Related Experiment Video
Updated: May 26, 2025
In Vivo Vascular Injury Readouts in Mouse Retina to Promote Reproducibility
Published on: April 21, 2022
Sensitivity of ophthalmologists, residents, and optometrists in identifying peripheral retinal tears on
Jesse J Jung1, Xavier Chan2, Kai Xiong Cheong2
1East Bay Retina Consultants, Inc., Oakland, CA, USA; Department of Ophthalmology, University of California, San Francisco, San Francisco, CA, USA.
Objective:
To compare the sensitivity of 3 groups of masked graders with varying levels of ophthalmic training to identify peripheral retinal breaks utilizing ultra-widefield orthogonal, directed peripheral steering, and auto-montaged images.
Design:
Retrospective observational cohort study.
Participants:
155 patients from a single vitreoretinal specialist's practice.
Methods:
221 eyes with pretreatment orthogonal, directed-peripheral steering, and auto-montage that underwent laser retinopexy for retinal tears between 2015 and 2021 were divided into 2 groups: treatment-naïve and control. Combined sensitivity and specificity of identifying all retinal breaks on orthogonal, directed-peripheral steering, and auto-montaged imaging were calculated compared with the gold standard of mydriatic, scleral depression examination. Linear probability modeling was performed to calculate the required surface area from auto-montage images to identify breaks that were missed initially on orthogonal images.
Results:
For orthogonal images, combined sensitivity was highest for ophthalmologists (67.53%), residents (62.34%), and then optometrists (55.84%). The sensitivity increased for orthogonal/steering (ophthalmologists [85.71%], residents [77.92%], and optometrists [67.53%]) and auto-montage (ophthalmologists [85.51%], residents (80.28%), and optometrists [75.00%]). To ensure identification of all tears with auto-montage that was initially missed on grading the orthogonal image, for every 10% increase in montage surface area, there was a 4.8 percentage point (%p) increase in the likelihood of detecting a retinal tear on montage image grading (p = 0.023).
Conclusions:
Masked graders had moderate sensitivity in identifying retinal breaks with ultra-widefield images. Even with directed-peripheral steering and auto-montage, optometrists had the lowest sensitivity compared to ophthalmology residents and general ophthalmologists and required increased surface area to identify all retinal breaks.
