PMID- 29653860 OWN - NLM STAT- MEDLINE DCOM- 20190905 LR - 20190906 IS - 1549-4713 (Electronic) IS - 0161-6420 (Linking) VI - 125 IP - 9 DP - 2018 Sep TI - A Deep Learning Algorithm for Prediction of Age-Related Eye Disease Study Severity Scale for Age-Related Macular Degeneration from Color Fundus Photography. PG - 1410-1420 LID - S0161-6420(17)33064-6 [pii] LID - 10.1016/j.ophtha.2018.02.037 [doi] AB - PURPOSE: Age-related macular degeneration (AMD) is a common threat to vision. While classification of disease stages is critical to understanding disease risk and progression, several systems based on color fundus photographs are known. Most of these require in-depth and time-consuming analysis of fundus images. Herein, we present an automated computer-based classification algorithm. DESIGN: Algorithm development for AMD classification based on a large collection of color fundus images. Validation is performed on a cross-sectional, population-based study. PARTICIPANTS: We included 120 656 manually graded color fundus images from 3654 Age-Related Eye Disease Study (AREDS) participants. AREDS participants were >55 years of age, and non-AMD sight-threatening diseases were excluded at recruitment. In addition, performance of our algorithm was evaluated in 5555 fundus images from the population-based Kooperative Gesundheitsforschung in der Region Augsburg (KORA; Cooperative Health Research in the Region of Augsburg) study. METHODS: We defined 13 classes (9 AREDS steps, 3 late AMD stages, and 1 for ungradable images) and trained several convolution deep learning architectures. An ensemble of network architectures improved prediction accuracy. An independent dataset was used to evaluate the performance of our algorithm in a population-based study. MAIN OUTCOME MEASURES: kappa Statistics and accuracy to evaluate the concordance between predicted and expert human grader classification. RESULTS: A network ensemble of 6 different neural net architectures predicted the 13 classes in the AREDS test set with a quadratic weighted kappa of 92% (95% confidence interval, 89%-92%) and an overall accuracy of 63.3%. In the independent KORA dataset, images wrongly classified as AMD were mainly the result of a macular reflex observed in young individuals. By restricting the KORA analysis to individuals >55 years of age and prior exclusion of other retinopathies, the weighted and unweighted kappa increased to 50% and 63%, respectively. Importantly, the algorithm detected 84.2% of all fundus images with definite signs of early or late AMD. Overall, 94.3% of healthy fundus images were classified correctly. CONCLUSIONS: Our deep learning algoritm revealed a weighted kappa outperforming human graders in the AREDS study and is suitable to classify AMD fundus images in other datasets using individuals >55 years of age. CI - Copyright (c) 2018 American Academy of Ophthalmology. Published by Elsevier Inc. All rights reserved. FAU - Grassmann, Felix AU - Grassmann F AD - Institute of Human Genetics, University of Regensburg, Regensburg, Germany. FAU - Mengelkamp, Judith AU - Mengelkamp J AD - Institute of Human Genetics, University of Regensburg, Regensburg, Germany; Regensburg Medical Image Computing, Ostbayerische Technische Hochschule Regensburg (OTH Regensburg), Regensburg, Germany. FAU - Brandl, Caroline AU - Brandl C AD - Institute of Human Genetics, University of Regensburg, Regensburg, Germany; Department of Ophthalmology, University Hospital Regensburg, Regensburg, Germany; Department of Genetic Epidemiology, Institute of Epidemiology, University of Regensburg, Regensburg, Germany. FAU - Harsch, Sebastian AU - Harsch S AD - Institute of Human Genetics, University of Regensburg, Regensburg, Germany. FAU - Zimmermann, Martina E AU - Zimmermann ME AD - Department of Ophthalmology, University Hospital Regensburg, Regensburg, Germany. FAU - Linkohr, Birgit AU - Linkohr B AD - Institute of Epidemiology II, Helmholtz Zentrum Munchen-German Research Center for Environmental Health, Neuherberg, Germany. FAU - Peters, Annette AU - Peters A AD - Institute of Epidemiology II, Helmholtz Zentrum Munchen-German Research Center for Environmental Health, Neuherberg, Germany. FAU - Heid, Iris M AU - Heid IM AD - Department of Ophthalmology, University Hospital Regensburg, Regensburg, Germany. FAU - Palm, Christoph AU - Palm C AD - Regensburg Medical Image Computing, Ostbayerische Technische Hochschule Regensburg (OTH Regensburg), Regensburg, Germany; Regensburg Center of Biomedical Engineering (RCBE), OTH Regensburg and Regensburg University, Regensburg, Germany. Electronic address: christoph.palm@oth-regensburg.de. FAU - Weber, Bernhard H F AU - Weber BHF AD - Institute of Human Genetics, University of Regensburg, Regensburg, Germany. Electronic address: bweb@klinik.uni-regensburg.de. LA - eng PT - Journal Article PT - Research Support, N.I.H., Extramural PT - Research Support, Non-U.S. Gov't DEP - 20180410 PL - United States TA - Ophthalmology JT - Ophthalmology JID - 7802443 SB - IM MH - Aged MH - *Algorithms MH - Cross-Sectional Studies MH - *Deep Learning MH - *Diagnostic Techniques, Ophthalmological MH - Female MH - Fundus Oculi MH - Humans MH - Macula Lutea/*pathology MH - Macular Degeneration/*diagnosis MH - Male MH - Middle Aged MH - Photography MH - Reproducibility of Results MH - Severity of Illness Index EDAT- 2018/04/15 06:00 MHDA- 2019/09/07 06:00 CRDT- 2018/04/15 06:00 PHST- 2017/10/05 00:00 [received] PHST- 2018/02/20 00:00 [revised] PHST- 2018/02/27 00:00 [accepted] PHST- 2018/04/15 06:00 [pubmed] PHST- 2019/09/07 06:00 [medline] PHST- 2018/04/15 06:00 [entrez] AID - S0161-6420(17)33064-6 [pii] AID - 10.1016/j.ophtha.2018.02.037 [doi] PST - ppublish SO - Ophthalmology. 2018 Sep;125(9):1410-1420. doi: 10.1016/j.ophtha.2018.02.037. Epub 2018 Apr 10.