Noise Cancelation from EMG : A Neural Network Approach

An Electromyography (EMG) signal is one of the physiological signals that contain many important information about the muscular and nervous systems to diagnose related disease. But recorded Electromyography signals includes environmental noise like power line interference and internal noise like Electrocardiography two most important noises of Electromyography signals. This paper presents an Electromyography signal denoising scheme based on adaptive neural networks, In this study two different neural networks were compared for their efficacy in cancelling power line interference and ECG The first neural network was Adaptive Linear Neuron (ADALINE) that was trained by Least mean squares (LMS) and the second neural network was multilayer perceptron that was trained by Back Propagation. This comparison showed that back propagation method is able to eliminate noise better than Adaptive Linear Neuron. Our criterion for comparing these networks with each others are two parameters , first by determined Mean Square Error and then by bonds of error in steady state

کلیدواژه ها:

Adeline ، Back Propagation ، Electromyography ، neural network ، noise cancellation