Page 29 - Computational Retinal Image Analysis

CHAPTER

The physics, instruments and

modalities of retinal imaging 3

b
a
a
b
Andrew R. Harvey , Guillem Carles , Adrian Bradu , Adrian Podoleanu
a School of Physics and Astronomy, University of Glasgow, Glasgow, United Kingdom
b School of Physical Sciences, University of Kent, Canterbury, United Kingdom
1 Introduction

The vital importance of the human eye has stimulated an enormous effort into imag-
ing the retina and other parts of the eye to provide clinical diagnostic information
on disease and for fundamental research. In this chapter we discuss how the physics
underpinning the formation of images of the retina impacts the salient features im-
portant for analysis of retinal images.
The fascination of the physicist with the eye dates back, at least, to the first gruesome
insertion by Newton, in 1665, of a ‘bodkin’ (needle) behind his eye ball [1]. Over the
succeeding three-and-a-half centuries we have developed a deep understanding of im-
age formation and vision in the eye. This has included the development of increasingly
elaborate imaging systems, originating with the first ophthalmoscopes at the time of von
Helmhotz [2] in 1851 and progressing through modern scanning-laser ophthalmoscopes
(SLO) for improved contrast and depth sectioning; optical-coherence tomography for
imaging the three-dimensional structure of the retina [3]; to emerging techniques such
as adaptive optics that enables the imaging of individual photoreceptors [4]. A vast range
of imaging techniques are used in ophthalmic research and clinical application. Over
the past two decades the digital recording of retinal images has become pervasive—and
with it the possibility of automated analysis of images for screening and providing ob-
jective metrics to aid diagnosis. The aim of this chapter is to highlight how the physics of
image formation within the eye leads to specific characteristics of retinal images. For ex-
ample: why is it that images recorded with a SLO appear different from those recorded
with a fundus camera? Why is it difficult to record images of individual blood cells and
the smallest capillaries? Why do arteries and veins appear so different?
The eye has more than a passing resemblance to a camera in terms of function
and shares many similar components. Like a modern camera, there is a lens for form-
ing an image, which can be varied for optimal focus at a range of distances (referred
to as accommodation by clinicians), an iris for adjusting the amount of transmitted
light, and an array of opto-electronic photoreceptors transduce the optical image into
an electronic signal for transmission and processing. Importantly, when we record

Computational Retinal Image Analysis. https://doi.org/10.1016/B978-0-08-102816-2.00003-4 19
© 2019 Elsevier Ltd. All rights reserved.