Development and Validation of a Deep Learning System for Diabetic Retinopathy and Related Eye Diseases Using Retinal Images From Multiethnic Populations With Diabetes
Problem
Detecting referable and vision-threatening diabetic retinopathy, possible glaucoma, and age-related macular degeneration from retinal images in multiethnic populations with diabetes.
Why it matters
Automated screening could extend eye-disease detection across large community and clinic-based diabetic populations.
What the AI did
A deep learning system was trained on tens of thousands of labelled retinal images per disease and validated on hundreds of thousands of images from multiethnic populations with diabetes.
Results
Evaluated on 494,661 retinal images; validated on 112,648 DR, 71,896 glaucoma, and 35,948 AMD images; performance metrics truncated in abstract.
Impact
The DLS was trained and validated for detecting multiple eye diseases across multiethnic diabetic populations.
Record
| Venue | JAMA |
|---|---|
| Published | 2017 |
| Work type | article |
| Authors | Daniel Shu Wei Ting, Carol Yim-lui Cheung, Gilbert Lim, Gavin Siew Wei Tan, Duc Quang Nguyen, Alfred Tau Liang Gan, Haslina Hamzah, Renata García-Franco et al. |
| Citations | 2,383 (OpenAlex, at import time) |
| References | 46 |
| Model or method | deep learning system (architecture not stated in source) |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1001/jama.2017.18152 |
| Open access | bronze |
| Keywords | deep learning, diabetic retinopathy, glaucoma, AMD, retinal screening |
Importance rating
significant — thousands to millions affected
Abstract quantifies large-scale validation: 494,661 retinal images across multiethnic diabetic populations, a national screening program and 10 cohorts, supporting screening of diabetic eye disease.