Mouse Cursor Control System Based on Facial Electromyogram and Mechanomyogram

Abstract

This paper reports the development of a mouse cursor control system as an assistive technology for upper arm amputees. The control is based on facial electromyogram (fEMG) and mechanomyogram (MMG) signals. The fEMG and MMG signals are collected for six words from eight subjects. A reference signal has been simulated based on the mean values of the signals representing the six words. The Euclidian distance between the Cepstral coefficients of the six words from that of the reference signal comprised the feature vector. Classification is through a probabilistic neural network. Six mouse cursor operations: up, down, left, right, left click and right click are reproduced. We have achieved an average classification rate of 91.5% using fEMG and 89.5% using MMG signal. The classification result is mapped into cursor operations through a switch based linear control.