Classification of EMG Signals Using Spectral Features Extracted from Dominant Motor Unit Action Potential

Abstract

In this paper, disease classification of electromyogram (EMG signal) based on the spectral features extracted from the dominant motor unit action potential (MUAP) is discussed. This scheme provides an improved accuracy and reduces the computational complexity to a great extend. The MUAPs are extracted from the EMG signal using a matlab program known as EMGLAB and the highest energy MUAP is selected as dominant MUAP. The main goal of this study is to extract the relevant spectral features for the classification so that the redundant features can be eliminated. For spectral feature extraction direct and DWT based methods are used. K-nearest neighborhood (KNN) classifier is used for the classification purpose. The performance is evaluated using three clinical dataset in terms of specificity sensitivity and accuracy. The results show that the classification based on the proposed method gives better accuracy than the existing methods for disease classification. Index Terms-Electromyography (EMG), motor unit action potential(MUAP), EMGLAB, amyotrophic lateral sclerosis(ALS), Myopathy, K-nearest neighborhood (KNN) classifier.