Image Processing and Retinopathy: A Novel Approach to Computer Driven Tracing of Vessel Network

摘要

Retinopathy is a major cause of blindness in the world with a higher incidence in senescent people. The most effective treatment to reduce visual loss is early detection of retinal damage through regular screening. We are studying the development of a software for quantifying abnormalities in the retinal vessel network with parameters highly sensitive to pathology progression. This paper deals with the study of an effective computer driven system of retinal vessel pattern segmentation, able to recognize even thin capillary branching. In particular, the proposed method is mainly based on the so called shape modelling techniques and is tuned up on real retina images. The current version of our method has been tested on several images differing for local retinal vessel pattern view and ocular background. Preliminary results on real retina images are promising and prompt us to follow this approach in building a computer driven segmentation procedure suitable in retina vessel pattern tracing.