Collation of a Few Retinal Vessel Segmentation Techniques: Is the Problem Solved?

Abstract

Morphological features of retinal blood vessels are used to diagnose and stage many ophthalmic disorders. Automated segmentation of retinal blood vessels can reduce the labor as well as the cost of the treatment process. Accurate segmentation around the optic disc, lesion area, vessel crossovers and bifurcations, handling central light reflex, and identification of minor vessels in the low contrast regions are some of the challenges faced in the robust segmentation of retinal vessels. Of these, some challenges have been addressed and some still need significant attention to be resolved. Although various techniques cater to different subsets of challenges, no single approach exists that has successfully addressed all the challenges. In this work we critically analyze six segmentation techniques, presenting their strengths and weaknesses. The advantage of this analysis is that the flaws in individual methods can be worked on to improve them individually and the strengths of different methods could be coupled to yield a more robust segmentation algorithm. © The Author(s), under exclusive licence to Springer Nature Singapore Pte Ltd. 2025.