New method for classifying MUAPs of EMG signal

In this study, a Self-Organizing Feature Map (SOFM) network was used to classify motor unit action potentials (MUAPs) in the electromyography (EMG) signal. The duration and peak-to-peak value was utilized as the typical features to construct a two-dimension vector. The EMG signal was recorded with a concentric needle electrode from musculus biceps brachii of a healthy man contracted at 20% Maximum Voluntary Contraction (MVC). It was digitized at a sampling frequency of 20 KHz. After clustering by SOFM network, three standard MUAP templates were obtained. In general, the results are consistent to a doctor's suggestions.