Spatial statistical modelling of capillary non-perfusion in the retina

Ian J C Maccormick,Peter J Diggle,Simon P Harding,Yalin Zheng,Yitian Zhao,Silvester Czanner,Gabriela Czanner

doi:10.1038/s41598-017-16620-x

Abstract

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 quantifies 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.

Highlights

The retinal microcirculation is exquisitely accessible to clinical observation, and unlike other organs, the retinal vasculature is arranged perpendicular to an optical axis
Automated segmentation provides the user with Capillary non-perfusion (CNP) metrics, such as the proportion of CNP-pixels in a given area, and has advantages over manual grading in terms of cost and reproducibility
In all 132 images, the inner circle automated overall CNP was found to be positively associated with manual grading of macular CNP (Supplementary Figure 3a, p = 0.11 two-sample t-test, p = 0.05 Mann-Whitney U test, n = 87 and 45) and gave excellent consistency (ICC = 0.88, p < 0.01)

Summary

Introduction

The retinal microcirculation is exquisitely accessible to clinical observation, and unlike other organs, the retinal vasculature is arranged perpendicular to an optical axis. 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[1]. There are several automated methods for segmenting areas of CNP from retinal images[6,7], including one developed by our group and applied to a subset of manually-graded images from children with malarial retinopathy[2,8]. We need new analytical tools to refine image segmentation data into biologically meaningful information Such information would have uses as an outcome measure in clinical trials for diabetic maculopathy and retinal vein occlusions, as well as clinical practice. An ideal spatial model must account for the biological topography of the tissue, and must be flexible to be adapted to specific problems of the image acquisition

Methods

Results

Conclusion