PREDICTION OF DIABETIC RETINOPATHY FROM RETINAL IMAGES USING FUZZY C-MEANS AND ARTIFICIAL NEURAL NETWORKS ^

Abstract

There are various kinds of the infections occur in the eyes. The very fatal disease which occurs in
the eye is the diabetic retinopathy (DR). It is the second largest disease that occurs amongst the
humans according to the WHO – United Nations survey. The major cause of vision loss is DR. For the
work, we use the publicly available Diabetic Retinopathy detection database available in Kaggle.
Raw retinal funduscopic images are hard to process by machine learning algorithms. The diabetes
retinopathy detection technique has the five phases. They are Image Acquisition, pre-processing,
vessel segmentation, vessel classification and disease prediction. Images on the retina provide
significant information about pathological changes caused by local eye disease, which reveals
diabetes, hypertension, atherosclerosis, vascular disease and stroke. So, in this work, the
implementation of automate segmentation approach is carried out based on active contour method to
provide regional information.