Blood Vessel Extraction and Bifurcations Detection Using Hessian Matrix of Gaussian and Euclidian Distance

One of the sign for diagnosing diabetic retinopathy is Intraretinal Microvascular Abnormalities (IRMA). IRMA is located in the superficial retina area that adjacent to the non-perfusion area resulting venous beading at least two quadrant in the fundus image. The difficulties in venous beading detection are the characteristics of the objects in retinal blood vessel images were varied. There are arteries and veins inside the fundus image. Two of these vessels also contain bifurcation. Bifurcation detection is a very crucial step to obtain the optimum result and proper classification between normal veins with the veins that have the beading. This study, the blood vessel and Eigen value of hessian matrix will be extracted from the fundus image. The extraction result then processed using morphological and Euclidian distance to detect the bifurcation point of the retinal fundus image. This step is the early stage of venous beading detection. Bifurcation detection was performed by combining morphological operation to eliminate the noises of fundus image as well as to compute Euclidian distance of the vessel. The result of this study is expected to detect bifurcation point accurately. The outcome of bifurcation features extraction will be used to classified normal veins from venous beading.