Review on Feature Extraction Methods in Neuromuscular Disease Diagnosis

Abstract

This paper mainly focuses on various feature selection methods that is followed for achieving accurate diagnosis of neuromuscular diseases such as Amyotrophic Lateral Sclerosis (ALS) and Myopathy. Since both of these has similarity in the Electromyography (EMG) waveform of normal patients, this will create more difficulties in terms of diagnosis. Hence, proper feature selection is the essential part in the diagnosis. Two feature selection methods were adopted for evaluation. In the first method, time domain and frequency domain features are taken from each frame of EMG signal and in the second method, Discrete Wavelet Transform (DWT) features like maximum DWT coefficient and mean value of high energy DWT coefficients were analysed. For the purpose of classification, the Multi-Support Vector Machine (MSVM) classifier is employed.