Efficient hybrid approach to segment and classify exudates for DR prediction

Diabetic retinopathy (DR) is initiated due to the severity of diabetes which can finally lead to an incurable blindness. It is a significant reason for optical damage that may cause blindness permanently. There are no main symptoms of DR appearing initially but its quantity and severity rises with the passage of time. Initial screening and diagnosis of DR may help to stop vision loss. Exudates (EXs) are one of the primary clinical symptoms of DR. In this manuscript, a computerized technique is proposed for DR detection based on EXs. The proposed system is consisting of four major steps. The first step is enhancement of region of interest using median filter and adaptive contrast enhancement method. After that, local variance and global threshold methods are utilized for candidate lesions segmentation. Moreover, texture features with multiple classifiers are applied for classification. The proposed method is evaluated in terms of sensitivity, specificity, accuracy and area under curve on DIARETDB1, MESSIDOR and local data sets.