CardioRetinaNet Joint Analysis of Retinal Images for Cardiovascular Disease Diagnosis and Age Estimation using Convolutional Neural Network

Abstract

In recent years, despite the consistent progress in the field of medicine, a rapid growth of patients with cardiovascular disease has been a problem globally. Cardiovascular disease, a general term for disorders of the heart or blood vessels, require high-speed and precise detection to control and avoid hostile consequences. Traditionally, computerized tomography scans, utilizing images retrieved from CT-scans, to calculate coronary artery calcium score were the dominant method for the diagnosis of cardiovascular diseases. This usual technique is problematic because of its lengthy procedure and the difficulty of getting the diagnosis due to the expensive cost. To address this problem, I propose a convolutional neural network-based cardiovascular disease diagnosis system using retinal images which are notably more cost-effective than computerized tomography scans. The proposed system takes retinal images as input and generates a categorical assessment of the severity of cardiovascular disease as output. Additionally, it provides a predicted age of the patient, contributing to an enhanced performance in cardiovascular disease classification. By incorporating these features, the proposed system aims to advance the accuracy of cardiovascular disease diagnosis. The experimental results clearly demonstrate that the proposed system attains a state-of-the-art performance in diagnosing cardiovascular disease. I expect that the proposed system will make a significant contribution to the utilization of retinal images as a biomarker for diagnosis of cardiovascular disease.