Automatic Pathological Analysis of Diabetic Retinopathy Using Minimal Feature Vector Classifier

Rising technologies in health care aim at reducing unnecessary visits to medical specialists, minimizing overall cost of treatment and optimizing the number of patients seen by each doctor. In this paper explores a method that recommends referring a patient with diabetes for Diabetic Retinopathy (DR) assessment based on the image classification outcome, which is especially useful in remote and rural areas. Two stages of classifications are used to decide the need for referring the doctor. First stage is lesion detectors. The outcome of the lesion detectors gives the information about different kinds of lesions based upon Nave Bayes classifier. Second stage is the referring stage, which is based on the lesion detectors decision scores and finding the high level feature vectors.