ECG Heartbeat Classification: A Comparative Performance Analysis between One and Two Dimensional Convolutional Neural Network

Abstract

Recently automatic categorization of Electrocardiogram (ECG) has garnered great attention as it is the most reliable measure for monitoring cardiovascular system functionality. Deep learning has been proven to be a very efficient and accurate approach in this field. A deep learning method based on two-dimensional deep convolutional neural network is proposed in this paper for classifying heartbeats in order to accurately detect five different arrhythmias in accordance with the AAMI EC57 standard. Its performance is compared with another method based on one-dimensional deep convolutional neural network by using PhysioNet's MIT-BIH dataset for evaluation. Both of these CNN methods proved to be better than some state of the art methods when implemented on the same dataset. The 2D CNN method outperformed all the other implemented methods with an average accuracy of 94.37%.