Classification of Eye Fundus using Machine Learning Techniques

This study presents a comprehensive approach for the automated classification of eye fundus images into healthy and diabetic categories utilizing machine learning techniques. With the escalating global prevalence of diabetes, early detection of diabetic retinopathy through efficient and accurate classification systems is crucial. In this research, a dataset comprising eye fundus images was meticulously curated, encompassing both healthy and diabetic cases. Two well-established classification algorithms, Support Vector Machines (SVM) and k-Nearest Neighbors (k-NN), were employed to discern between these categories. The primary focus was on achieving high classification accuracy, which serves as a pivotal indicator of the model's performance. Rigorous experimentation, including data preprocessing, feature extraction, and hyperparameter tuning, was carried out to optimize the classifiers. The results showcased the potential of SVM and k-NN in accurately classifying eye fundus images, with both methods achieving commendable accuracy rates, thus demonstrating their efficacy as diagnostic tools for diabetic retinopathy in clinical settings. This research contributes to the advancement of automated screening systems for diabetic eye diseases, offering a promising avenue for early intervention and improved patient outcomes.

Keywords: Fundus Images, Artificial intelligence, Support Vector Machines, k-Nearest Neighbors (k-NN), machine Learning, Accuracy.