PURPOSE: To evaluate the association of visible retinal area (VRA,mm2) on ultrawide field images (UWFI) with diabetic retinopathy (DR) severity and detection of predominantly peripheral lesions (PPL).
METHODS: Nonmydriatic optomap plus UWFI (California/TX; Optos plc, UK) were acquired as part of a DR teleophthalmology program (JVN) and evaluated at a central reading center for DR severity and PPL. Standard care mydriatic UWFI were acquired at an academic retina practice (BEI). All UWFI from 11/6/2017-11/6/2019 were reviewed. Manual lid lifting was standard for JVN imaging and beginning 11/6/2018 was required for BEI imaging if more than 10% of the retina was obscured by lid/lash artifacts. Fully automated algorithms were used to determine VRA (Dice coefficient:0.848) and hemorrhage and/or microaneurysm counts [(HMA),AUC:0.90-0.95]. PPL-HMA were defined as present when at least 1 field had greater HMA number in the peripheral retina than within the corresponding ETDRS field.
RESULTS: Images from 5,919 eyes with gradable UWFI were analyzed. Theoretical maximal UWFI VRA was 923.9. Mean VRA was 665.1±167.6 for all eyes (N=5,919), 550.8±240.7 for nonmydriatic JVN (N=1,418,24.0%) and 701.1±115.3 for mydriatic BEI images (N=4,501,76.0%). Instituting manual lid lifting guidelines for BEI images increased VRA by 10% (688.1±112.0 to 757.0±69.7,p<0.0001) and mean number of HMA in UWFI fields by 41.7% (4.8 to 6.8,p<0.0001). VRA was moderately correlated with increasing likelihood of PPL-HMA (All,BEI,JVN: r=0.33,0.29,0.36;p<0.0001). In JVN images, increasing VRA was associated with more PPL (Q1:4.8%,Q2:9.0%,Q3:11.8%,Q4:9.0%,p=0.01), PPL leading to more severe DR (3.4%,7.1%,8.7%,5.6%,p=0.02) and more PPL-HMA (23.7%,45.8%,60.6%,69.2%,p<0.0001). The findings remained significant after correcting for diabetes duration and DR severity (p=0.02,0.03,<0.0001). There were no significant associations between VRA and DR severity.
CONCLUSIONS: Using fully automated VRA and HMA detection algorithms, pupillary dilation and manual lid lifting were shown to substantially increase HMA detection and VRA. VRA also correlated with detection of PPL. Given the importance of HMA and PPL for determining risk of DR progression, these findings emphasize the importance of maximizing VRA in addition to image quality for optimal risk assessment in clinical trials and teleophthalmology programs.
