Effective de-noising of ECG by optimised adaptive thresholding on noisy modes

Abstract

Electrocardiograph (ECG) denoising is a preliminary step in QRS complex detection and ECG beat classification process. An ECG denoising methodology is proposed that uses ensemble empirical mode decomposition (EEMD) for decomposing input ECG to its intrinsic mode functions (IMFs). Then noisy-IMFs are identified by the proposed method that uses squared distance similarity measure. On selected noisy-IMFs, hard thresholding, by the evolutionary optimisation algorithms (EOAs)-based thresholds, is applied. For selecting the best performing EOA, algorithms namely artificial bee colony, particle swarm optimisation (PSO), and cuckoo search (CS) optimisation are implemented. The proposed methodology is tested with the Massachusetts Institute of Technology and Beth Israel Hospital (MIT-BIH) arrhythmia database and is evaluated in terms of signal-to-noise ratio (SNR) and mean-square error (MSE). In comparison with state-of-the-art methods, the proposed method is found to be the best attaining maximum SNR and MSE for PSO and CS algorithm optimised thresholding.