Anatomical mapping of SEMG signal quality and controllability in skeletal muscle groups for myoelectric prosthesis

Abstract

This thesis presents the hardware implementation for myoelectric signal processing and the experimental evaluation of myoelectric signals to characterize the controllability of the muscle groups in the upper body for controlling the myoelectric prosthetic device. Digital filters were implemented to improve the quality of raw myoelectric signals acquired from the targeted muscle groups. The 5th order median filter implementation provided the reliable noise reduction for the electrophysiological noise observed in the abdominal muscle groups. The real-time onset detection algorithm was implemented to determine the onset and the offset of myoelectric signals and to generate discrete control signals for the prosthetic device. The experiment was designed to investigate the adequacy of utilizing myoelectric signals from the muscle groups in the upper body–deltoids, pectoralis majors, latissimus dorsi, and external obliques–as used in the control of myoelectric prosthetic devices. The voluntary muscle contraction capability of each targeted muscle group was evaluated during the experiment. It was demonstrated that the precise and accurate myoelectric control was achieved using the deltoids muscle group. However, the pectoralis majors and the external obliques were proven to be more appropriate to apply to fast switching on/off control. The combinations of the myoelectric signals acquired from the deltoids and the latissimus dorsi were investigated to generate multiple output stages, and 4 discrete states of myoelectric output were obtained using those muscle groups simultaneously.