Classification of ECG using convolutional neural network (CNN)

Abstract

Electrocardiogram (ECG) gives the clear record on electrical activities of heart. This record can be used to diagnose various heart diseases. An approach is proposed to automatically detect the myocardial infraction using ECG signals. In this work, a convolutional neural network (CNN) algorithm is implemented for the automated detection of a normal and Abnormal ECG signals (Left bundle branch block beat (LBBB), Right bundle branch block beat (RBBB), Atrial premature beat (APB) and Paced beat (PB)). The feature extraction and signal classification both are carried in a single CNN unit. MIT-BIH arrhythmia database is used to obtain the five different classes of ECG signals. This proposed classifier accurately classifies the signals with reduced classification time. So, in clinical settings this method can be implemented to help the clinicians in the diagnosis of myocardial infarction.