Electromyography Signal Pattern Recognition for Movement of Shoulder

Abstract

Pectoralis major and deltoid are two muscles that are associated with the movement of the shoulder. Electromyography (EMG) signal acquired from these two muscles can be used to classify the movement of the shoulder based on pattern recognition. In this paper, an experiment for EMG data collection involves eight healthy male subjects who perform four shoulder movements which are flexion, extension, internal rotation and external rotation. Feature extraction of EMG data is done using root mean square (RMS), variance (VAR) and zero crossing (ZC). For pattern recognition, the classifiers that are used are Support Vector Machine (SVM), K-Nearest Neighbour (KNN), Linear Discriminant Analysis (LDA) and Quadratic Discriminant Analysis (QDA). Classification results shows highest accuracy on ZC feature using an SVM classifier with cubic kernel. The study on shoulder movement using EMG of pectoralis and deltoid muscles could be extended on arm amputees based on hypothesis that the EMG signal could be utilized for control of robotic prosthetic arm.