Decomposition of superimposed waveforms of EMG signal using fuzzy neural network

This paper discusses the decomposition of superimposed waveforms in needle electromyography (EMG) signals into their constituent motor unit action potentials (MUAPs). The decomposition method is based on information diffusion theory using the BP network. This method was tested on real and simulated EMG data that was recorded at force levels up to 20% of maximum voluntary contractions (MVS) and the results are satisfactory. Because of the good quality of information diffusion, the method is promising in decomposing EMG signals contracted at high force levels.