Automatic Screening and Classification of Diabetic Retinopathy

Diabetic Retinopathy (DR) is a microvascular complications caused by increase of insulin in blood, leading to blindness or vision loss because of changes in blood vessels of retina. DR is highly preventable with regular screening and timely intervention of lesions which can help ophthalmologists in detecting at an early stage. The background or non-proliferative DR contains four types of lesions, i.e. microaneurysms, hemorrhages, hard exudates and soft exudates. This paper presents a novel automatic approach for detecting DR in eye fundus images by employing image processing techniques. The proposed system consists of preprocessing, feature extraction using Gray Level Co-Occurrence Matrix (GLCM), and classification is done using Support Vector Machine (SVM) and k Nearest Neighbor (kNN). The proposed system uses genetic algorithm to evaluate and test publicly available retinal image database using performance parameters such as sensitivity, specificity and accuracy.