Myopathy detection and classification based on the continuous wavelet transform

Abstract

Electromyography (EMG) technique is often used for diagnosis of neuromuscular diseases such as myopathy that affects the muscle and causes many changes in the electromyography signal characteristics. This paper presents a new method for analysis and classification of normal and myopathy EMG signals based on the continuous wavelet transform (CWT). Classification algorithms, namely Support Vector Machine (SVM), k-Nearest Neighbor (k-NN), Decision Tree (DT), Discriminant Analysis (DA) and Native Bayes (NB) were used in our study. Five features were extracted from the CWT and employed them as input features to the classifiers. Results were evaluated and subsequently, a comparison was made in terms of performance markers, namely, accuracy, sensitivity, and specificity to ensure the efficacy of individual classifiers as well as the number and the combination of the feature sets. Results showed that k-NN classifier with an association of four features delivered the best performances with an accuracy of 93.68%.