SCIEntIfIC RepoRtS | 7: 16792 | DOI:10.1038/s41598-017-16620-x
Spatial statistical modelling of
capillary non-perfusion in the retina
Ian J. C. MacCormick
, Yalin Zheng
, Silvester Czanner
, Yitian Zhao
, Peter J. Diggle
Simon P. Harding
& Gabriela Czanner
Manual grading of lesions in retinal images is relevant to clinical management and clinical trials, but it
is time-consuming and expensive. Furthermore, it collects only limited information - such as lesion size
or frequency. The spatial distribution of lesions is ignored, even though it may contribute to the overall
clinical assessment of disease severity, and correspond to microvascular and physiological topography.
Capillary non-perfusion (CNP) lesions are central to the pathogenesis of major causes of vision loss.
Here we propose a novel method to analyse CNP using spatial statistical modelling. This quanties
the percentage of CNP-pixels in each of 48 sectors and then characterises the spatial distribution with
goniometric functions. We applied our spatial approach to a set of images from patients with malarial
retinopathy, and found it compares favourably with the raw percentage of CNP-pixels and also with
manual grading. Furthermore, we were able to quantify a biological characteristic of macular CNP
in malaria that had previously only been described subjectively: clustering at the temporal raphe.
Microvascular location is likely to be biologically relevant to many diseases, and so our spatial approach
may be applicable to a diverse range of pathological features in the retina and other organs.
e retinal microcirculation is exquisitely accessible to clinical observation, and unlike other organs, the retinal
vasculature is arranged perpendicular to an optical axis. Consequently alterations to small vessel ow can be eas-
ily mapped using techniques such as uorescein angiography (FA) and optical coherence tomography angiogra-
phy (OCT-A). Capillary non- perfusion (CNP) appears as distinctive dark areas with geographic boundaries, and
develops when blood fails to reach areas of the capillary bed (Fig.1a–c). It is a feature of several major causes of
blindness including diabetic maculopathy, retinal vein occlusion, and retinal artery occlusion
. CNP also occurs
in malarial retinopathy, and can be graded manually according to a validated scheme
. Malarial retinopathy is
seen in children and adults with cerebral malaria, which has a high mortality rate. e retina and brain sustain
similar damage in cerebral malaria, and several retinal signs are associated with death (reviewed in
As with grading schemes for retinal vein occlusion
the malarial retinopathy grading scheme assesses the
overall area of CNP in various large retinal regions. Manual grading is necessarily semi-quantitative, time con-
suming and costly. ese and other constraints mean that manual grading is impractical for large image datasets,
and at best captures only a tiny fraction of the biological information contained in a retinal image.
Beyond lesion type, frequency, and area, spatial characteristics may be a particularly relevant aspect of a ret-
inal image. is is because the retinal microvasculature is not homogenous but rather composed of regions with
dierent vascular topology, geometry, and corresponding haemorheology and physiology. For example, the foveal
avascular zone is a unique region where the retina is supplied solely from the underlying choriocapillaris. e
perifoveal region has one capillary layer, which forms an oxygen diusion watershed with the choriocapillaris.
e temporal macula and horizontal raphe contain a further watershed between superior and inferior temporal
. erefore, the biological meaning of CNP in one sub-region of the macula may be dierent from that of
a lesion of similar size in an adjacent region. Current grading techniques are too coarse to allow such distinctions.
Department of Eye & Vision Science, Institute of Ageing and Chronic Disease, University of Liverpool, 6 West
Derby Street, Liverpool, L7 8TX, United Kingdom.
Malawi-Liverpool Wellcome Trust Clinical Research Programme,
Queen Elizabeth Central Hospital, Blantyre, Malawi.
Centre for Clinical Brain Sciences, University of Edinburgh,
Edinburgh, United Kingdom.
School of Computing, Mathematics and Digital Technology, Faculty of Science
and Engineering, Manchester Metropolitan University, Manchester, M1 5GD, United Kingdom.
Cixi Institute of
Biomedical Engineering, Ningbo Institute of Industrial Technology, Chinese Academy of Sciences, Ningbo, China.
CHICAS, Lancaster Medical School, Lancaster University, Lancaster, LA1 4YB, United Kingdom.
St Paul’s Eye Unit,
Royal Liverpool University Hospital, Liverpool, L7 8XP, United Kingdom.
Department of Biostatistics, Institute
of Translational Medicine, University of Liverpool, 1-5 Brownlow Street, Liverpool, L69 3GL, United Kingdom.
Correspondence and requests for materials should be addressed to G.C. (email: email@example.com)
Received: 4 August 2017
Accepted: 10 November 2017
Published: xx xx xxxx