Feng Li1, Yuguang Wang1, Tianyi Xu2, Lin Dong1, Lei Yan1, Minshan Jiang3, Xuedian Zhang4,5, Hong Jiang6, Zhizheng Wu7, Haidong Zou8. 1. School of Optical-Electrical and Computer Engineering, University of Shanghai for Science and Technology, Shanghai, China. 2. Deparement of Anesthesia, Shanghai Ninth People's Hospital, Shanghai JiaoTong University School of Medicine, Shanghai, China. 3. School of Optical-Electrical and Computer Engineering, University of Shanghai for Science and Technology, Shanghai, China. jiangmsc@gmail.com. 4. School of Optical-Electrical and Computer Engineering, University of Shanghai for Science and Technology, Shanghai, China. xdzhang@usst.edu.cn. 5. Shanghai Institute of Intelligent Science and Technology, Tongji University, Shanghai, China. xdzhang@usst.edu.cn. 6. Deparement of Anesthesia, Shanghai Ninth People's Hospital, Shanghai JiaoTong University School of Medicine, Shanghai, China. jianghongjiuyuan@163.com. 7. Department of Precision Mechanical Engineering, Shanghai University, Shanghai, China. 8. Department of Ophthalmology, Shanghai First People's Hospital, Shanghai, China.
Abstract
OBJECTIVES: To present and validate a deep ensemble algorithm to detect diabetic retinopathy (DR) and diabetic macular oedema (DMO) using retinal fundus images. METHODS: A total of 8739 retinal fundus images were collected from a retrospective cohort of 3285 patients. For detecting DR and DMO, a multiple improved Inception-v4 ensembling approach was developed. We measured the algorithm's performance and made a comparison with that of human experts on our primary dataset, while its generalization was assessed on the publicly available Messidor-2 dataset. Also, we investigated systematically the impact of the size and number of input images used in training on model's performance, respectively. Further, the time budget of training/inference versus model performance was analyzed. RESULTS: On our primary test dataset, the model achieved an 0.992 (95% CI, 0.989-0.995) AUC corresponding to 0.925 (95% CI, 0.916-0.936) sensitivity and 0.961 (95% CI, 0.950-0.972) specificity for referable DR, while the sensitivity and specificity for ophthalmologists ranged from 0.845 to 0.936, and from 0.912 to 0.971, respectively. For referable DMO, our model generated an AUC of 0.994 (95% CI, 0.992-0.996) with a 0.930 (95% CI, 0.919-0.941) sensitivity and 0.971 (95% CI, 0.965-0.978) specificity, whereas ophthalmologists obtained sensitivities ranging between 0.852 and 0.946, and specificities ranging between 0.926 and 0.985. CONCLUSION: This study showed that the deep ensemble model exhibited excellent performance in detecting DR and DMO, and had good robustness and generalization, which could potentially help support and expand DR/DMO screening programs.
OBJECTIVES: To present and validate a deep ensemble algorithm to detect diabetic retinopathy (DR) and diabetic macular oedema (DMO) using retinal fundus images. METHODS: A total of 8739 retinal fundus images were collected from a retrospective cohort of 3285 patients. For detecting DR and DMO, a multiple improved Inception-v4 ensembling approach was developed. We measured the algorithm's performance and made a comparison with that of human experts on our primary dataset, while its generalization was assessed on the publicly available Messidor-2 dataset. Also, we investigated systematically the impact of the size and number of input images used in training on model's performance, respectively. Further, the time budget of training/inference versus model performance was analyzed. RESULTS: On our primary test dataset, the model achieved an 0.992 (95% CI, 0.989-0.995) AUC corresponding to 0.925 (95% CI, 0.916-0.936) sensitivity and 0.961 (95% CI, 0.950-0.972) specificity for referable DR, while the sensitivity and specificity for ophthalmologists ranged from 0.845 to 0.936, and from 0.912 to 0.971, respectively. For referable DMO, our model generated an AUC of 0.994 (95% CI, 0.992-0.996) with a 0.930 (95% CI, 0.919-0.941) sensitivity and 0.971 (95% CI, 0.965-0.978) specificity, whereas ophthalmologists obtained sensitivities ranging between 0.852 and 0.946, and specificities ranging between 0.926 and 0.985. CONCLUSION: This study showed that the deep ensemble model exhibited excellent performance in detecting DR and DMO, and had good robustness and generalization, which could potentially help support and expand DR/DMO screening programs.
Authors: Philippe M Burlina; Neil Joshi; Katia D Pacheco; T Y Alvin Liu; Neil M Bressler Journal: JAMA Ophthalmol Date: 2019-03-01 Impact factor: 7.389
Authors: C P Wilkinson; Frederick L Ferris; Ronald E Klein; Paul P Lee; Carl David Agardh; Matthew Davis; Diana Dills; Anselm Kampik; R Pararajasegaram; Juan T Verdaguer Journal: Ophthalmology Date: 2003-09 Impact factor: 12.079
Authors: Stuart Keel; Pei Ying Lee; Jane Scheetz; Zhixi Li; Mark A Kotowicz; Richard J MacIsaac; Mingguang He Journal: Sci Rep Date: 2018-03-12 Impact factor: 4.379
Authors: Jaakko Sahlsten; Joel Jaskari; Jyri Kivinen; Lauri Turunen; Esa Jaanio; Kustaa Hietala; Kimmo Kaski Journal: Sci Rep Date: 2019-07-24 Impact factor: 4.379