Hybrid system for automatic classification of Diabetic Retinopathy using fundus images

Diabetes over long duration causes damage of tiny blood vessel that nourishes the retina. This blood vessel leaks the blood and other fluids that cause swelling of retinal tissues. This leads to the Diabetic Retinopathy. Prolonged Diabetic Retinopathy may lead to vision loss. To prevent blindness, accurate detection at early stage is essential. We present a system for automatic classification of subjects as Diabetic Retinopathy and Normal by using fundus images which gives fast results when there is mass screening of Diabetic Retinopathy. Clinical features like blood vessel area, exudates area, bifurcation point and Non-segmented texture based features were extracted from fundus images by using different techniques which helps in classifying the subjects. SVM a classifier is used, which classify subjects into Normal and DR based on features extracted.