Analysis of ECG signals to classify abnormal patterns by employing Artificial Neural Network and Discrete Wavelet Coefficients

Abstract

Electrocardiogram (ECG) is one of the most important tools to diagnose the heart health. Patients with arrhythmias diagnosed by ECG are commonly seen in clinical practice. Incorrectly interpreted ECGs might result in inappropriate clinical decisions. It is necessary to reduce physicians' pressure and increase accuracy in diagnostic and therapeutic processes with automatic diagnosis of various cardiac disorders and arrhythmias. The following article suggests classifying the different types of cardiac arrhythmias from ECG signals. The arrhythmia classification is based on the morphological features of the signal and features extracted by a Discrete Wavelet Transform (DWT). The Artificial Neural Network (ANN) has been used as a classifier. MIT-BIH ECG arrhythmia database acquired from pysionet.org is used for analysis purposes. ANN is used to classify four groups, including three types of arrhythmias, Atrial Premature Beat (APB), Left Bundle Branch Block (LBBB), Right Bundle Branch Block (RBBB), and Normal group (N). The K-fold cross-validation method results show that using morphological features and wavelet coefficients together, the average classification accuracy is 93%. Average classification accuracy is 83% when only the signal's morphological features are used, and if only the wavelet coefficients are used, it is 88%. In this way, the highest classification accuracy is obtained using both categories of features.