Detection of Diabetic Retinopathy Using K-Means Clustering and Self-Organizing Map

Diabetes Mellitus (DM) is a metabolic disorder characterized by hyperglycaemia (high blood sugar) which is increasingly affecting the world population. As diabetes progresses over time, many vital parts of the body are affected. We concentrate on the effect of diabetes on eye which leads to diabetic retinopathy. In diabetic retinopathy, the retinal microvasculature is subjected to progressive pathological alterations leading to complications like retinal non-perfusion, increase in vascular permeability and pathologic proliferation of retinal blood vessels. This work we have classified digital fundus images in to two classes: (i) normal and diabetes retinopathy comprising of mild non-proliferative diabetic retinopathy, moderate non-proliferative diabetic retinopathy, severe non-proliferative diabetic retinopathy and proliferative diabetic retinopathy using Self Organizing Map (SOM) classifier. The features are extracted from the retinal images using K-means algorithm and fed to the SOM classifier for classification. We have shown clear separation between the two classes using our proposed model.