Power of Cepstrum meets EMG: Detecting ALS and Myopathy

This study is devoted to the classification of neuromuscular diseases (NMDs) including ALS and myopathy by using Electromyography (EMG) signals and a machine learning approach. An elliptic bandpass filter was utilized for the preprocessing of the raw EMG signals. The mean and standard deviation of Mel frequency cepstral coefficients (MFCC) and Gammatone frequency cepstral coefficients (GTCC) were retrieved as cepstral domain features. K-Nearest Neighbors (KNN) classifier, which was trained to classify the EMG signals outperformed other classifiers and demonstrated the highest accuracy of 98.7%. This research demonstrates the effectiveness of using cepstral domain features and the KNN classifier for accurate and reliable classification of NMDs by using EMG signals.