Blue-light reflectance imaging for identification and margin delineation of peripheral full-thickness retinal breaks: a retrospective image-based analysis

Abstract Background To evaluate whether blue-light reflectance (BLR) imaging improves identification and margin delineation of previous stablished peripheral full-thickness retinal breaks (FTRB) compared with color imaging, green-light reflectance (GLR), and red-light reflectance (RLR), particularly in lightly pigmented fundi. Methods This retrospective, observational, cross-sectional image analysis included 89 widefield retinal images from 81 eyes of 73 patients with peripheral full-thickness retinal breaks (FTRB) acquired at a single center. Color images obtained with a Broad Line Fundus Imaging system (CLARUS 700; Carl Zeiss Meditec) were decomposed into blue, green, and red monochromatic reflectance images. Two masked retina specialists independently graded each modality for FTRB identification and margin delineation, with adjudication of discrepancies. Analyses were performed in the overall cohort and stratified by fundus pigmentation (58 normally pigmented and 31 lightly pigmented images) using Cochran’s Q test and exact McNemar tests with Bonferroni correction. Results FTRB identification differed significantly across modalities in the overall cohort ( P < 0.001), with identification rates of 93.3% for color imaging, 88.8% for GLR, 100% for BLR, and 30.3% for RLR. BLR identified more FTRB than GLR ( P = 0.002), whereas the difference compared with color imaging did not remain significant after correction. In lightly pigmented fundi, BLR identified all FTRB and outperformed GLR (100% vs. 74.2%; P = 0.008). Margin delineation also differed significantly across modalities ( P < 0.001), with rates of 78.7% for color imaging, 70.8% for GLR, 97.8% for BLR, and 13.5% for RLR. BLR achieved higher rates of margin delineation than both color imaging and GLR (both P < 0.001). In lightly pigmented fundi, margins were delineated in 96.8% of images with BLR, compared with 54.8% for color imaging, 38.7% for GLR, and 0% for RLR. Conclusions BLR achieved the highest rates of FTRB identification and margin delineation, with the greatest relative benefit observed in lightly pigmented fundi. These findings suggest that BLR may serve as a practical adjunct to color imaging within the context of a retrospective, image-based analysis.