Camera type affects retinal AI accuracy and image quality costs
From Canon to Nikon: Camera Domain Shift and Potential Equity Implications in Retinal AI
Computer Vision and Pattern RecognitionComputers and Society
Summary
Diabetic retinopathy, which can cause blindness, needs regular eye screenings. The authors found that different camera brands used for taking retinal pictures affect the quality of images and the accuracy of AI that diagnoses eye diseases. Even though some cameras cost more, cheaper cameras sometimes produce lower-quality images that hurt AI performance. They tested AI on many tasks and found that camera differences cause varying results that aren’t fixed by changing AI settings or image resolution. This means choosing the camera matters for reliable AI eye screening.
What this means in practice
- •For medical device manufacturers: Design retinal imaging devices considering how camera choice impacts AI accuracy and prevent performance variability across hardware.
- •For healthcare ai deployers: Adjust retinal disease AI screening to account for differences in camera systems to ensure consistent diagnostic reliability.
Authors
Lisa Y. W. Tang
Abstract
Diabetic retinopathy is a leading cause of preventable blindness, yet access to regular screening remains limited worldwide. AI offers a scalable approach, but its performance can be sensitive to image quality and acquisition conditions. In a Brazilian cohort of 8,524 patients, retinal images acquired with the Nikon NF5050 were more than twice as likely to have inadequate image quality compared with those acquired with the Canon CR-2, despite refurbished Canon systems costing more than twice as much as Nikon systems. This potential cost-quality trade-off highlights an important consideration for the reliability of AI systems deployed across heterogeneous imaging systems. Using a frozen pretrained retinal foundation model, we systematically evaluate whether camera-associated performance differences persist across 12 retinal disease classification tasks, four input resolutions, and five downstream classifiers. We find that camera-associated performance differences vary substantially across tasks, and that neither higher input resolution nor classifier choice consistently mitigates cross-camera variation. These findings identify camera platform as a potentially overlooked source of performance variability, with implications for the safety and reliability of retinal AI when deployed across heterogeneous imaging systems. We release all source code, evaluation scripts, and per-task results on GitHub to support reproducible benchmarking.