Weakly-supervised Deep Learning for the Improvement of Cardiovascular Disease (CVD) Prediction using Retinal Fundus Images

This research explores new methods of predicting Cardiovascular Disease (CVD) from retinal fundus images using Deep Learning algorithms. Over the past decades, there is increasing recognition that the retinal microvasculature can be used to directly visualise human circulation system non-invasively. The main benefit of this research is the translational value of deep learning in the prediction of CVD from retinal images, which subsequently will enable the implementation of tele-retinal screening to predict CVD risks in underserved rural areas through remote medicine. Results have shown an accuracy of 91.00% and 73.89% for the prediction of heart attack and hypertension respectively from retinal images.