An Effective Automated Technique for Retinal Disease Identification in Diabetic Retinopathy without Manually Labeled Kit

Abstract

Diabetes is a rapidly increasing common disease among people in worldwide which causes dysfunction of various organs. Diabetic retinopathy (DR) is the main cause of blindness in adults. Exudates are among the primary symptoms of diabetic retinopathy. So, regular screening is needed for early detection of exudates which could decrease the chances of blindness. The quick growth of diabetes pushes the limits of the current DR screening capabilities for which the digital imaging of the eye fundus algorithm automatic image analysis algorithm is the conventional solution. In this work, new exudates detection method is given which has overcome the limitations of labeled lesion training sets such as: time consumption, complexity and more probability of error. In this we present a new concept to normalize the fundus image and directly compare our method with an implementation. Here, we introduce two variations of a new exudates segmentation method that comes under the category of thresholding methods and find out the diabetics as well as other eye diseases.