Vessel Segmentation from Color Retinal Images with Varying Contrast and Central Reflex Properties

Clinical research suggests that changes in the retinal blood vessels (e.g., vessel caliber) are important indicators for earlier diagnosis of diabetes and cardiovascular diseases. Reliable vessel detection or segmentation is a prerequisite for quantifiable retinal blood vessel analysis for predicting these diseases. However, the segmentation of blood vessels is complicated by its huge variations such as abrupt changes in local contrast, a wide range of vessel width and central reflex in the vessel. In this paper, we propose a novel technique to detect retinal blood vessels which is able to address these issues. The core of the technique is a new vessel edge tracking method which combines the method of finding pattern of vessel start point and pixel grouping and profiling techniques. An edge profile checking method is developed for filtering noise and other objects, and tracking the real vessel edges. From the filtered edges a rule based technique is adopted for grouping the edges of individual vessels. Experimental results show that 92.4% success rate in the identification of vessel start-points and 82.01% success rate in tracking the major vessels.