Research On Macula Area Localization Of Fundus Image Based On The Segmentation Of Retinal Vessel Ends

In retinal image, temporal to the optic nerve head is the macula which is darker withno blood vessels. It is the area utilized in activities that require discerning sharp detailssuch as reading. Retinopathy in the area of macula, termed maculopathy, is associated witha high risk of visual loss. Hence, accurate determination of the location of the fovea isessential for critical treatment decisions such as laser for diabetic maculopathy which willbe furthered in our future clinical work.The method for macula detection has been discussed in some papers these years. Inconclusion, the methodology can be divided into two kinds, the algorithm using directimaging processing based on the intensity characteristics of macula and the algorithmbased on morphological information between fovea (center of macula) and other structureslike OD and Vessel Arch. However, identifying macula is still a big challenge toresearchers.Based on the theory that macula is an avascular area which conations no blood vessel,we proposal the method to localize the macula after the accurate segmentation of bloodvessel ends in fundus images. Firstly we do some preprocessing such as contrast adjustingto make the fundus images easy for the followed processing. Secondly multi-scale analysismethod is used to segment vessel with different width. Then the ending of the vessels ismarked to be used for further fitting into circle area which is deemed as our macula area.The testing of our methodology shows a good detection of macula, which means ourmethod is accurate and robust for the disease-infected retinal images that don’t have clearmacula image characteristics. However, the better algorithm for automatic and fast maculalocalization should be considered in our future work and this procedure has the potentialto be used for monitoring fundus-laser-surgery systems.