Better detection of retinal abnormalities by accurate detection of blood vessels in retina

Early diagnosis and treating the diseases causing loss of vision is one of the most important steps to prevent blindness. Among these diseases are diabetes and degeneration of fovea. One way to diagnose these diseases is to analyze the retinal fundus images. In order to detect the abnormalities of retina, one would better to first localize the anatomical parts of retina. Several methods have been presented and most of them have proposed some methods for vessel structure detection. The objective is to detect all of the abnormal spots and to detect them correctly. In this work a new method for detecting blood vessel tree based on morphological operators is proposed. After vessel detection, the abnormal spots in Retinal fundus images are detected more accurately. The proposed method is applied on 40 fundus images of Nikookari Database. The average sensitivity of the method is 85.82% and the average specificity is 99.98%.