Publication

Abstract : Diabetic retinopathy (DR) is a serious eye condition linked to diabetes, resulting from high blood sugar levels. It can cause varying degrees of vision impairment, potentially leading to blindness. Early symptoms include blurred vision, changes in color vision, and dark spots. Detecting DR early is crucial to preventing vision loss, as it typically appears as small red spots on the retina from hemorrhages or microaneurysms. This paper presents a method that utilizes Convolutional Neural Networks (CNN) to identify crucial features such as haemorrhages, exudates, bleeding, and microaneurysms, thereby enhancing the accuracy of diabetes-related predictions. The CNN approach enables the processing of real-time data, simplifying the extraction of key features from retinal images without necessitating extensive annotation or segmentation. This approach not only streamlines the detection and treatment process during the early stages but also provides a preliminary safeguard for patient health.