Signal analysis of electromyogram by artificial neural network

During strong contraction, electromyogram (EMG) becomes a noise-like “interference pattern” composed of trains of motor-unit action potential (MUAP). With its adaptive properties, an artificial neural network (ANN) system is proposed and applied to the analysis of EMG for MUAP´s detection. Features of MUAPs are extracted and fed into the ANN system for on-line training in which the number of classes is not fixed. Then the ANN recognises the signal based on the properties of the training samples. The performance of the system has been tested with different configurations of the ANN and different parameters of computer-simulated EMG signals. The system gives a recognition rate of about 80% for one MUAP with a firing rate of 5 Hz. The recognition rate decreases to 70% or less if the firing rate or the number of different MUAPs increases