EMG pattern analysis and classification by neural network

It is proposed that electromyographic (EMG) patterns can be analyzed and classified by neural networks for EMG-controlled prosthetic members. The thrust of this work is that neural networks make EMG pattern recognition much easier and more efficient; thus, use of EMG control in a prosthetic arm/hand would involve much less physical and mental effort on the part of the subject. The FFT-analyzed pattern for a one-channel surface EMG of the flexor digitorium superficialis can be recognized by the standard back-propagation neural network. Classification of the five finger movements based on the recognition of their patterns is successfully accomplished.<
>