Automated quantitative characterisation of retinal vascular leakage and microaneurysms in ultra-widefield fluorescein angiography

Abstract

Ultra-widefield fluorescein angiography (UWFA) is an emerging imaging modality used to characterise pathology in the retinal vasculature such as microaneurysms (MAs) and vascular leakage. Despite its potential value for diagnosis and disease surveillance, objective quantitative assessment of retinal pathology by UWFA is currently limited because it requires laborious manual segmentation by trained human graders. In this report, we describe a novel fully automated software platform, which segments MAs and leakage areas in native and dewarped UWFA images with retinal vascular disease. Comparison of the algorithm with human grader-generated gold standards demonstrated significant strong correlations for MA and leakage areas (intraclass correlation coefficient (ICC)=0.78-0.87 and ICC=0.70-0.86, respectively, p=2.1×10-7 to 3.5×10-10 and p=7.8×10-6 to 1.3×10-9, respectively). These results suggest the algorithm performs similarly to human graders in MA and leakage segmentation and may be of significant utility in clinical and research settings.

Keywords: Diagnostic tests/Investigation; Imaging; Macula; Retina.

Conflict of interest statement

Competing interests: None declared.

© Article author(s) (or their employer(s) unless otherwise stated in the text of the article) 2017. All rights reserved. No commercial use is permitted unless otherwise expressly granted.

Figures

Figure 1. Automated microaneurysm detection

A) An early time point ultra-widefield fluorescein angiogram image is selected. B) Initial detection of candidate microaneurysms (MAs). C) Candidate MAs are filtered based on peripheral intensity gradient. D) MAs are superimposed upon original early phase image.

Figure 2. Automated leakage detection

A) A sample late time point ultra-widefield fluorescein angiogram (UWFA) image taken by the Optos 200Tx System. B) “Flattened” late image with removal of intensity gradient from optic disc to the periphery. C) Removal of vessels utilizing early time point UWFA image. D) Leakage areas are superimposed upon flattened late UWFA image.

Figure 3. Microaneurysm detection

A) Ultra-widefield angiogram analyzed by B) grader 1, C) grader 2, and D) automated algorithm. The number of microaneurysms detected by grader 1, grader 2 and algorithm were 355, 295, and 335 respectively.

Figure 4. Leakage detection

A) A dewarped ultra-widefield angiogram analyzed by B) grader 1, C) grader 2, and D) algorithm. Leakage areas of grader 1, grader 2, and algorithm were 122854, 101632, and 150793 pixels respectively.

Figure 5. Comparison of algorithm detection to expert reader gold standard identification of pathologic features

Correlation of A) MA counts and B) leakage areas segmented by algorithm and gold standard. Leakage areas were measured in pixels. Lines represent line of best fit by linear regression. Bland-Altman plots of C) MA and D) leakage comparing manual segmentation to automated segmentation.