Tiarnan D L Keenan1, Qingyu Chen2, Elvira Agrón3, Yih-Chung Tham4, Jocelyn Hui Lin Goh5, Xiaofeng Lei6, Yi Pin Ng6, Yong Liu7, Xinxing Xu7, Ching-Yu Cheng8, Mukharram M Bikbov9, Jost B Jonas10, Sanjeeb Bhandari3, Geoffrey K Broadhead3, Marcus H Colyer11, Jonathan Corsini12, Chantal Cousineau-Krieger3, William Gensheimer13, David Grasic3, Tania Lamba14, M Teresa Magone3, Michele Maiberger14, Arnold Oshinsky14, Boonkit Purt15, Soo Y Shin14, Alisa T Thavikulwat3, Zhiyong Lu16, Emily Y Chew17. 1. Division of Epidemiology and Clinical Applications, National Eye Institute, National Institutes of Health, Bethesda, Maryland. Electronic address: tiarnan.keenan@nih.gov. 2. National Center for Biotechnology Information, National Library of Medicine, National Institutes of Health, Bethesda, Maryland. Electronic address: qingyu.chen@nih.gov. 3. Division of Epidemiology and Clinical Applications, National Eye Institute, National Institutes of Health, Bethesda, Maryland. 4. Singapore Eye Research Institute, Singapore National Eye Centre, Singapore; Duke-NUS Medical School, Singapore. 5. Singapore Eye Research Institute, Singapore National Eye Centre, Singapore. 6. Institute of High Performance Computing, A∗STAR, Singapore. 7. Duke-NUS Medical School, Singapore; Institute of High Performance Computing, A∗STAR, Singapore. 8. Singapore Eye Research Institute, Singapore National Eye Centre, Singapore; Duke-NUS Medical School, Singapore; Institute of High Performance Computing, A∗STAR, Singapore; Department of Ophthalmology, Yong Loo Lin School of Medicine, National University of Singapore, Singapore. 9. Ufa Eye Research Institute, Ufa, Russia. 10. Department of Ophthalmology, Medical Faculty Mannheim, Heidelberg University, Mannheim, Germany; Institute of Molecular and Clinical Ophthalmology, Basel, Switzerland; Privatpraxis Prof Jonas und Dr Panda-Jonas, Heidelberg, Germany. 11. Department of Ophthalmology, Madigan Army Medical Center, Tacoma, Washington; Department of Surgery, Uniformed Services University of the Health Sciences, Bethesda, Maryland. 12. Warfighter Eye Center, Malcolm Grow Medical Clinics and Surgery Center, Joint Base Andrews, Maryland. 13. White River Junction Veterans Affairs Medical Center, White River Junction, Vermont; Geisel School of Medicine, Dartmouth, New Hampshire. 14. Washington DC Veterans Affairs Medical Center, Washington, D.C. 15. Department of Surgery, Uniformed Services University of the Health Sciences, Bethesda, Maryland; Department of Ophthalmology, Walter Reed National Military Medical Center, Bethesda, Maryland. 16. National Center for Biotechnology Information, National Library of Medicine, National Institutes of Health, Bethesda, Maryland. Electronic address: luzh@ncbi.nlm.nih.gov. 17. Division of Epidemiology and Clinical Applications, National Eye Institute, National Institutes of Health, Bethesda, Maryland. Electronic address: echew@nei.nih.gov.
Abstract
PURPOSE: To develop deep learning models to perform automated diagnosis and quantitative classification of age-related cataract from anterior segment photographs. DESIGN: DeepLensNet was trained by applying deep learning models to the Age-Related Eye Disease Study (AREDS) dataset. PARTICIPANTS: A total of 18 999 photographs (6333 triplets) from longitudinal follow-up of 1137 eyes (576 AREDS participants). METHODS: Deep learning models were trained to detect and quantify nuclear sclerosis (NS; scale 0.9-7.1) from 45-degree slit-lamp photographs and cortical lens opacity (CLO; scale 0%-100%) and posterior subcapsular cataract (PSC; scale 0%-100%) from retroillumination photographs. DeepLensNet performance was compared with that of 14 ophthalmologists and 24 medical students. MAIN OUTCOME MEASURES: Mean squared error (MSE). RESULTS: On the full test set, mean MSE for DeepLensNet was 0.23 (standard deviation [SD], 0.01) for NS, 13.1 (SD, 1.6) for CLO, and 16.6 (SD, 2.4) for PSC. On a subset of the test set (substantially enriched for positive cases of CLO and PSC), for NS, mean MSE for DeepLensNet was 0.23 (SD, 0.02), compared with 0.98 (SD, 0.24; P = 0.000001) for the ophthalmologists and 1.24 (SD, 0.34; P = 0.000005) for the medical students. For CLO, mean MSE was 53.5 (SD, 14.8), compared with 134.9 (SD, 89.9; P = 0.003) for the ophthalmologists and 433.6 (SD, 962.1; P = 0.0007) for the medical students. For PSC, mean MSE was 171.9 (SD, 38.9), compared with 176.8 (SD, 98.0; P = 0.67) for the ophthalmologists and 398.2 (SD, 645.4; P = 0.18) for the medical students. In external validation on the Singapore Malay Eye Study (sampled to reflect the cataract severity distribution in AREDS), the MSE for DeepSeeNet was 1.27 for NS and 25.5 for PSC. CONCLUSIONS: DeepLensNet performed automated and quantitative classification of cataract severity for all 3 types of age-related cataract. For the 2 most common types (NS and CLO), the accuracy was significantly superior to that of ophthalmologists; for the least common type (PSC), it was similar. DeepLensNet may have wide potential applications in both clinical and research domains. In the future, such approaches may increase the accessibility of cataract assessment globally. The code and models are available at https://github.com/ncbi/deeplensnet.
PURPOSE: To develop deep learning models to perform automated diagnosis and quantitative classification of age-related cataract from anterior segment photographs. DESIGN: DeepLensNet was trained by applying deep learning models to the Age-Related Eye Disease Study (AREDS) dataset. PARTICIPANTS: A total of 18 999 photographs (6333 triplets) from longitudinal follow-up of 1137 eyes (576 AREDS participants). METHODS: Deep learning models were trained to detect and quantify nuclear sclerosis (NS; scale 0.9-7.1) from 45-degree slit-lamp photographs and cortical lens opacity (CLO; scale 0%-100%) and posterior subcapsular cataract (PSC; scale 0%-100%) from retroillumination photographs. DeepLensNet performance was compared with that of 14 ophthalmologists and 24 medical students. MAIN OUTCOME MEASURES: Mean squared error (MSE). RESULTS: On the full test set, mean MSE for DeepLensNet was 0.23 (standard deviation [SD], 0.01) for NS, 13.1 (SD, 1.6) for CLO, and 16.6 (SD, 2.4) for PSC. On a subset of the test set (substantially enriched for positive cases of CLO and PSC), for NS, mean MSE for DeepLensNet was 0.23 (SD, 0.02), compared with 0.98 (SD, 0.24; P = 0.000001) for the ophthalmologists and 1.24 (SD, 0.34; P = 0.000005) for the medical students. For CLO, mean MSE was 53.5 (SD, 14.8), compared with 134.9 (SD, 89.9; P = 0.003) for the ophthalmologists and 433.6 (SD, 962.1; P = 0.0007) for the medical students. For PSC, mean MSE was 171.9 (SD, 38.9), compared with 176.8 (SD, 98.0; P = 0.67) for the ophthalmologists and 398.2 (SD, 645.4; P = 0.18) for the medical students. In external validation on the Singapore Malay Eye Study (sampled to reflect the cataract severity distribution in AREDS), the MSE for DeepSeeNet was 1.27 for NS and 25.5 for PSC. CONCLUSIONS: DeepLensNet performed automated and quantitative classification of cataract severity for all 3 types of age-related cataract. For the 2 most common types (NS and CLO), the accuracy was significantly superior to that of ophthalmologists; for the least common type (PSC), it was similar. DeepLensNet may have wide potential applications in both clinical and research domains. In the future, such approaches may increase the accessibility of cataract assessment globally. The code and models are available at https://github.com/ncbi/deeplensnet.
Authors: Carol Yim-lui Cheung; Huiqi Li; Ecosse L Lamoureux; Paul Mitchell; Jie Jin Wang; Ava Grace Tan; Lily K Johari; Jiang Liu; Joo Hwee Lim; Tin Aung; Tien Yin Wong Journal: Invest Ophthalmol Vis Sci Date: 2011-03-10 Impact factor: 4.799
Authors: Tiarnan D L Keenan; Qingyu Chen; Yifan Peng; Amitha Domalpally; Elvira Agrón; Christopher K Hwang; Alisa T Thavikulwat; Debora H Lee; Daniel Li; Wai T Wong; Zhiyong Lu; Emily Y Chew Journal: Ophthalmology Date: 2020-05-21 Impact factor: 12.079
Authors: Saad M Khan; Xiaoxuan Liu; Siddharth Nath; Edward Korot; Livia Faes; Siegfried K Wagner; Pearse A Keane; Neil J Sebire; Matthew J Burton; Alastair K Denniston Journal: Lancet Digit Health Date: 2020-10-01