ABSTRACT
The World Health Organization (WHO) predicted that in the year 2012 about 347 million people worldwide would be suffering from diabetes, and more than 80% deaths occur due to diabetes in different countries. WHO projects that diabetes will be the seventh major cause of death in 2030. Diabetic retinopathy begins with a leakage of blood or fluid from the retinal blood vessels and it damages the retina. Non-proliferative diabetic retinopathy (NPDR) is an early stage of diabetic retinopathy and it is categorized by three stages: mild, moderate, and severe. The mild stage is characterized by the presence of minimum microaneurysms. The moderate stage is defined by the presence of haemorrhages, microaneurysms, and hard exudates, whereas the severe stage is determined by the existence of blockages in vessels, leading to blood deprivation in several areas of the retina. The performance of the proposed algorithm was tested on online databases such as STARE, DRIVE, DiarectDB0, DiarectDB1, and SASWADE, and it was seen that the classification of NPDR lesions using multilayer perceptron results in 94% accuracy.
