Prediction of Cardiovascular Disease Risk From Retinal Vasculature Using a Quantitative Diagnostic Approach With CVD-Net in DR and HR Patients

The global incidence of diabetes is increasing significantly each year, posing serious health risks if not diagnosed at an early stage. Elevated blood glucose levels can lead to complications such as Diabetic Retinopathy (DR), Hypertensive Retinopathy (HR), Cardiovascular Diseases (CVD), and renal failure. Background The objective of this study is to establish an association between retinopathy and CVD, which is driven by shared pathogenic mechanisms including inflammation, microvascular damage, oxidative stress, and endothelial dysfunction. Diabetes and hypertension substantially increase the likelihood of DR and HR, which in turn elevate the risk of developing CVD. Objective This work is to forecast the risk of cardiovascular diseases by analyzing retinal vasculature and associated clinical risk factors in patients with DR and HR. Methods The method involves a quantitative diagnostic approach using morphological and physiological attributes of the retinal vascular system, such as the Arteriole-to-Venule ratio (AVR) and the Cup-to-Disc ratio (CDR), which serve as biomarkers. In addition, systemic risk indicators including age, gender, Body Mass Index (BMI), smoking habits, and alcohol consumption are incorporated. Deep Learning (DL) techniques are employed to detect DR, HR, and CVD and to quantify their characteristic features. Results The results indicate that early investigations based on the proposed approach can effectively identify patients at high risk, potentially preventing up to 90% of CVD cases when detected during the early stages. Implications This study highlight the potential of retinal imaging, when combined with deep learning, to serve as a reliable, non-invasive tool for early cardiovascular risk prediction in diabetic and hypertensive individuals, enabling timely clinical intervention.