Abstract
Purpose :
Triage of patients with visual complications of Diabetic Retinopathy (DR) in the emergency department (ED), urgent care centers, or primary care offices can be challenging due to limited availability of ophthalmic specialists. In this pilot study, we deployed non-mydriatic automated Ocular Coherence Tomography (OCT) technology (or tele-OCT-B) in conjunction with 3D macular mapping software to assess the ability of non-ophthalmologists to triage posterior pole OCT-B images for Diabetic Macular Edema (DME).
Methods :
Macular OCT-B scans of 41 eyes (21 patients with mean age 56.5 years and male to female ratio of 3:4) were captured in an academic retina clinic using Topcon Maestro 3D OCT-1. 3D topographical maps were rendered. A retinal physician assessed each OCT-B image and 3D map for the presence of DME and the severity level as described by the International Clinical DME Severity Scale. The fundus photo images from Topcon OCT camera were reviewed for DR. Two third-year medical students independently reviewed the topographic maps for the presence of DME. Cohen’s kappa (κ) was used to test concordance between student and retina specialist (gold standard) for identifying DME. Statistical analysis was performed using IBM SPSS software version 25.
Results :
38/41 eyes were included in the final analysis. Non-ophthalmologist graders had an average sensitivity of 70% (SD 4.71) and specificity of 80% (SD 3.08) in detecting DME (Table 1). Inter-grader concordance between non-ophthalmologists was substantial (κ=0.78, p<0.00001). Between non-ophthalmologists and the retina specialist, average concordance in detecting any DME was κ=0.50 (SD 0.008). When stratified by clinical DME severity, on average, non-ophthalmologists correctly classified 8/9 (90%) eyes with moderate to severe DME, 1/5 (20%) eyes with mild DME, and 14.5/15 (97%) eyes with no DME.
Conclusions :
While concordance between non-ophthalmologists and the retinal specialist was modest for detection of DME, non-ophthalmologists correctly identified most moderate to severe DME cases on OCT-B with high agreement. OCT-B 3D map may be an invaluable tool to use with non-mydriatic fundus camera in the ED or primary care offices in screening diabetic patients for vision threatening DME who need urgent ophthalmic referral, especially during a pandemic.
This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.