A deep learning system has analysed retinal images to detect diabetic retinopathy, glaucoma and age-related macular degeneration
Artificial intelligence has been used to detect diabetic retinopathy, glaucoma and age-related macular degeneration (AMD) in a study published in JAMA.
Researchers trained a deep learning system to recognise the eye conditions then assessed its accuracy across 71,896 retinal images from 14,880 patients.
They found that the system could detect referable diabetic retinopathy with a sensitivity of 90.5% and specificity of 91.6%.
The artificial intelligence technology was even more accurate when detecting vision-threatening diabetic retinopathy – displaying 100% sensitivity and 91.1% specificity.
In terms of possible glaucoma, the system showed 96.4% sensitivity and 87.2% specificity, while the technology detected AMD with 93.2% sensitivity and 88.7% specificity.
The results of the deep learning system were compared to the performance of retinal specialists, general ophthalmologists, trained graders and optometrists as a reference standard.
The authors concluded: “Further research is necessary to evaluate the applicability of the deep learning system in health care settings and the utility of the deep learning system to improve vision outcomes.”