Hypertension and Heart Attack Detection System Based on Retinal Image Analysis

Abstract

This study tackles the important problem of delayed and erroneous identification of hypertension and heart attack risk, which continue to be major causes of cardiovascular deaths worldwide. Traditional diagnostic approaches, such blood pressure monitoring and electrocardiograms, have limitations. They can be intrusive, unreliable, or not suited for large-scale screening. As a result, many instances go undetected until serious organ damage occurs. The suggested method uses retinal fundus imaging, paired with deep learning, to identify tiny microvascular changes that indicate systemic cardiovascular problems. The main objective is to design a diagnostic method that is non-invasive, inexpensive, and automated. This method should help with early detection in both clinical settings and places with little resources. The proposed system processes retinal pictures using vessel segmentation, feature extraction, and classification with a Convolutional Neural Network (CNN). This produces understandable results, which are then improved by heatmap visualizations. This study is new because it combines training on many retinal datasets, detailed understanding of vascular features, and explainable artificial intelligence. This approach aims to find biomarkers connected to high blood pressure and heart attacks that might not be easily seen by doctors.