Visualizing Diabetic Retinopathy: A Deep Learning Approach to Enhance Fundus Image Analysis

Diabetic retinopathy is an eye disorder caused by high blood sugar levels. It is critical to understand that diabetic retinopathy develops only when a patient has diabetes and goes undiagnosed and untreated. The proposed framework improves the detection of diabetic retinopathy. Le-Net is utilized for input image approval, and it has a 99% accuracy rate. When the brightest spot method is used, the ROI extraction process is accurate to 98.67%. Furthermore, using the U-Net design to separate the optic disc and optic cup yielded dice coefficients of 0.93 and 0.87, respectively. The proposed technique demonstrated the reliability and potential of the system by achieving 99% grouping precision, review, particularity, and awareness utilizing SVM, Neural Network, and Adaboost classifiers.