Denoising Artifacts from Cardiac Signal Using Normalized Variable Step Size LMS Algorithm

In this paper, an efficient Error Data Normalized Variable Step Size Least Mean Square (EDNVSSLMS) adaptive algorithm is presented to enhance the quality of an Electrocardiogram (ECG) signal. Due to physiological and non-physiological effects, ECG signals usually undergo numerous artifacts such as Baseline Wander, Muscle artifact, Power Line Interference and Electrode Motion artifacts. The proposed EDNVSSLMS algorithm de-noise these artifacts with better Peak Signal to Noise Ratio, misadjustment and convergence rate compared to other LMS based algorithms, while preserving important clinical wave features morphologies. Also, based on o EDNVSSLMS algorithm, we implemented sign and block based EDNVSS algorithms. Finally we have applied these algorithms to ECG signals corrupted with noise. The performance results shows that Block Based EDNVSSLMS algorithm gives better elimination of noises in the ECG signal with less misadjustment and high peak to signal noise ratio. For less computational complexity and for fast communication of the signal is main, then we select Sign Regressor EDNVSS algorithm. Copyright © 2015 IFSA Publishing, S. L.