Study: Non-Expert Graders Accurately Identify Neovascularization with Swept Source OCT-A

Optometrist examing patient’s eyes
Researchers evaluated clinicians' ability to detect neovascularization after a brief training on OCT-A.

Ophthalmologists of all levels of grading experience were able to identify neovascularization (NV) in patients with diabetic retinopathy (DR) using swept source optical coherence tomography angiography (OCT-A) as accurately as with fluorescein angiography (FA) after undergoing a brief OCT-A training, researchers found in a study published in the American Journal of Ophthalmology.

Previous studies have indicated that imaging researchers can use swept source OCT-A to identify NV in DR with high sensitivity, but investigators wanted to determine whether this could also be achieved in the clinical setting.

The researchers collated 47 paired swept source OCT-A and FA images from 24 patients with severe non-proliferative diabetic retinopathy (NPDR) or proliferative diabetic retinopathy (PDR) and a high clinical suspicion for the presence of NV. The patients had been imaged with both ultra-widefield FA and OCT-A at the same visit. They determined via consensus-adjudication that NV was present in 36 of the FA images and 35 of the OCT-A images and that there were concordant ground truth grades for the images in all but 1 eye, which demonstrated NV on FA but not OCT-A images.

A dozen ophthalmologists (2 ophthalmology residents, 6 vitreoretinal fellows and 4 vitreoretinal attending physicians from 3 academic ophthalmology departments) reviewed a training slideshow explaining the characteristics of NV on swept source OCT-A and passed an OCT-A training set. Previously, 4 graders (2 attending retina specialists and 2 fellows) had graded OCT-A images for research, and 3 graders had served as lead authors on published research involving OCT-A.

The graders were asked to grade whether NV was present in each image. In the OCT-A grading set, graders were first asked to grade using just the vitreoretinal interface (VRI) and total retinal en face slabs and then, 6 months later, to repeat the grading of the same VRI and total retinal en face slabs along with all corresponding OCT-A B-scans.

Researchers found no significant difference in the percentage of correct grading between the types of images (P =.92). Including B-scans did not result in a significant increase of correct grading of NV on OCT-A (87.8% vs 86.7% if excluded, P =.62).

Residents were more likely to grade the FA incorrectly but OCT-A correctly compared with fellows and attendings (P =.04), who did not show asymmetry in their grading on FA and OCT-A.

Of the 47 FA/OCT-A imaging pairs, 6 demonstrated 33% or larger discrepancy in percentage correct grading between FA and OCT-A; 3 of them demonstrated a greater percentage correct grading on OCT-A while 3 had a greater percentage correct grading on FA.

Limitations to the study include graders’ inability to access the entire sequence of FA images and that the FA images in the study were cropped to the size of the corresponding 12mm x 12mm OCT-A image, so areas of NV outside that area may have been excluded from the field of view.

Disclosure: Several study authors declared affiliations with the biotech or pharmaceutical industries. Please see the original reference for a full list of authors’ disclosures.


Al-khersan H, Russell JF, Lazzarini TA, et al. Comparison between graders in detection of diabetic neovascularization with swept source oct angiography and fluorescein angiography. Am J Ophthalmol. Published online December 10, 2020. doi:10.1016/j.ajo.2020.11.020