SIMULATION AND COMPARATIVE ANALYSIS OF AN ENHANCED NOISE CANCELLATION IN ECG SIGNAL DENOISING VIA NORMALIZED LEAST MEANS SQUARE (NLMS) ALGORITHM

Abstract

The clinical utility of ECG signal can be compromised by the presence of unwanted components and interference in the ECG signals which do not only obscures critical cardiac information but also introduces inaccuracies in the analysis, leading to misdiagnoses and suboptimal patient care. However, noise in ECG signals arises from multiple sources, including muscular activities, baseline wanders, power line interference and electrode artifacts. This work propose a robust enhanced filtering techniques using the Normalized Least Means Square (NLMS) Algorithm designed to successfully cancel noise out of ECG signal while preserving the key diagnostic features. Extensive simulations were conducted using standard PhysioNet ECG Database and the Massachusetts Institute of Technology- Beth Israel Hospital (MIT-BIH) Arrhythmia Database, the MIT-BIH Normal Sinus Rhythm Database, and the MIT-BIH QT Database to validate performance across diverse noise condition. In the most challenging scenario, where the input Signals-to-Noise (SNR) reached as low as -5 dB, the filter achieved a SNR output quantifying approximately 88.96dB. The filter across different noise type, ECG datasets, and signals-to-noise ratio (SNR) levels suggest that the enhanced filtering approach has potential for broader applicability in various clinical contexts and can be integrated into a diverse array of monitoring system.