NNERVE: Neural Network Extraction of Repetitive Vectors for Electromyography. I. Algorithm

Abstract

Artificial neural network (ANN) based signal processing methods have been shown to have significant robustness in processing complex, degraded, noisy, and unstable signals. A novel approach to automated electromyogram (EMG) signal decomposition, using an ANN processing architecture, is presented in this paper. Due to the lack of a priori knowledge of motor unit action potential (MUAP) morphology, the EMG decomposition must be performed in an unsupervised manner. An ANN classifier, consisting of a multilayer perceptron neural network and employing a novel unsupervised training strategy, is proposed. The ANN learns repetitive appearances of MUAP waveforms from their suspected occurrences in a filtered EMG signal in an autoassociative learning task. The same training waveforms are fed into the trained ANN and the output of the ANN is fed back to its input, giving rise to a dynamic retrieval net classifier. For each waveform in the data, the network discovers a feature vector associated with that waveform. For each waveform, classification is achieved by comparing its feature vector with those of the other waveforms. Firing information of each MUAP is further used to refine the classification results of the ANN classifier. Then, individual MUAP waveform shapes are derived and their firing tables are created.

Electromyographic (EMG) Decomposition

“…In general, as the number of superpositions increases, the density of the legitimate clusters decreases, which can make clustering more difficult. A variety of clustering methods have been used in iEMG decomposition, including hierarchical clustering (13,36), partitioning (39), template matching (10), neural networks (40,41), and densitybased clustering (42).…”

Comparison of AR parametric methods with subspace-based methods for EMG signal classification using stand-alone and merged neural network models

“…Eight MUP features were used as input [8]. Hassoun et al performed classification by using the time domain waveform as input into a 3-layer ANN with a "pseudo-unsupervised" learning algorithm [9,10].…”

EMG Signal Decomposition Using Motor Unit Potential Train Validity

“…Raw-data (time samples) and first-or second -derivative of time samples [1], [5][6][7], [19], [25], [27], [72], [74], [76][77][78][79][80], [82], [86], [87], power spectrum and Fourier transform coefficients [45], [49], [50], [62], wavelet coefficients [53][54][55], [59], [59], [79], [81], [83], [88][89][90][91][92][93][94], and principal components of wavelet coefficients [95] are features that have been used to represent and assign MUPs to MUPTs. Using power spectrum coefficients [62] or wavelet coefficients of MUPs decreases the dimensionality of the feature space and hence may improve the processing time.…”

“…In many of these algorithms MU firing pattern information is used passively or actively along with MUP shape information to assign an individual MUP to the correct train. [2], [5], [6], [108], artificial intelligence-based MAPCs [25], [122][123][124], artificial neural networks [86], [87], fuzzy logic-based classifiers [68], [77], [78], [82], [88], [89], [125], certainty-based classifiers [1], [7], [27], [72], [76], [79], [80], [82], [125], matched template filter classifiers [2], [5], [6], [45], [49], [50], [75], and multi-classifiers [27], [79], [82], [88], [89], [125] [86], [87] an auto-associative algorithm is used to extract the number of possible MUPTs and their MUP templates.…”

“…Several existing decomposition procedures are based solely on clustering [53], [58], [65][66][67], [69], [74], [91], [93], [94], [108][109][110][111]. However, generally, a clustering technique is followed by supervised classification [1], [2], [7], [19], [27], [54], [55], [62], [72], [73], [76][77][78][79][80], [80][81][82], [86][87][88][89], [112], [113] and both clustering and supervised classification tasks are repeated across a number of iterations. Throughout multiple iterations, the detected MUPs are assigned to the corresponding MUPTs until either the extracted MUPTs are stable or some termination criteria are met.…”