Classification of diabetic retinopathy severity levels of transformed images using K-means and thresholding method

Abstract

Diabetic Retinopathy is one of the major eye complications which occurs because of diabetes. The main reason behind this disease is damage to the blood vessels and due to which lack of oxygen supply in retina. This paper is an extension work of the previously done This paper categorizes the severity levels of diabetic retinopathy by using clustering approach and thresholding method. Two approaches are implemented in this paper Spatial Domain and Frequency Domain for calculating Feature Vectors. Clustering is applied to these Feature Vectors and classification is done by using thresholding method. The classification with encouraging results is obtained i.e. Sensitivity, Specificity and Accuracy is 94%, 100% and 97% respectively.