The Decomposition of Multi-channel Surface Electromyogram Based on Waveform Clustering Convolution Kernel Compensation

It is important to obtain the motor unit information from multi-channel surface electromyogram (SEMG). A new decomposition method called waveform clustering convolution kernel compensation is proposed in this paper, which is classified based on waveform to improve performance. According to the number of clusters, the classification of SEMG waveform is described using a minimum distance classifier. The simulation results and experiment results indicate that the proposed algorithm in this paper can exactly decompose firing pattern of motor unit in comparison with classic convolution kernel compensation. This approach potentially offers a new tool to sensitively obtain muscle function and could more accurately guide advances in the evaluation of rehabilitation.