Cardiac beat classification based on wavelet analysis of empirical mode decomposed ECG signals

Abstract

The information of electrocardiograms (ECG) signal is the most important bio- electrical message of human body, which reflects the basic law of heart activity. To improve the efficiency and accuracy of the diagnosis of cardiovascular diseases, it has a very important significance. For ECG beat classification, a wide range of signal processing techniques extracting features from time, frequency and time frequency domains have been reported in the literature. Since, ECG is a nonsta- tionary signal, time frequency analysis can perform better than the conventional time or frequency analysis methods. But, development of a multi-class beat classi- fication method, which is simple yet effective in handling practical conditions such as lack of enough training dataset and random selection of training and testing dataset, is still a challenging task. In the empirical mode decomposition (EMD) domain, the basic functions are directly derived from the original signal without the knowledge of any previous value of the signal.In this thesis, first the intrinsic mode functions (IMFs) are extracted by using the EMD and then the discrete wavelet packet decomposition (WPD) is performed only on the selected dominant IMFs. Both approximate and detail WPD coefficients of the dominant IMF are taken into consideration. It is found that some higher order statistics of these EMD-WPD coefficients corresponding to different beat classes exhibit distinguishing charac- teristics and these statistical parameters are chosen as the desired features. It is proposed and shown that smoothed three point central difference for an ECG sig- nal namely dECG signal and modified dECG signal can further enhance the level of discrimination as it also includes the effect of P and T waves apart from QRS complex of an ECG beat. Each of the proposed sets of feature when fed to Eu- clidean distance based k-Nearest Neighbor (k-NN) classifier can classify different cardiac beats with randomly selected training and testing dataset. Simulations are carried out to evaluate the performance of the proposed methods in terms of sensi- tivity, specificity, selectivity and accuracy. It is shown that the proposed methods outperform the state-of-the-art method with greater effectiveness.