Comparative analysis of various adaptive algorithms for the enhancement of ECG recording

Abstract

Electrocardiogram (ECG) is used as a diagnostic tool for the newlineidentification and interpretation of cardiac disease. During recording and newlinetransmission, the ECG signal gets corrupted by artifacts emanating from newlinenearby sources. The dominant artifacts found in most of the ECG recordings newlineare Powerline Interference(PLI)and Gaussian noise. The presence of such newlineartifacts covers a significant feature of the ECG signal so that it is difficult to newlineidentify them. For the proper identification and interpretation, an ECG signal newlinewith good quality is required. Therefore, separating artifacts from the ECG newlinesignal is one of the important steps in the ECG analysis. Several ECG noise newlinereduction methods have been proposed to separate valid ECG signal newlinecomponents from unwanted artifacts. Of all methods, the Filtering technique newlinebased on the Adaptive method is the most suitable choice for processing bio newlinesignal like ECG signal. In this thesis, two adaptive methods based on newlineEmpirical Mode Decomposition (EMD) and Synchrosqueezing Transform newline(SST) are proposed to filter artifacts in the ECG signal. EMD is a recently developed technique used for processing a nonlinear, non-stationary signal. It is a fully adaptive, data-driven techniques suitable for analyzing the biomedical signal. It is based on the principles of partitioning input signal into various intrinsic mode functions. According to EMD, any kind of signal can be decomposed into a finite set of intrinsic mode newlinefunctions (IMFs) with a residue where lower-order IMF carries highfrequency newlineinformation and higher-order poses low-frequency content of the signal. An IMF is defined as a function with a number of extreme and zero crossings at least differ by one. The main objective of ECG denoising is to recover useful signal components from corrupted ECG so as to present ECG for better interpretation. newline newline