Main Article Content
Abstract
Diabetic Retinopathy (DR) is the primary cause of vision loss and the early symptoms of this disease are exudates. Prior detection and treatment may prevent the vision loss. Automatic detection of exudate in retinal images is a challenging task. In this paper the two techniques Support Vector Machine (SVM) and Deep Convolution Neutral Network (DCNN) are used for retinal exudate detection and extraction. The SVM technique is applied to enhance and calculate the features of retinal image to recognize exudates. The DCNN technique extracts the blood vessels from the retina images. From these blood vessels the retina image is detected as exudate or non-exudate. Both the techniques are compared for area under the curve, specificity, accuracy and sensitivity. From the simulation results it is evident that DCNN technique is superior compared to SVM technique. The SVM and DCNN techniques are implemented using MATLAB software.