level fatiguing contractions

Changes in surface electromyographic (EMG) amplitude during sustained, fatiguing contractions are commonly attributed to variations in muscle fiber conduction velocity (MFCV), motor unit firing rates, transmembrane action potentials and the synchronization or recruitment of motor units. However, the relative contribution of each factor remains unclear. Analytical relationships relating changes in MFCV and mean motor unit firing rates to the root mean square (RMS) and average rectified (AR) value of the surface EMG signal are derived. The relationships are then confirmed using model simulation. The simulations and analysis illustrate the different behaviors of the surface EMG RMS and AR value with changing MFCV and firing rate, as the level of motor unit superposition varies. Levels of firing rate modulation and short-term synchronization that, combined with variations in MFCV, could cause changes in EMG amplitude similar to those observed during sustained isometric contraction of the brachioradialis at 80% of maximum voluntary contraction were estimated. While it is not possible to draw conclusions about changes in neural control without further information about the underlying motor unit activation patterns, the examples presented illustrate how a combined analytical and simulation approach may provide insight into the manner in which different factors affect EMG amplitude during sustained isometric contractions.

Muscle fatigue and fatigue-related biomechanical changes during a cyclic lifting task

STUDY DESIGN

Electromyographic and biomechanical methods were utilized to investigate correlations between indexes of localized muscle fatigue and changes in the kinematics and kinetics of motion during a cyclic lifting task.

SUMMARY OF BACKGROUND DATA

Recent advances in time-frequency analysis procedures for electromyographicic signal processing provide a new way of studying localized muscle fatigue during dynamic contractions. These methods provide a means to investigate fatigue-related functional impairments in patients with low back pain.

OBJECTIVES

To study the relationship between localized muscle fatigue and the biomechanics of lifting and lowering a weighted box. Fatigue-related changes in the electromyographicic signal of trunk and limb muscles were evaluated and compared to kinematic and kinetic measures in order to determine whether lifting strategy is modified with fatigue.

METHODS

A total of 14 healthy male subjects (26 +/- 5 years) cyclically lifted and lowered a 13 kg box (12 lifts/min) for 4.5 minutes. A 5-second static maximum lifting task was included immediately before and after the cyclic lifting task to measure changes in lifting strength and static electromyographicic fatigue indexes. Electromyographic signals from 14 muscle sites (including paravertebral and limb muscles) were measured. Changes in the electromyographicic Instantaneous Median Frequency, a fatigue index, were computed using time-frequency analysis methods. This index was compared with more standardized measures of fatigue, such as those based on electromyographicic median frequency acquired during a static trunk extension test, subjective fatigue measures, and maximal static lifting strength. Biomechanical measures were gathered using a motion analysis system to study kinematic and kinetic changes during the lifting task.

RESULTS

During the cyclic lifting task, the electromyographic Instantaneous Median Frequency significantly decreased over time in the paravertebral muscles, but not in the limb muscles. Paravertebral electromyographicic Instantaneous Median Frequency changes were consistent with self-reports of fatigue as well as decreases in trunk extension strength. The magnitude of muscle-specific changes in electromyographicic Instantaneous Median Frequency was not significantly correlated with electromyographicic median frequency changes from the static trunk extension task. The load of the box relative to the maximal static lifting strength significantly affected the electromyographicic Instantaneous Median Frequency changes of paravertebral back muscles. Significant changes with fatigue during the task were found in the angular displacements at the knee, hip, trunk, and elbow. These biomechanical changes were associated with increased peak torque and forces at the L4-L5 vertebral segment.

CONCLUSIONS

Our results demonstrate correlation between localized muscle fatigue and biomechanical adaptations that occur during a cyclic lifting task. This new technique may provide researchers and clinicians with a means to investigate fatigue-related effects of repetitive work tasks or assessment procedures that might be useful in improving education, lifting ergonomy, and back school programs. Although both the dynamic and static tasks resulted in spectral shifts in the electromyographicic data, the fact that these methods led to different muscle-specific findings indicates that they should not be considered as equivalent assessment procedures.

Multichannel surface EMG: basic aspects and clinical utility

The generation of the surface electromyogram (sEMG) is described with regard to the properties of the single muscle fiber action potential as source, the physical aspects of volume conduction and recording configuration, and the properties and firing pattern of motor units (MUs). The spatial aspect of the motor unit action potential (MUP) is emphasized in relation to the results of high-density, multichannel sEMG measurements. The endplate zone, depth, size, and position of MUs can be estimated. The use of muscle fiber conduction velocity measurements in channelopathies and the changes in pathological fatigue are described. Using the unique patterns of spatial spread of MUPs over the skin (MU fingerprint), MU classification and the determination of firing moments is done noninvasively. Clinical applications of high-density sEMG measurements are reviewed. Emerging possibilities provided by MUP size and fingerprint measurements in neuromuscular disease and motor control are discussed. We conclude that multichannel sEMG adds unique, and sometimes indispensable, spatial information to our knowledge of the motor unit.

Deconvolution estimation of motor unit conduction velocity distribution

A conduction velocity distribution (CVD) estimator that incorporates volume conductor modeling of the muscle voluntary response is introduced in this paper. The CVD estimates are obtained from two correlation functions, an autocorrelation and a cross, computed from myoelectric signal recorded at the skin surface. The performance of the proposed estimator is evaluated for simulated and experimental data. The study includes assessment of the estimator bias and standard deviation, as well as its sensitivity to errors in the model parameters. Simulations show its good performance in terms of estimator bias. A filtering technique also helps reduce its variance. However, the inaccuracy introduced in the estimation of model parameters considerably deteriorates the estimator performance.

Noninvasive detection of fibrillation potentials in skeletal muscle

The presence of spontaneous muscle activity was determined by analysis of the power spectra of computer-model-generated sequences of spontaneous activity and additive noise. The modeling results identified the frequency band of 100-300 Hz as the band of peak signal-to-noise ratio for the detection of fibrillation potentials. Animal experiments were conducted in which the left sciatic nerves of three rats were transected. Measurements were taken 14 days following surgery with Ag/AgCl gel electrodes on the skin surface. Data was recorded from the gastrocnemius muscle on both the normal and denervated side for all three rats. The normal data and the denervated data yielded no discernible difference in the time-domain. Spectral analysis, however, demonstrated a clear and quantifiable difference between denervated and normal muscle signals. The average difference between the denervated and normal power spectral densities for the frequency band from 100 Hz to 300 Hz was 3.43, 1.90, and 3.02 dB for the three rats. The additional energy observed in the signals recorded from denervated muscles suggests that the single fiber spontaneous muscle activity that occurs in denervated muscle can be noninvasively detected. The potential diagnostic utility of noninvasive fibrillation potential detection is discussed and suggestions for future experiments are made.

Reliability of EMG time-frequency measures of fatigue during repetitive lifting

PURPOSE

To test the short-term and long-term reliability of time-frequency electromyographic (EMG) measures of fatigue during repetitive dynamic lifting and compare it with the reliability of median frequency (MF) EMG measures of fatigue during static lifting.

METHODS

Fourteen' healthy male subjects (26 +/- 5 years) repetitively (12 lifts/min) lifted and lowered a box (29 x 25 x 23 cm, 13 kg) for 4.5 min during 3 different tests on 2 different days. EMG data and the biomechanics of motion were recorded. Before and after dynamic lifting, static maximum lifting tests were performed. At the end of each of the two sessions, subjects performed a static lift at 80% of their maximum lifting force for 30 s.

RESULTS

Significant fatigue-related changes were observed during the lifting exercise via EMG time-frequency analysis at the paravertebral L5, L2, T10, and vastus lateralis (VL) electrode sites. Two parameters for assessing fatigue during dynamic contractions [i.e., the Instantaneous Median Frequency (IMDF) and its time dependent change] were shown to be reproducible both in the short-term (2 h) and long-term (2 wk). The corresponding ICCs reflecting the reproducibility of values between sessions were 96.9% (L5), 98.1% (L2), 90.1% (T10), 96.4% (UT), 98.0% (GM), 89.5% (VL), and 99.0% (BF), respectively. For most EMG recording sites, the reliability of the IMDF measures was not dependent upon the postural strategy that the subject used to accomplish the lifting task or on the subject's strength as measured via the static maximum lifting test. A comparison between the ICC values of the IMDF measures and those of the parameters utilized to assess fatigue during static sustained lifts [i.e., the Median Frequency (MDF) and its change during the test] revealed equally good reproducibility for most EMG recording sites. The respective ICC values that took into account time dependent trends for the IMDF parameter were 87.1% (L5), 62.4% (L2), 90.1% (T10), 0% (UT), 72.7% (GM), 45.4% (VL), and 100% (BF), and for the MDF parameter 94.9% (L5), 73.0% (L2), 80.9% (T10), 100% (UT), 89% (GM), 91.7% (VL), and 90.9% (BF), respectively.

CONCLUSIONS

The time-frequency approach allows one to derive EMG spectral parameters that can be used to monitor muscle fatigue during dynamic real-world tasks such as lifting.

A multiple-layer finite-element model of the surface EMG signal

The effect of skin, muscle, fat, and bone tissue on simulated surface electromyographic (EMG) signals was examined using a finite-element model. The amplitude and frequency content of the surface potential were observed to increase when the outer layer of a homogeneous muscle model was replaced with highly resistive skin or fat tissue. The rate at which the surface potential decreased as the fiber was moved deeper within the muscle also increased. Similarly, the rate at which the surface potential decayed around the surface of the model, for a constant fiber depth, increased. When layers of subcutaneous fat of increasing thickness were then added to the model, EMG amplitude, frequency content, and the rate of decay of the surface EMG signal around the limb decreased, due to the increased distance between the electrodes and the active fiber. The influence of bone on the surface potential was observed to vary considerably, depending on its location. When located close to the surface of the volume conductor, the surface EMG signal between the bone and the source and directly over the bone increased, accompanied by a slight decrease on the side of the bone distal to the active fiber. The results emphasize the importance of distinguishing between the effects of material properties and the distance between source and electrode when considering the influence of subcutaneous tissue, and suggest possible distortions in the surface EMG signal in regions where a bone is located close to the skin surface.

The use of a neural network system in the identification of motor unit characteristics from surface detected action potentials: a simulation study

The surface detected motor unit action potential (MUAP) morphology depends on many physiological and anatomical characteristics of the contracting muscle that are not directly accessible to measurement. In this paper, a neural network based approach is proposed to estimate the motor unit (MU) parameters from a simulated single surface MUAP. We have developed an estimation system that is composed of the following stages: conduction velocity estimation, signal dimension reduction, MU parameters estimation, and number of MU fibres estimation. The parameter estimation stage employs four multilayer neural networks trained on simulated MUAPs corresponding to various ranges of MU parameters. In the estimation mode, this module produces four MU parameters sets. The selected set of the five muscle characteristics is that which minimises an error criterion on a signal reconstructed from the estimated parameters. The proposed system is tested with several simulated MUAPs signals with additive white noise in order to evaluate its performance. It is shown that the technique performs well when the signal to noise ratio is greater than 20 dB.

Electromyogram median frequency, spectral compression and muscle fibre conduction velocity during sustained sub-maximal contraction of the brachioradialis muscle

Changes in the median frequency of the power spectrum of the surface electromyogram (EMG) are commonly used to detect muscle fatigue. Previous research has indicated that changes in the median frequency are related to decreases in muscle fibre conduction velocity (MFCV) during sustained fatiguing contractions. However, in experimental studies the median frequency has been consistently observed to decrease by a relatively greater amount than MFCV. In this paper, a new estimate of EMG frequency compression, the Spectral Compression Estimate (SCE), is compared with the median frequency of the EMG power spectrum, the median frequency of the EMG amplitude spectrum and MFCV measured during sustained, isometric, fatiguing contractions of the brachioradialis muscle at 30, 50 and 80% maximum voluntary contraction (MVC). The SCE is found to provide a better estimate of the observed changes in MFCV than the median frequency of either the EMG power spectrum or EMG amplitude spectrum.

Assessment of single motor unit conduction velocity during sustained contractions of the tibialis anterior muscle with advanced spike triggered averaging

This paper describes an improved spike triggered averaging technique for the assessment of control properties and conduction velocity (CV) of single motor units (MUs) of the tibialis anterior muscle during voluntary muscle contractions. The method is based on the detection of multi-channel surface EMG signals (with linear electrode arrays) and intramuscularly recorded single MU action potentials (MUAPs). Intramuscular electrodes were inserted in the muscle taking into account the MU structural properties (innervation zone, tendon locations, length of the fibers), assessed by the linear array surface EMG detection technique. An algorithm for intramuscular EMG signal decomposition is used to identify single MUAP trains. The MUAPs detected by the intramuscular EMG decomposition algorithm were used to trigger and average the multi-channel EMG signals. CV of single averaged surface MUAPs was estimated by the use of advanced signal processing methods based on multi-channel recordings which allow to consistently reduce the variance of CV estimates compared with traditional two channel delay estimators. The number of averaged potentials can thus be limited, resulting in high temporal resolution CV estimates. The developed technique was tested on recordings from the tibialis anterior muscle in 11 volunteers during fatigue. It was shown that the method allows the assessment of single MU CV changes (fatigue) as small as 0.1 m/s with less than 2 s data epochs. The method allows reliable assessment of firing rate and conduction properties of single MUs with applications for the investigation of central and peripheral fatigue mechanisms.

A model of the muscle action potential for describing the leading edge, terminal wave, and slow afterwave

The leading edge, terminal wave, and slow afterwave of the motor-unit action potential (MUAP) are produced by changes in the strength of electrical sources in the muscle fibers rather than by movement of sources. The latencies and shapes of these features are, therefore, determined primarily by the motor-unit (MU) architecture and the intracellular action potential (IAP), rather than by the volume-conduction characteristics of the limb. We present a simple model to explain these relationships. The MUAP is modeled as the convolution of a source function related to the IAP and a weighting function related to the MU architecture. The IAP waveform is modeled as the sum of a spike and a slow repolarization phase. The MU architecture is modeled by assuming that the individual fibers lie along a single equivalent axis but that their action potentials have dispersed initiation and termination times. The model is illustrated by simulating experimentally recorded MUAPs and compound muscle action potentials.

Finite element modeling of electromagnetic signal propagation in a phantom arm

Improving the control of artificial arms remains a considerable challenge. It may be possible to graft remaining peripheral nerves in an amputated limb to spare muscles in or near the residual limb and use these nerve-muscle grafts as additional myoelectric control signals. This would allow simultaneous control of multiple, degrees of freedom (DOF) and could greatly improve the control of artificial limbs. For this technique to be successful, the electromyography (EMG) signals from the nerve-muscle grafts would need to be independent of each other with minimal crosstalk. To study EMG signal propagation and quantify crosstalk, finite element (FE) models were developed in a phantom-arm model. The models were validated with experimental data collected by applying sinusoidal excitations to a phantom-arm model and recording the surface electric potential distribution. There was a very high correlation (r > 0.99) between the FEM data and the experimental data, with the error in signal magnitude generally less than 5%. Simulations were then performed using muscle dielectric properties with static, complex, and full electromagnetic solvers. The results indicate that significant displacement currents can develop (> 50% of total current) and that the fall-off of surface signal power varies with how the signal source is modeled. Index Terms-Control, electromyography (EMG), finite element (FE), modeling, prosthesis.

Time-frequency parameters of the surface myoelectric signal for assessing muscle fatigue during cyclic dynamic contractions

The time-dependent shift in the spectral content of the surface myoelectric signal to lower frequencies has proven to be a useful tool for assessing localized muscle fatigue. Unfortunately, the technique has been restricted to constant-force, isometric contractions because of limitations in the processing methods used to obtain spectral estimates. A novel approach is proposed for calculating spectral parameters from the surface myoelectric signal during cyclic dynamic contractions. The procedure was developed using Cohen class time-frequency transforms to define the instantaneous median and mean frequency during cyclic dynamic contractions. Changes in muscle length, force, and electrode position contribute to the nonstationarity of the surface myoelectric signal. These factors, unrelated to localized fatigue, can be constrained and isolated for cyclic dynamic contractions, where they are assumed to be constant for identical phases of each cycle. Estimation errors for the instantaneous median and mean frequency are calculated from synthesized signals. It is shown that the instantaneous median frequency is affected by an error slightly lower than that related to the instantaneous mean frequency. In addition, we present a sample application to surface myoelectric signals recorded from the first dorsal interosseous muscle during repetitive abduction/adduction of the index finger against resistance. Results indicate that the variability of the instantaneous median frequency is related to the repeatability of the biomechanics of the exercise.

A novel approach for precise simulation of the EMG signal detected by surface electrodes

We propose a new electromyogram generation and detection model. The volume conductor is described as a nonhomogeneous (layered) and anisotropic medium constituted by muscle, fat and skin tissues. The surface potential detected in space domain is obtained from the application of a two-dimensional spatial filter to the input current density source. The effects of electrode configuration, electrode size and inclination of the fibers with respect to the detection system are included in the transfer function of the filter. Computation of the signal in space domain is performed by applying the Radon transform; this permits to draw considerations about spectral dips and clear misunderstandings in previous theoretical derivations. The effects of generation and extinction of the action potentials at the fiber end plate and at the tendons are included by modeling the source current, without any approximation of its shape, as a function of space and time and by using again the Radon transform. The approach, based on the separation of the temporal and spatial properties of the muscle fiber action potential and of the volume conductor, includes the capacitive tissue properties.

Estimation of single motor unit conduction velocity from surface electromyogram signals detected with linear electrode arrays

This work addresses the problem of estimating the conduction velocity (CV) of single motor unit (MU) action potentials from surface EMG signals detected with linear electrode arrays during voluntary muscle contractions. In ideal conditions, that is without shape or scale changes of the propagating signals and with additive white Gaussian noise, the maximum likelihood (ML) is the optimum estimator of delay. Nevertheless, other methods with computational advantages can be proposed; among them, a modified version of the beamforming algorithm is presented and compared with the ML estimator. In real cases, the resolution in delay estimation in the time domain is limited because of the sampling process. Transformation to the frequency domain allows a continuous estimation. A fast, high-resolution implementation of the presented multichannel techniques in the frequency domain is proposed. This approach is affected by a negligible decrease in performance with respect to ideal interpolation. Application of the ML estimator, based on two-channel information, to ten firings of each of three MUs provides a CV estimate affected by a standard deviation of 0.5 m s(-1); the modified beamforming and ML estimators based on five channels provide a CV standard deviation of less than 0.1 m s(-1) and allow the detection of statistically significant differences between the CVs of the three MUs. CV can therefore be used for MU classification.

Spectral compression of the electromyographic signal due to decreasing muscle fiber conduction velocity

Spectral compression of the electromyographic (EMG) signal, due largely to decreasing muscle fiber conduction velocity, is commonly used as an indication of muscle fatigue. Current methods of estimating conduction velocity using characteristic frequencies such as the median frequency of the power spectrum, are based on an assumption of uniform spectral compression. To examine changes in the EMG frequency spectrum during fatigue, muscle fiber conduction velocity was measured during sustained, isometric contractions of the biceps brachii. Compression of the EMG power and amplitude spectra was simultaneously examined using the median frequency and an alternative method-the spectral distribution technique. The spectral distribution technique consistently gave a better estimate of the relative change in muscle fiber conduction velocity than either of the median frequencies. This was further examined using a physiologically based EMG simulation model, which confirmed these findings. The model indicated that firing statistics can significantly influence spectral compression, particularly the behavior of characteristic frequencies in the vicinity of the firing rates. The relative change in the median frequency, whether of the amplitude or frequency spectrum, was consistently greater than the relative change in conduction velocity. The most accurate indication of the relative change in conduction velocity was obtained by calculating the mean shift in the midfrequency region of the EMG amplitude spectrum using the spectral distribution technique.

Noninvasive estimation of motor unit conduction velocity distribution using linear electrode arrays

Determining the conduction velocity of motor unit action potentials is one of the most important problems in surface electromyography. The estimate of one average conduction velocity value depends on a variety of uncontrollable factors. More meaningful information is obtained from the estimation of the distribution of the different delays in the myoelectric signals. A solution to the problem is the separation and characterization of the individual components propagating at different velocities. A technique, based on surface electrode array recording, is proposed to estimate motor unit conduction velocity distribution. The method consists in the identification of the single action potentials in the time scale domain (with the continuous wavelet transform) and in the estimation of their conduction velocities based on the beamforming algorithm. The performances of the technique have been evaluated using simulated and real myoelectric signals. The results demonstrate that the technique is accurate and reliable. The method may be useful for the diagnosis of neuromuscular disorders, for the monitoring of muscle fatigue and for noninvasive investigation of individual motor units.

Modeling of surface myoelectric signals--Part II: Model-based signal interpretation

Experimental electromyogram (EMG) data from the human biceps brachii were simulated using the model described in [10] of this work. A multichannel linear electrode array, spanning the length of the biceps, was used to detect monopolar and bipolar signals, from which double differential signals were computed, during either voluntary or electrically elicited isometric contractions. For relatively low-level voluntary contractions (10%-30% of maximum force) individual firings of three to four-different motor units were identified and their waveforms were closely approximated by the model. Motor unit parameters such as depth, size, fiber orientation and length, location of innervation and tendonous zones, propagation velocity, and source width were estimated using the model. Two applications of the model are described. The first analyzes the effects of electrode rotation with respect to the muscle fiber direction and shows the possibility of conduction velocity (CV) over- and under-estimation. The second focuses on the myoelectric manifestations of fatigue during a sustained electrically elicited contraction and the interrelationship between muscle fiber CV, spectral and amplitude variables, and the length of the depolarization zone. It is concluded that a) surface EMG detection using an electrode array, when combined with a model of signal propagation, provides a useful method for understanding the physiological and anatomical determinants of EMG waveform characteristics and b) the model provides a way for the interpretation of fatigue plots.