Description

Blood Vessel Segmentation using SegNet -Matlab -Image processing project

Diabetic Retinopathy (DR) is spreading alarmingly in developing countries like India. If not treated early, it may leads to complete blindness. For early treatment, it must be diagnosed early which required complete automation in diagnostic process. The first step of such automatic diagnosis is the blood vessel segmentation. Though there are numerous algorithm mostly in classical methods, are available in litterers, deep neural networks are most promising developed by recent researchers. In this paper, the standard SegNet is used with some modification and simulated with standard publicly available fundus image databases. Simulated result achieves high performance with overall accuracy 0.9298 and precision is 0.80 and its result is much better than the manual way of a skilled ophthalmologist.