Machine Learning and Deep Learning for Cardiovascular Disease Detection in ECG Images

Abstract

This research focuses on the critical global issue of cardiovascular diseases, particularly heart conditions, a leading cause of mortality. Timely prediction is essential, and Electrocardiogram (ECG), a cost-effective and noninvasive tool, plays a pivotal role in monitoring heart activity. To enhance predictive accuracy, this project employs deep learning techniques, specifically transfer learning from neural networks like Squeeze Net and Alex Net, along with a specialized Convolution Neural Network (CNN) architecture. These techniques aim to identify four significant cardiac abnormalities: abnormal heartbeat, myocardial infarction, history of myocardial infarction, and normal cases. The model's uniqueness lies in its exceptional performance, achieved by extracting crucial features through a combination of deep learning and traditional machine learning algorithms. This research underscores the transformative impact of artificial intelligence on healthcare, significantly advancing medical condition predictions through image analysis.