35 - A system for automatic cardiac arrhythmia recognition using electrocardiogram signal

Abstract

Electrocardiogram (ECG) signal plays a dynamic role in finding cardiac arrhythmias of the heart. Any deviation of the cardiac rhythm from normal sinus rhythm is known as an arrhythmia or irregular heartbeat. Cardiac arrhythmias occur in a short duration; therefore, it is difficult to distinguish them using the human eye directly. Thus, a computer-aided diagnosis tool is essential to detect the cardiac arrhythmias automatically. Diagnosis at an early stage of cardiac arrhythmia may facilitate to decrease the mortality rate of the heart patients. In this chapter, two different techniques are proposed for ECG arrhythmia classification. The first method is based on a convolutional neural network (CNN), whereas the second method uses random forest classifier. The following two steps are followed by both the methods: (1) preprocessing, (2) feature extraction and classification. The preprocessing steps involve normalization of the ECG signal followed by R-peak detection. In the first method, CNN involves both steps, feature extraction and classification, but in the second method, features are extracted using dual-tree complex wavelet transform (DTCWT) and classification is performed by the random forest technique. Extracted DTCWT-based morphological features are combined with four other temporal features. The resultant combined feature vector is applied to the random forest classifier for automated identification of the cardiac arrhythmia. The performance of the techniques is evaluated on Massachusetts institute of technology-Beth Israel hospital (MIT-BIH) arrhythmia database, and five types of ECG beats—namely, paced (P), premature ventricular contraction (V), right bundle branch block (R), left bundle branch block (L), and normal (N) beat—are classified in this work. The experimental result shows that the proposed method is better compared to the earlier reported techniques.