Temporal registration of vessels in retinal images

Abstract

Fundus image of human retina serves as a significant avenue for doctors to monitor and probe the progress and severity of various diseases. Cardiovascular diseases such as stroke and coronary heart disease are the leading causes of mortality worldwide, narrowing of the retinal arterioles has been suggested to predict cardiovascular disease. Temporal registration of retinal images provides an important groundwork for doctors to monitor such narrowing. In this paper, we describe a tree matching approach to register retinal images. We model each vessel in retinal image as a tree, called Vessel Feature Tree (VFT). We design matching functions and a global alignment algorithm to compute the best match between vessels in two images. Experiment results on 300 pairs of real-world retina images indicate that the proposed approach is able to achieve an accuracy of 93% and we prove that our proposed algorithm outperforms other latest developed algorithms.