An Automated Tool for Quantitative Assessment of Retinal Vasculopathy

Abstract

Purpose : The aim of this study was to develop and evaluate a quantitative tool that identifies four hallmarks of retinal vasculopathy: (i) microaneurysms (MAs); (ii) blood-retina barrier (BRB) leakage; (iii) reduced blood flow; and (iv) non-perfusion.

Methods : The tool was designed to analyze fluorescein angiography (FA) data and generate quantitative maps of the four aforementioned vasculopathy markers. The generated maps were then segmented using the ETDRS (Early Treatment Diabetic Retinopathy Study) grid, and a disease-severity-index was calculated for each grid-region based on the overall presence of vasculopathy markers. The algorithm was evaluated in 120 patients with diabetic retinopathy (DR), who underwent FA imaging at 3 time-points: baseline, and at 1- and 2- year follow-ups. Half of the patients underwent imaging before and after treatment (pan-retinal photocoagulation and/or anti-vascular endothelial growth factor).

Results : Our analysis suggests that the earliest marker of DR is the appearance of MAs, followed by reduced retinal perfusion and flow, and then – increased BRB leakage. Our preliminary results further show that treatment significantly reduces the extent and degree of BRB leakage and increases blood flow (p<0.05). Importantly, the algorithm was able to identify patients who responded to treatment and those who did not (20-30%, depending on treatment).

Conclusions : Our findings highlight the ability of the developed tool to detect and quantify biomarkers of retinal vasculopathy and objectively assess disease severity and response to treatment. Coupled with a graphical user interface, the tool stands to allow ophthalmologists to perform computer-assisted FA analysis in clinical settings.

This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.