Blood Vessel Segmentation in HRT Images for Glaucoma Early Detection

摘要

In this paper, we present a solution to the blood vessel segmentation problem, with the long term goal of automatically diagnosing early stages of glaucoma. The images are obtained from the Heidelberg Retina Tomograph. We introduce two approaches. Firstly, we present the Thresholding Approach and secondly, the Clean Edge Map algorithm. The algorithms share their first steps - adjusting the lighting and cutting out of the pupil. The preprocessed images are then modified by combination of tools such as the Canny operator or thresholds. The results are finally shown using an implemented GUI.