Detection of Healthy and Neuropathy Electromyograms Employing Stockwell Transform

In this paper, Stockwell transform based feature extraction technique is proposed for detection and classification healthy and neuropathy electromyography (EMG) signals. The healthy and neuropathy EMG signals were acquired from an online database and Stockwell transform was applied separately on each category of EMG signals. From the magnitude of S-matrix, suitable features parameters were extracted and statistical significance of the selected features were assessed using parametric t-test. Support vector machines (SVM) classifier was used for the purpose of classification of EMG signals. A mean accuracy of 99.92% was achieved employing polynomial kernel function with polynomial index ‘m=3’. Experimental investigations and comparative study with existing literatures revealed that the proposed methodology can be potentially implemented to develop a computerized disease diagnostic system for detection for classification of healthy and neuropathy EMG signals.