Expert System for Diagnosis of Multiple Neuromuscular Disorders using EMG Signals

Abstract

Age factors and muscular diseases like amyotrophic lateral sclerosis (ALS) and myopathy significantly reduce muscle activity. Early and accurate diagnosis of ALS and myopathy is of great significance for maintaining better life quality. Electromyogram (EMG) signals of the Biceps Brachii muscles are widely used for the diagnosis of ALS and myopathy through a computer-aided automated system. This results in early diagnosis of the diseases which is helpful in symptom management in the patients. In this article, EMG signals were first filtered using Empirical Mode Decomposition (EMD) through reconstruction of the signal using appropriate Intrinsic Mode Functions (IMFs). Preprocessed signals were reconstructed using relative energy-based thresholding of IMFs. Important features of preprocessed EMG signals were extracted using Mel Frequency Cepstral Coefficients (MFCC) and Gammatone Cepstral Coefficient (GTCC). The final feature vector was constructed by a combination of MFCC and GTCC features. These features were used to train and test Cubic Support Vector Machine (C-SVM). C-SVM yielded the best results of 91.1% mean accuracy for distinguishing between Normal, ALS, and Myopathy signals. The proposed method was compared with a range of other state-of-the-art classification methods. The results of this research advocate the effectiveness of the proposed framework for the accurate diagnosis of neuromuscular diseases in clinical environments.