Classification of muscle fatigue condition using multi-sensors

Abstract

The aim of this work is to assess the muscle fatigue condition using multimodal system. Muscle fatigue is a common muscle condition which experiences in our daily activity. There were 20 subjects participated in this study. Electromyogram (EMG) (shows the electrical activity of the muscle), Mechanomyogram (MMG) (shows a mechanical activity of the muscle) and Acoustic myogram (AMG) (is audible produced when the muscle was contracted) were used in this study. EMG, MMG and AMG were recorded continuously from hamstring muscle, according to the data acquisition protocol. The recorded signals were segmented into fatigue and non-fatigue. Time domain, frequency domain and time-frequency domain features were extracted from the myograms. The extracted features were classified using k-nearest neighbor. The mean accuracy of EMG, MMG and AMG was 87.10%, 81.40% and 67.23% respectively. The mean accuracy of the multimodal system was 92.07%. In this paper, we also have discussed the effect of single myogram and multi modal myograms.