Comparison of neovascularization detection in proliferative diabetic retinopathy using widefield swept source optical coherence tomography and fluorescein angiography

Abstract

Purpose : We compare the ability of resident ophthalmologists to identify neovascularization (NV) in patients with proliferative diabetic retinopathy (PDR) using widefield (WF) swept source (SS) optical coherence tomography angiography (OCTA) and fluorescein angiography (FA) to help examine how easily this new technology could be more widely deployed to less experienced users.

Methods : Eyes with PDR based on clinical examination were imaged with 15mm x 15mm WF SS-OCTA and FA no more than 1 week apart. These images were scrambled to create a grading set consisting of anonymized SS-OCTA images of the retina slab with the corresponding B-scans (hereby referred to as SS-OCTA with corresponding B-scan), SS-OCTA images of the retina slab and the vitreoretinal interface (VRI) angio slab (hereby referred to as the VRI slab), and FA image set consisting of an early phase image and a late phase image (hereby referred to as FA) for each eye. After IRB approval, participants were asked to identify NV in each image and scored on the number of NV identified correctly. Additionally, participants were timed on how long they spent on each FA and VRI slab.

Results : 10 eyes from 10 patients were included in the study for a grading set consisting of 30 images. The mean number of neovascularization per image was 2.03. 11 resident physicians participated in the study. Overall, resident physicians correctly identified 65.6% of NV using FA, 50.8% of NV using SS-OCTA with corresponding B-scan, and 78.0% of NV using the VRI slab. There was no statistically significant difference in resident physician’s ability to detect NV across the three imaging modalities (P = 0.12). Participants spent an average of 4.8 seconds less for each VRI slab compared to FA (13.5 seconds versus 18.3 seconds, P = 0.04).

Conclusions : Detection rates of NV by resident physicians using SS-OCTA was comparable to that of using FA. Additionally, resident physicians were able to identify NV more quickly on VRI slab compared to FA. Our results suggest that SS-OCTA may be an appropriate imaging modality for the detection of NV in PDR patients, even with less experienced OCTA graders.

This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.

 

An example of FA (A), SS-OCTA with corresponding B-scan (B), and VRI slab (C) of PDR patients included in the grading set. NV shown in yellow arrowhead.