Customized CNN Architecture for ECG Image-Based Classification of Cardiovascular Diseases

Outbreaks Cardiovascular diseases (CVDs) are still the main reason behind deaths globally; because they have a great effect on the total disease burden. As of recent estimates, approximately 32 percent of all global deaths are attributed to cardiovascular diseases, which translates to around 18.6 million deaths annually. Electrocardiography (ECG) is a fundamental tool in cardiology, traditionally analyzed in its raw signal format. However, recent advancements in machine learning and image processing suggest that analyzing ECG data as images might offer superior predictive capabilities for cardiovascular events. CNN models are especially having a more notable impact in this field. In this perspective, we have used ECG images and made improvements using Convolutional neural networks and outperformed the existing models with 98 percent accuracy. Xception model is the one that served as our base upon which we made some personal modifications resulting in a customized Convolutional Neural Network (CNN) to conduct this research.