Unsupervised pattern recognition for the classification of EMGsignals

The shapes and firing rates of motor unit action potentials (MUAPs) in an electromyographic (EMG) signal provide an important source of information for the diagnosis of neuromuscular disorders. In order to extract this information from EMG signals recorded at low to moderate force levels, it is required: i) to identify the MUAPs composing the EMG signal, ii) to classify MUAPs with similar shape, and iii) to decompose the superimposed MUAP waveforms into their constituent MUAPs. For the classification of MUAPs two different pattern recognition techniques are presented: i) an artificial neural network (ANN) technique based on unsupervised learning, using a modified version of the self-organizing feature maps (SOFM) algorithm and learning vector quantization (LVQ) and ii) a statistical pattern recognition technique based on the Euclidean distance. A total of 1213 MUAPs obtained from 12 normal subjects, 13 subjects suffering from myopathy, and 15 subjects suffering from motor neuron disease were analyzed. The success rate for the ANN technique was 97.6% and for the statistical technique 95.3%. For the decomposition of the superimposed waveforms, a technique using crosscorrelation for MUAP's alignment, and a combination of Euclidean distance and area measures in order to classify the decomposed waveforms is presented. The success rate for the decomposition procedure was 90%     

A model for the generation of synthetic intramuscular EMG signals to test decomposition algorithms

As more and more intramuscular electromyogram (EMG) decomposition programs are being developed, there is a growing need for evaluating and comparing their performances. One way to achieve this goal is to generate synthetic EMG signals having known features. Features of interest are: the number of channels acquired (number of detection surfaces), the number of detected motor unit action potential (MUAP) trains, their time-varying firing rates, the degree of shape similarity among MUAPs belonging to the same motor unit (MU) or to different MUs, the degree of MUAP superposition, the MU activation intervals, the amount and type of additive noise. A model is proposed to generate one or more channels of intramuscular EMG starting from a library of real MUAPs represented in a 16-dimensional space using their Associated Hermite expansion. The MUAP shapes, regularity of repetition rate, degree of superposition, activation intervals, etc. may be time variable and are described quantitatively by a number of parameters which define a stochastic process (the model) with known statistical features. The desired amount of noise may be added to the synthetic signal which may then be processed by the decomposition algorithm under test to evaluate its capability of recovering the signal features.     

上肢康复机器人实验平台肌电信号采集系统的设计

表面肌电信号是肌肉收缩的同时伴随的一种电压信号,是一种复杂的表皮下肌电信号活动在皮肤表面处的时间和空间上综合得出的结果,能够反映出神经、肌肉的功能状态。正是其在相同肌群规律性和在不同肌群差异性,使得利用肌电信号作为人机接口来控制上肢康复机器人成为可能。本文的主要内容是肌电信号采集系统的设计,将从硬件电路以及软件设计两部分进行阐述。其中硬件电路主要由表面电极、信号调理、NI-USB-6210数据采集卡和上位机四部分组成;系统软件采用虚拟仪器开发平台LabVIEW编程,完成肌电信号实时采集、滤波处理、数据存储等功能。     

Spatial Filtering of Noninvasive Multielectrode EMG: Part I?Introduction to Measuring Technique and Applications

Complementary to its conventional applications, surface EMG is also suited to gain more detailed information on the functional state of a muscle, when measurement configurations with smaller pickup areas are used. A new category of suitable measurement configurations is obtained by application of the spatial filtering principle to electromyography. In a spatial filter unit, the signals of several recording electrodes are combined to form one output signal channel. The filter characteristic is determined by the weighting factors used and by the geometrical arrangement of the electrodes. Extended multielectrode arrays and multichannel recording make possible the detection of correlated excitations at different sites of the muscle. Even in high levels of muscle contraction, single motor unit impulses that are suitably shaped by filtering can be repeatedly recognized in the surface EMG signal. In clinical studies, pathologically shaped impulses have been identified indicating multiple innervation zones. The initiation and the propagation of excitation within single motor units can be detected with improved accuracy even from very small muscles.     

A Nonstationary Model for the Electromyogram

A theoretical model of the electromyographic (EMG) signal has been developed. In the model, the neural pulse train inputs were considered to be point processes which passed through linear, time-invariant systems that represented the respective motor unit action potential. The outputs were then summed to produce the EMG. It was assumed, that in the production of muscle force, the controlled parameter was the number of active motor units, n(t). The model then showed that the EMG can be represented as an amplitude modulation process of the form EMG = [Kn(t)1/2 w(t) with the stochastic process, w(t), having the spectral and probability characteristics of the EMG during a constant contraction. Various assumptions made in the model development have been verified by experiments.     

Automatic decomposition of selective needle-detected myoelectricsignals

A procedure for the storage and documentation of myoelectric signals has been developed that consists of a selective needle signal detection protocol, a data collection-compression routine, an adaptive signal decomposition algorithm, and an error filter. The collection-compression routine stores only fixed-length signal epochs that contain motor unit action potentials (MUAPs) detected during individual motor unit firings. The decomposition algorithm assigns the collected MUAPs to candidate motor units, based on template matching using power-spectrum domain features and firing-time criteria calculated from the motor units' firing statistics. Power spectrum features allow the use of Nyquist sampling rates and remove the need for template alignment. The algorithm is adaptive and attempts to minimize dependent errors. The error filter, using firing statistics, accounts for unresolved superpositions and other decomposition errors. Using a standard TECA single-fiber needle electrode, signal recorded during isometric, constant, or slow force-varying contractions of up to 50% of the maximal voluntary contraction level, have been successfully analyzed     

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.     

Gesture-based control and EMG decomposition

This paper presents two probabilistic developments for the use with electromyograms (EMGs). First described is a neuroelectric interface for virtual device control based on gesture recognition. The second development is a Bayesian method for decomposing EMGs into individual motor unit action potentials (MUAPs). This Bayesian decomposition method allows for distinguishing individual muscle groups with the goal of enhancing gesture recognition. All examples presented rely upon sampling EMG data from a subject's forearm. The gesture-based recognition uses pattern recognition software that has been trained to identify gestures from among a given set of gestures. The pattern recognition software consists of hidden Markov models, which are used to recognize the gestures as they are being performed in real time from moving averages of EMGs. Two experiments were conducted to examine the feasibility of this interface technology. The first replicated a virtual joystick interface, and the second replicated a keyboard. Moving averages of EMGs do not provide an easy distinction between fine muscle groups. To better distinguish between different fine motor skill muscle groups, we present a Bayesian algorithm to separate surface EMGs into representative MUAPs. The algorithm is based on differential variable component analysis, which was originally developed for electroencephalograms. The algorithm uses a simple forward model representing a mixture of MUAPs as seen across multiple channels. The parameters of this model are iteratively optimized for each component. Results are presented on both synthetic and experimental EMG data. The synthetic case has additive white noise and is compared with known components. The experimental EMG data were obtained using a custom linear electrode array designed for this study.     

The Effect of Electromyographic Jitter on Single Motor Unit EMG Potentials

Variability between successive discharges of the single motor unit potential in the biceps brachii muscle, due to electromyographic (EMG) jitter, has been investigated. This jitter results from random arrival times of single fiber potentials at the electrode. A computer model has been used to generate single motor unit potentials incorporating the effects of EMG jitter. A computed variance peak was found in the fast rising edge of the motor unit potential for electrode sites outside of the motor unit territory. This peak was also observed in experimental data recorded from human subjects. The peak variance outside of the motor unit territoxy has also been mathematically related to the number of fibers in the motor unit, jitter, and the slope of the mean action potential at the center of the fast rising edge.     

An Analysis of the Electromyogram by Fourier, Simulation and Experimental Techniques

The electromyogram of a single motor unit is studied by considering it as a time function defined by a convolution integral where a point process input passes through a filter whose impulse response is the shape of a single motor unit action potential. The interspike intervals are assumed to be normally distributed, independent random variables. Simulation is performed on a digital computer. The theoretical analysis shows that the absolute value of the ensemble average of the Fourier transform of the simulated EMG approaches the absolute value of the Fourier transform of the motor unit potential. This has been confirmed by simulation except at the very low end of the spectrum. These results are compared with the Fourier transforms of the recorded surface EMG data from human muscles.     

Nonlinear Spatial Filtering of Multichannel Surface Electromyogram Signals During Low Force Contractions

This study introduces the application of nonlinear spatial filters to help identify single motor unit discharge from multiple channel surface electromyogram (EMG) signals during low force contractions. The nonlinear spatial filters simultaneously take into account the instantaneous amplitude and frequency information of a signal. This property was used to enhance motor unit action potentials (MUAPs) in the surface EMG record. The advantages of nonlinear spatial filtering for surface MUAP enhancement were investigated using both simulation and experimental approaches. The simulation results indicate that when compared with various linear spatial filters, nonlinear spatial filtering achieved higher SNR and higher kurtosis of the surface EMG distribution. Over a broad range of SNR and kurtosis levels for the input signal, nonlinear spatial filters achieved at least 32 times greater SNR and 11% higher kurtosis for correlated noise, and at least 15 times greater SNR and 1.7 times higher kurtosis for independent noise, across electrode array channels. The improvements offered by nonlinear spatial filters were further documented by applying them to experimental surface EMG array recordings. Compared with linear spatial filters, nonlinear spatial filters achieved at least nine times greater SNR and 25% higher kurtosis. It follows that nonlinear spatial filters represent a potentially useful supplement to linear spatial filters for detection of motor unit activity in surface EMG at low force contractions.     

Motor Unit Firing Rate During Static Contraction Indicated by the Surface EMG Power Spectrum

In this study, power spectral density functions (PSDF's) were computed of interference EMG of various facial and jaw-elevator muscles during nonfatiguing submaximal static contractions, recorded with surface electrodes. A distinct peak was found in the PSDF's in the frequency region below 40 Hz. It was shown that the peak was due to genuine EMG activity and that it could not be considered as an artifact, which was caused by electrode displacements during contraction. An increase of contraction strength resulted in a shift of the peak to higher frequencies and a decrease of peak amplitude relative to the power spectral estimates above 40 Hz, which were shown to be determined by the shape of the motor unit (MU) action potentials. In accordance with mathematical models of the EMG PSDF, it was demonstrated that the peak indicates the dominant firing rate of the sampled MU's. Our results suggest that this can be defined as the firing rate of the first recruited low-threshold MU's, which may be expected to dominate the interference EMG signal because of their preponderance in number. The data further suggest that the peak can be more readily observed in PSDF's of facial and jaw-elevator muscles than in PSDF's of limb muscles. This might be related to differences in MU firing statistics.     

Motor unit number index (MUNIX)

The surface-recorded compound muscle action potential (CMAP) and electromyographic (EMG) interference pattern is used to compute the motor unit number index (MUNIX). The MUNIX demonstrated all known changes in the number of motor units in normal subjects, and in patients with amyotrophic lateral sclerosis (ALS). In normal subjects MUNIX decreased slightly with age and showed excellent reproducibility. In many ALS patients MUNIX was reduced even when the CMAP was normal. Lower MUNIX values were seen in weaker muscles. This is a noninvasive method that requires minimal electrical stimulation. It is performed in less than 5 min. This makes it suitable for serial EMG investigations.     

Simulation of surface EMG signals for a multilayer volume conductor with a superficial bone or blood vessel

This study analytically describes surface electromyogram (EMG) signals generated by a planar multilayer volume conductor constituted by different subdomains modeling muscle, bone (or blood vessel), fat, and skin tissues. The bone is cylindrical in shape, with a semicircular section. The flat portion of the boundary of the bone subdomain is interfaced with the fat layer tissue, the remaining part of the boundary is in contact with the muscle layer. The volume conductor is a model of physiological tissues in which the bone is superficial, as in the case of the tibia bone, backbone, and bones of the forearm. The muscle fibers are considered parallel to the axes of the bone, so that the model is space invariant in the direction of propagation of the action potential. The proposed model, being analytical, allows faster simulations of surface EMG with respect to previously developed models including bone or blood vessels based on the finite-element method. Surface EMG signals are studied by simulating a library of single-fiber action potentials (SFAP) of fibers in different locations within the muscle domain, simulating the generation, propagation, and extinction of the action potential. The decay of the amplitude of the SFAPs in the direction transversal to the fibers is assessed. The decay in the direction of the bone has a lower rate with respect to the opposite direction. Similar results are obtained by simulating motor unit action potentials (MUAPs) constituted by 100 fibers with territory 5 mm2. M waves and interference EMG signals are also simulated based on the library of SFAPs. Again, the decay of the amplitude of the simulated interference EMG signals is lower approaching the bone with respect to going farther from it. The findings of this study indicate the effect of a superficial bone in enhancing the EMG signals in the transversal direction with respect to the fibers of the considered muscle. This increases the effect of crosstalk. The same mathematical method used to simulate a superficial bone can be applied to simulate other physiological tissues. For example, superficial blood vessels (e.g., basilic vein, brachial artery) can influence the recorded EMG signals. As the electrical conductivity of blood is high (it is of the same order as the longitudinal conductivity in the muscle), the effect on EMG signals is opposite compared to the effect of a superficial bone.     

Evaluation of the Effectiveness of EMG Parameters in the Study of Neurogenic Diseases-A Statistical Approach Using Clinical and Simulated Data

Of interest here is the problem of determining to what extent combinations of parameters derived from the EMG signal allow 1) discriminating two subclasses of neurogenic myopathies, and 2) recognizing different morphologies of the motor unit action potential underlying a measured EMG signal. EMG signals measured on clinical subjects and computer-simulated EMG signals were collected in a database and used cooperatively in this study. Suitable statistical models were developed which allow testing hypotheses on the role of accepted EMG parameters for the two purposes named above, and deriving new suitable combinations of EMG parameters. Results support the hypothesis that frequency-domain parameters are very clearly related to the morphology of the motor unit action potential. However, the attempt to use them in order to discriminate the two pathologic subclasses considered appears to be jeopardized by the fact that the signal may be measured in territories which do not reflect the morphology of the motor unit action potential dominant in such subclasses. On the basis of time-domain parameters, a significant discrimination was obtained between the two subclasses, and such discrimination is related mainly to a time-domain parameter which has already proved successful in the discrimination between myopathic and normal subjects. Data corroborate the hypothesis that the diagnostic yield improves when time-domain EMG parameters are measured at recruitment.     

Simulation Techniques in Electromyography

A motor unit action potential (MUAP) recorded in clinical electromyography (EMG) is the spatial and temporal summation of the action potentials (AP's) from all muscle fibers in a motor unit (MU). An important determinant of MUAP waveform characteristics is the size of the recording electrode. In this paper, we have described the use of a modified line source model of single muscle fiber action potentials to simulate MUAP's as recorded by single fiber (SF) EMG, concentric needle (CN) EMG, and macro-EMG electrodes. Results indicate that SFEMG recordings from a normal MU contain mainly the AP's of the closest one to three muscle fibers of the MU. The amplitude, area, and duration of the simulated CNEMG MUAP's are determined mainly by the number and size of muscle fibers within a semicircular territory of 0.5, 1.5, and 2.5 mm, respectively, around the tip of the electrode. The amplitude and area of simulated macro-EMG MUAP's increase with the number of muscle fibers in the MU.     

Spectral Analysis of Surface Motor Unit Action Potentials and Surface Interference Electromyogram

Surface interference electromyograms (EMG's) were recorded from the tibial muscle of a healthy subject during 50 percent maximal contraction and single motor unit action potentials (MUAP's) were isolated by averaging from the interference pattern. The formation of the EMG was simulated by summing isolated MUAP's according to the statistical properties of the corresponding motor unit discharges. Power spectral density functions (PSDF's) were finally computed for single MUAP's as well as for simulated and experimental EMG's and compared with each other.     

Bayesian network modeling for discovering "dependent synergies" among muscles in reaching movements

The coordinated activities of muscles during reaching movements can be characterized by appropriate analysis of simultaneously-recorded surface electromyograms (sEMGs). Many recent sEMG studies have analyzed muscle synergies using statistical methods such as Independent Component Analysis, which commonly assume a small set of influences upstream of the muscles (e.g., originating from the motor cortex) produce the sEMG signals. Traditionally only the amplitude of the sEMG signal was investigated. Here, we present a fundamentally different approach and model sEMG signals after the effects of amplitude have been minimized. We develop the framework of Bayesian networks (BNs) for modeling muscle activities and for analyzing the overall muscle network structure. Instead of assuming that synergies may be independently activated, we assume that neuronal activity driving a given muscle may be conditionally dependent upon neurons driving other muscles. We call the resulting interactions between muscle activity patterns "dependent synergies". The learned BN networks were explored for the purpose of classification across subjects based on hand dominance or affliction by stroke. Network structure features were investigated as classification input features and it was determined that specific edge connection patterns of 3-node subnetworks were selectively recruited during reaching movements and were differentially recruited after stroke compared to normal control subjects. The resulting classification was robust to inter-subject and within-group variability and yielded excellent classification performance. The proposed framework extends muscle synergy analysis and provides a framework for thinking about muscle activity interactions in motor control.     

A software package for the decomposition of long-term multichannel EMG signals using wavelet coefficients

This paper presents a method to decompose multichannel long-term intramuscular electromyogram (EMG) signals. In contrast to existing decomposition methods which only support short registration periods or single-channel recordings of signals of constant muscle effort, the decomposition software EMG-LODEC (ElectroMyoGram LOng-term DEComposition) is especially designed for multichannel long-term recordings of signals of slight muscle movements. A wavelet-based, hierarchical cluster analysis algorithm estimates the number of classes [motor units (MUs)], distinguishes single MUAPs from superpositions, and sets up the shape of the template for each class. Using three channels and a weighted averaging method to track action potential (AP) shape changes improve the analysis. In the last step, nonclassified segments, i.e., segments containing superimposed APs, are decomposed into their units using class-mean signals. Based on experiments on simulated and long-term recorded EMG signals, our software is capable of providing reliable decompositions with satisfying accuracy. EMG-LODEC is suitable for the study of MU discharge patterns and recruitment order in healthy subjects and patients during long-term measurements.     

Singularity characteristics of needle EMG IP signals

Clinical electromyography (EMG) interference pattern (IP) signals can reveal more diagnostic information than their constituents, the motor unit action potentials (MUAPs). Singularities and irregular structures typically characterize the mathematically defined content of information in signals. In this paper, a wavelet transform method is used to detect and quantify the singularity characteristics of EMG IP signals using the Lipschitz exponent (LE) and measures derived from it. The performance of the method is assessed in terms of its ability to discriminate healthy, myopathic and neuropathic subjects and how it compares with traditionally used Turns Analysis (TA) methods and a method recently developed by the authors, interscale wavelet maximum (ISWM). Highly significant intergroup differences were found using the LE method. Most of the singularity measures have a performance similar to that of ISWM and considerably better than that of TA. Some measures such as the ratio of the mean LE value to the number of singular points in the signal have considerably superior performance to both methods. These findings add weight to the view that wavelet analysis methods offer an effective way forward in the quantitative analysis of EMG IP signal to assist the clinician in the diagnosis of neuromuscular disorders.