Multi-class support vector machine classifier in EMG diagnosis

The shapes 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 the EMG signals recorded at low to moderate force levels, it is required to: i) identify the MUAPs composed by the EMG signal, ii) cluster the MUAPs with similar shapes, iii) extract the features of the MUAP clusters and iv) classify the MUAPs according to pathology. In this work, three techniques for segmentation of EMG signal are presented: i) segmentation by identifying the peaks of the MUAPs, ii) by finding the beginning extraction point (BEP) and ending extraction point (EEP) of MUAPs and iii) by using discrete wavelet transform (DWT). For the clustering of MUAPs, statistical pattern recognition technique based on euclidian distance is used. The autoregressive (AR) features of the clusters are computed and are given to a multi-class support vector machine (SVM) classifier for their classification. A total of 12 EMG signals obtained from 3 normal (NOR), 5 myopathic (MYO) and 4 motor neuron diseased (MND) subjects were analyzed. The success rate for the segmentation technique used peaks to extract MUAPs was highest (95.90%) and for the statistical pattern recognition technique was 93.13%. The classification accuracy of multi-class SVM with AR features was 100%.