Deep learning-based approaches for myocardial infarction detection: A comprehensive review recent advances and emerging challenges

Abstract

Myocardial infarction (MI) is a severe heart disease requiring immediate and accurate detection for effective treatment. Deep learning (DL) algorithms have recently shown promise in enhancing MI diagnostic accuracy from electrocardiography (ECG) and echocardiogram (ECHO). This review presents a comprehensive literature overview focusing on recent innovative research on DL algorithms in ECG and ECHO analysis for MI identification. We examined relevant studies employing DL models, analyzing datasets, model architectures, preprocessing approaches, and performance measures. The findings reveal that DL-based algorithms substantially improve MI detection in terms of accuracy, sensitivity, specificity, and overall diagnostic performance. This is crucial for quicker, more reliable diagnoses and reducing the risk of complications. DL-based ECG and ECHO analyses emerge as pivotal tools for early and efficient MI identification. This review contributes to understanding the latest DL advancements in ECG and ECHO analysis for MI diagnosis, offering important directions for future research.