Diabetic Retinopathy Prediction Using Machine Learning

Abstract

Diabetic Retinopathy (DR) is a significant complication of diabetes and a leading cause of blindness worldwide. It occurs when high blood sugar levels cause damage to the blood vessels in the retina, leading to leakage and other retinal issues. Early detection and classification of DR lesions are crucial to prevent vision loss. While manual diagnosis of retinal fundus images by ophthalmologists is effective, it is often time-consuming, labor-intensive, costly, and carries a risk of misdiagnosis. In recent years, machine learning has become a prominent tool in enhancing performance across various fields, including medical image classification. This chapter evaluates classifiers such as Support Vector Machines, Decision Trees, Logistic Regression, k-Nearest Neighbors, and Artificial Neural Networks to identify the most effective approach for DR classification. Additionally, it reviews available DR Datasets and discusses several challenging issues that require further research. Comparisons with previous studies indicate satisfactory results. Furthermore, in diabetes prediction, our findings highlight those models such as Logistic regression (LR), Support Vector Machine (SVM), decision Tree (DT), Artificial Neural Network (ANN), and K Nearest neighbor (KNN) provide good predictive performance, making them valuable techniques for early detection. These classifiers have also been applied to diabetic retinopathy prediction, demonstrating their ability to analyze retinal fundus images and distinguish between different stages of DR. This research aims to improve DR diagnosis by demonstrating the efficacy of different Machine Learning ML classifiers, thereby aiding in the development of accurate and efficient computer-aided diagnostic systems for early detection and management.