Cardiovascular Risk Assessment With Retinal Images Using Densenet-121

In response to the escalating global mortality rates attributed to heart disease, this project presents a novel non-invasive approach for early heart attack prediction using retinal image data.Departing from traditional invasive methods, such as blood tests and ECGs, the project leverages deep learning method applied to retinal scans sourced from the Diabetic Retinopathy 224x224 dataset on Kaggle.By scrutinizing subtle changes in retinal blood vessels, including size variations and background illumination uniformity, the system can detect potential indicators of impending heart attacks.Through the utilization of distinct deep learning algorithm Densenet-121, with achieved accuracy of 97.15%, the project demonstrates robust predictive capabilities.By autonomously learning from the dataset, this algorithm generates trained model capable of accurately identifying individuals at elevated risk of heart disease, thus paving the way for early intervention and improved patient outcomes.