Diagnosing Glaucoma From Fundus Images By Detecting Blood Vessels And Bend Points: An Image Processing Algorithm Approach
Abstract
The study examined the use of automated image processing in diagnosing glaucoma. The study obtained the data used for the study from a fundus image data base online. From the fundus images the optic disc and optic cup were segmented based on an image processing algorithm proposed by the study. The optic disc was segmented using a framework that improves accuracy by removing nose and blobs. The optic cup was segmented by tracking the blood vessel and their bend points which are joined together to obtain an optic cup contour. The result of the Cup-to-Disc-Ratio (CDR) was then compared to the ground truth of the fundus images and it was observed to have high similarity. The performance of the algorithm was also accessed and it was observed to have an accuracy of 98.7%.