Noise Removal from ECG Signals by Adaptive Filter Based on Variable Step Size LMS Using Evolutionary Algorithms

Abstract

Nowadays, the electrocardiogram (ECG) signal is widely used to detect cardiovascular diseases. Several studies are conducted on noise removal of ECG signal based on the adaptive filter with least-mean Square (LMS) algorithm. In this paper, for improving the traditional LMS method, the evolutionary algorithms are used to select the variable optimal step size of LMS, causing the least error between the main and filtered ECG signals. The proposed Adaptive Noise Cancellation System (ANC) includes Wavelet Transform and IIR-Notch filter to reduce the baseline Wander and Power Line Interference noises. Afterward, an additive white noise generator unit is employed to evaluate the performance of the three adaptive models involving GA-LMS, PSO-LMS, and GA-PSO-LMS algorithms in terms of Signal to Noise Ratio (SNR) and Mean Square Error (MSE). Eventually, to evaluate the performance of the proposed models in terms of the MSE and SNR criteria, we conduct comprehensive experiments on the ECG records of the MIT -BIH database. The obtained results of variable step size, GA-LMS, PSO-LMS, and hybrid GA-PSO-LMS, demonstrate more efficiency in filtered signal compared to constant step size LMS. Besides, in most cases, the Hybrid GA-PSO-LMS method has superiority over two other proposed methods concerning the SNR and MSE criteria.