Deep learning accurately classifies geographic atrophy on near-infrared reflectance images.
NEAR-INFRARED REFLECTANCE IMAGING FOR THE ASSESSMENT OF GEOGRAPHIC ATROPHY USING DEEP LEARNING.
Aviv Fineberg … Orly Gal-Or
Retina (Philadelphia, Pa.) · 2025
Purpose
The aim of this study was to develop and evaluate a fully automated deep-learning-based approach for detecting geographic atrophy on near-infrared reflectance imaging.
Methods
Near-infrared reflectance images of patients aged 50 years or older with geographic atrophy, confirmed by two retinal specialists, were analyzed at Rabin Medical Center.
98.5%
Results
98.5%
The deep learning model correctly identified geographic atrophy in 98.5% of cases.
More results
The classification data set contained 330 images, and the localization data set included 659 images.
For geographic atrophy localization, YOLOv8-Large achieved 91% sensitivity, 91% precision, IoU of 84%, and DICE coefficient of 88%.
“
Conclusion · 1 of 2
Geographic atrophy can be reliably identified using near-infrared reflectance images.
Conclusion · 2 of 2
Deep learning models can assist in evaluating geographic atrophy on this routinely available imaging modality, aiding in the selection of patients who may benefit from emerging therapies.
