A Lightweight Customized Convolutional Neural Network based Cardiac Health Detection

Collectively, Cardio Vascular Diseases (CVDs) are the most dangerous global threat to human health. Timely and accurate cardiovascular disease (CVD) diagnosis is essential for saving lives. This paper proposes a deep learning-based approach for a heart health assessment system that uses a customized Convolutional Neural Network (CNN) model to classify and detect cardiac abnormalities in MRI scans. Highly distinguishing normal and pathological conditions is achieved using a system leveraging Customized CNN feature extraction capabilities. The proposed custom CNN involved a diverse dataset for training, which includes both healthy and diseased ones, to enhance cross-case robustness. Appropriate use of the proposed model can reduce the challenges of diagnosing late-stage cardiovascular disease (CVD) and promote preventive healthcare strategies by offering a contactless, costeffective healthcare solution. The proposed CNN model yields an accuracy of 85.06% in identifying patients of different categories.