Artificial Intelligence in Ophthalmology: Artery-Vein Ratio as a Biomarker for Hypertensive Retinopathy Detection

A systematic approach for the timely detection of hypertensive retinopathy (HR) is done using deep learning-based artery-vein (AV) ratio analysis. HR is a pivotal marker of systemic vascular damage, highlighting the need for early detection to enable effective intervention. AV-CapsNet model utilizes a specialized capsule network architecture to extract and classify features robustly. By examining retinal images to calculate the arteriovenous ratio (AVR), intricate patterns of arteries and veins are discerned. The CapsNet enhances the model's ability to detect subtle HR-related changes by capturing hierarchical feature relationships. Rigorous testing across various datasets, such as VICAVR and INSPIRE AVR, underscores AV-CapsNet's superiority, achieving an accuracy of 98.8%, precision of 97.7%, and recall of 95.5% in HR detection. The proposed deep learning framework not only advances medical image analysis but also holds promise for revolutionizing HR diagnostics, facilitating timely medical interventions, and enhancing patient outcomes.