Denoising ECG Signals by Using Extended Kalman Filter to Train Multi-Layer Perceptron Neural Network

Abstract

The purpose of this paper is to study a denoising scheme for ECG signals by using extended Kalman filter based on Multilayer Perceptron Neural Network. A comparison with other enhancement conventional filters, such as, Wiener, wavelet, median and least mean square filters has been investigated. This approach is evaluated on several ECG by artificially adding white and colored Gaussian noises, and real non-stationary muscle artifact to visually inspect clean ECG recordings. It is also evaluated on studying the mean square error and Peak signal to noise ratio of the filters outputs. On the basis of these two parameters, a comparative analysis has been presented to explore the efficient denoising capability of the proposed method. The results of this simulation show the effectiveness of this approach.