Classification of ECG images by Deep Learning to support the diagnosis of cardiovascular diseases

Abstract

In this paper, we have proposed the ECG classification model to partly support medical professionals. The model is built by deep learning technique with MIT-BIH training data. Specifically, the convolutional neural network (CNN) will be built to take input of images cropped from the ECG images and the output labeled as the corresponding arrhythmia. The obtained model is reliable when the test results have a relatively high accuracy of 97 to 99%.