Page 102 - Computational Retinal Image Analysis

CHAPTER

Retinal vascular analysis:

Segmentation, tracing, and 6

beyond

a,b
c
e
d
Li Cheng , Xingzheng Lyu , He Zhao , Huazhu Fu , Huiqi Li d
a A*STAR, Bioinformatics Institute, Singapore, Singapore
b ECE, University of Alberta, Edmonton, AB, Canada
c Zhejiang University, Hangzhou, China
d Beijing Institute of Technology, Beijing, China
e A*STAR, Institute for Infocomm Research, Singapore, Singapore

1 Introduction
As the proverb goes, the eye is a window to your soul. In fact, the eye is also a window
to your health, as many diseases such as eye diseases, diabetes, cardiovascular, and
systemic diseases manifest themselves via different structures, especially the blood
vessels, in an eye. For example, it is of clinical interest to examine the width of artery
or vein vessels, the branching ratio, the vessel length between junction points, etc.
As illustrated in, for example, Ref. [1], stemmed from the ophthalmic artery, the
arterial inflow to the eye is further branched. In particular, the central retinal artery
travels along the optic nerve and spreads out in branches to cover the retina. Being
parallel and countercurrent to the central retinal artery, the venous outflow exits
the eye through the central retinal vein, again along the optic nerve. Together, the
retinal vasculature is responsible for supplying oxygen and nutrients and removing
metabolic waste from the retina. Impaired oxygen supply can as well damage the
health of the retina.
Readouts from vessel structure or vasculature have been linked to various
diseases. Subtle vessel changes may also occur during the early stage of disease
cycle. The departure of vasculature geometry from normal state may also underlie
the development of other signs such as retinal lesions. For example, it is known that
the vasculature geometry including vessel diameter and tortuosity provide predictive
information for hypertension, diabetes, Alzheimer disease, and cardiovascular
disease [2–8]. It is thus used frequently in reliable and automated procedures in
analyzing the vasculature and its associated properties. This chapter will focus on

Computational Retinal Image Analysis. https://doi.org/10.1016/B978-0-08-102816-2.00006-X 95
© 2019 Elsevier Ltd. All rights reserved.