Optimizing electrode positions on forearm to increase SNR and myoelectric pattern recognition performance

Abstract

With the advances in electromyography-based human–computer interaction, particularly in myoelectric prosthetic hands, the position of electromyography electrodes has gained less attention from researchers. However, the performance of hand movement recognition significantly varies with the change of electrode position around the forearm muscles and along the length of the forearm muscles. To find the robust position on the forearm, this study performs a comprehensive and systematic investigation of the feasibility of hand movement recognition using electromyogram (EMG) signals concurrently recorded for sixteen electrode arrays on the forearm. The obtained results indicate that the electrode array placed on the middle of the extensor digitorum communis and extensor digiti minimi performs better than other electrode positions around the forearm. In addition, multiple electrode arrays placed near to elbow joint achieve a significantly higher signal to noise ratio and signal to movement artifact ratio. In addition to improved EMG signal quality, the recommended positions on the forearm significantly overperform in hand movement recognition and it is validated with five well-known feature extraction methods, various window sizes, and three classification algorithms. In this study, the electrode arrays near to elbow joint on the forearm achieve an F1 score of 98.28% to 98.80% with a linear discriminant analysis classifier. Therefore, this study clearly demonstrates the feasibility of recommended forearm positions for hand movement recognition. As these electrode positions are near to elbow joint, so, it is expected that these positions will be available in most amputees and utilized for improved hand movement recognition.