Approximation And Analysis Of ECG Signals Using Polynomials

Abstract

An electrocardiogram (ECG) signal is the most vital bio-medical signal that represents newlineelectrical activity of heart over time. ECG signals are required for interpretation newlineand diagnosis of cardiac related issues which are obtained as small electrical potentials newlinedue to cardiac functioning with respect to time, by placing electrodes on specific newlinelocations of skin. ECG signals are usually corrupted with various types of unwanted newlineinterferences in the form of artifacts/noises which distort the ECG signal, thus preventing newlinecorrect interpretation, monitoring and diagnosis. Existing signal enhancement newlinetechniques reduce specific noises to some extent, but are not able to retain clinically newlineimportant features of ECG signals. Therefore, these noises must be reduced for better newlinemedical evaluation. Continuous recording of ECG is required for monitoring of critical newlinecases, the number of such cases are increasing at an alarming rate leading to voluminous newlinesize of recorded ECG data. Moreover, due to insufficient number of cardiologist newlineto handle all cases, ECG data needs to be transmitted via communication channels for newlineanalysis and interpretation purpose consuming large channel bandwidth. Hence, storage newlineand transmission of such a huge data is impossible without compression. newlineThe main objective of this research is to provide an efficient, reliable and flexible newlineECG approximation model that approximates complex ECG signals upto significant newlinelevels through a series of steps. Firstly, ECG signal restoration algorithms have been newlinedeveloped to remove spurious data, utilizing the concept of total variation majorizationminimization newlineoptimization approach using first, second and combined difference total newlinevariation. Next, a polynomial model is developed based on Lagrange-Chebyshev interpolation newlinetechnique with chebyshev nodes to compress the enhanced signal. Efficiency newlineof model is further improved by characterizing the signal at significant points with the newlineBottom-up algorithm. The proposed models are tested on 20 complex ECG signals newlinetaken from MIT-BIH