Page 302 - Computational Retinal Image Analysis

300 CHAPTER 15 Retinal biomarkers and cardiovascular disease

changes may be biomarkers of CVD and may thus be potentially useful to improve
CVD risk prediction and stratification.
This chapter aims to review our current understanding of the relationship between
retinal vascular changes and CVD in a clinical perspective.

2 The concept of retinal vascular imaging
As the retina and other end organs (e.g., brain, heart) share similar anatomical fea-
tures and physiological properties (e.g., nonanastomotic end-arteries, blood-brain
and blood-retina barrier), the retinal vessels, measuring 100–300 μm in size, offer a
unique and easily accessible “window” to study the health and disease of the human
microcirculation [11]. While this concept is well known [13], it is only with the in-
troduction of digital retinal photography in the last two decades that retinal vascular
changes could be easily imaged, objectively measured and regularly monitored in a
precise manner.
Retinal vascular changes can be classified as qualitative and quantitative [14].
Panel 1 summaries the glossary of retinal vascular changes measured from retinal
photographs. Qualitative retinal vascular changes can be further classified into clas-
sic retinopathy signs (usually considered to be related to hypertension or diabetes)
and retinal arteriolar wall signs (traditionally thought to be more related to athero-
sclerosis). Retinopathy signs include microaneurysm, retinal hemorrhages, hard exu-
dates and cotton wool spots while retinal arteriolar wall signs include focal arteriolar
narrowing, arteriovenous nicking and opacification of arteriolar wall [15, 16]. Fig. 1
shows examples of qualitative retinopathy and Fig. 2 shows examples of retinal ar-
teriolar wall signs.
With advancements in image processing technologies, quantitative retinal vascu-
lature analysis from digital retinal photographs can be further measured with com-
puter software and standardized photographic protocols. Earlier software such as
the Retinal Analysis (RA; University of Wisconsin, Madison, WI) and Integrative
Vessel Analysis (IVAN; University of Wisconsin, Madison; WI) were widely used
for measuring retinal arteriolar caliber and retinal venular caliber from digital reti-
nal photographs in numerous large population-based studies [17, 18]. Standardized
protocols including measuring retinal vessels coursing through a specified area of a
standardized grid (0.5–1.0 disc diameter from the disc margin), and converting indi-
vidual vessels into summary variables. The revised Knudtson-Parr-Hubbard formula
is widely used to summarize the retinal arteriolar and venular calibers of the large six
arterioles and venules as central retinal artery equivalent (CRAE) and central retinal
vein equivalent (CRVE) respectively [19].
Subsequent research has focused on assessment of other retinal vasculature in-
cluding branching and geometry. It is known that the branching pattern of vascular
networks develops to minimize the energy required to maintain efficient blood flow,
based on Murray’s principle [14, 20–22]. Deviations or alterations from optimal ar-
chitecture are speculated to result in impaired microcirculatory transport, reduced