EMG Feature Selection for Diagnosis of Neuromuscular Disorders

Electromyography is one of the diagnostic techniques used for detecting neuromuscular disorders. The aim of this study is to classify disorders as neuropathy or myopathy based on EMG signals. We first applied the discrete wavelet transform (DWT) and power spectral density to extract general features associated with neuropathy, and then we used another filter in the second phase of this process based on the ReliefF algorithm to determine the most relevant features. The performance of our algorithm was evaluated using the Random Forest and K-NN classifiers. The results obtained showed that the Random Forest classifier with ReliefF selection method yielded the best performance with an accuracy of 88.8%, sensitivity of 95.33% and specificity of 80%.