Probabilistic Feature Extraction Techniques for Electrocardiogram Signal-A Review

Abstract

Abstract Extraction of features for an ECG signal plays a vital role in making the diagnosis of most of the diseases associated with cardiac muscle as well as the different states of arrhythmias. In this study, a literature review over the comprehensive manner has been done for the probabilistic feature extraction technique of an electrocardiogram signal in making the analysis of various classification methods of ECG arrhythmia signals that has been proposed over the past years of research. The ECG arrhythmia classification methods includes few digital signal processing techniques, Fuzzy Logic techniques, Hidden Markov Model, Support Vector Machines, Genetic Algorithm, Particle Swarm Optimization, Artificial Neural Networks, Transductive Transfer Learning etc. with which each individual approach exhibits its own disadvantages and advantages. For diagnosing the heart’s clinical condition, Echocardiogram is considered to be an essential tool but it consumes more time for processing and analyzing the ECG rhythms which possess huge number of heart beats. Therefore, it is in the urge that the system of diagnosis has to be made automated for classifying the clinical states of heart rhythms and heart beats in making the diagnosis very accurately and in the précised manner. For analyzing the electrocardiogram signal subsequently, the fundamental characteristics in terms of features like peak amplitudes, time intervals are very much required in determining the heart functions.