TY  - JOUR
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.
AU  - Grassmann, F.*
AU  - Mengelkamp, J.*
AU  - Brandl, C.*
AU  - Harsch, S.*
AU  - Zimmermann, M.E.*
AU  - Linkohr, B.
AU  - Peters, A.
AU  - Heid, I.M.*
AU  - Palm, C.*
AU  - Weber, B.H.F.*
C1  - 53402
C2  - 44844
CY  - 360 Park Ave South, New York, Ny 10010-1710 Usa
SP  - 1410-1420
TI  - A deep learning algorithm for prediction of age-related eye disease study severity scale for age-related macular degeneration from color fundus photography.
JO  - Ophthalmology
VL  - 125
IS  - 9
PB  - Elsevier Science Inc
PY  - 2018
SN  - 0161-6420
ER  - 

TY  - JOUR
AB  - PURPOSE: To investigate whether myopia is becoming more common across Europe and explore whether increasing education levels, an important environmental risk factor for myopia, might explain any temporal trend. DESIGN: Meta-analysis of population-based, cross-sectional studies from the European Eye Epidemiology (E(3)) Consortium. PARTICIPANTS: The E(3) Consortium is a collaborative network of epidemiological studies of common eye diseases in adults across Europe. Refractive data were available for 61 946 participants from 15 population-based studies performed between 1990 and 2013; participants had a range of median ages from 44 to 78 years. METHODS: Noncycloplegic refraction, year of birth, and highest educational level achieved were obtained for all participants. Myopia was defined as a mean spherical equivalent ≤-0.75 diopters. A random-effects meta-analysis of age-specific myopia prevalence was performed, with sequential analyses stratified by year of birth and highest level of educational attainment. MAIN OUTCOME MEASURES: Variation in age-specific myopia prevalence for differing years of birth and educational level. RESULTS: There was a significant cohort effect for increasing myopia prevalence across more recent birth decades; age-standardized myopia prevalence increased from 17.8% (95% confidence interval [CI], 17.6-18.1) to 23.5% (95% CI, 23.2-23.7) in those born between 1910 and 1939 compared with 1940 and 1979 (P = 0.03). Education was significantly associated with myopia; for those completing primary, secondary, and higher education, the age-standardized prevalences were 25.4% (CI, 25.0-25.8), 29.1% (CI, 28.8-29.5), and 36.6% (CI, 36.1-37.2), respectively. Although more recent birth cohorts were more educated, this did not fully explain the cohort effect. Compared with the reference risk of participants born in the 1920s with only primary education, higher education or being born in the 1960s doubled the myopia prevalence ratio-2.43 (CI, 1.26-4.17) and 2.62 (CI, 1.31-5.00), respectively-whereas individuals born in the 1960s and completing higher education had approximately 4 times the reference risk: a prevalence ratio of 3.76 (CI, 2.21-6.57). CONCLUSIONS: Myopia is becoming more common in Europe; although education levels have increased and are associated with myopia, higher education seems to be an additive rather than explanatory factor. Increasing levels of myopia carry significant clinical and economic implications, with more people at risk of the sight-threatening complications associated with high myopia.
AU  - Williams, K.M.*
AU  - Bertelsen, G.*
AU  - Cumberland, P.M.*
AU  - Wolfram, C.*
AU  - Verhoeven, V.J.*
AU  - Anastasopoulos, E.*
AU  - Buitendijk, G.H.*
AU  - Cougnard-Grégoire, A.*
AU  - Creuzot-Garcher, C.*
AU  - Erke, M.G.*
AU  - Hogg, R.*
AU  - Höhn, R.*
AU  - Hysi, P.*
AU  - Khawaja, A.P.*
AU  - Korobelnik, J.F.*
AU  - Ried, J.S.
AU  - Vingerling, J.R.*
AU  - Bron, A.*
AU  - Dartigues, J.F.*
AU  - Fletcher, A.*
AU  - Hofman, A.*
AU  - Kuijpers, R.W.*
AU  - Luben, R.N.*
AU  - Oxele, K.*
AU  - Topouzis, F.*
AU  - von Hanno, T.*
AU  - Mirshahi, A.*
AU  - Foster, P.J.*
AU  - van Duijn, C.M.*
AU  - Pfeiffer, N.*
AU  - Delcourt, C.*
AU  - Klaver, C.C.*
AU  - Rahi, J.S.*
AU  - Hammond, C.J.*
C1  - 44890
C2  - 37170
CY  - New York
SP  - 1489-1497
TI  - Increasing prevalence of myopia in Europe and the impact of education.
JO  - Ophthalmology
VL  - 122
IS  - 7
PB  - Elsevier Science Inc
PY  - 2015
SN  - 0161-6420
ER  - 

TY  - JOUR
AB  - PURPOSE: Intravitreal anti-vascular endothelial growth factor (VEGF) injections are currently the standard treatment for neovascular age-related macular degeneration (AMD), but a broad range of response rates has been observed. We evaluated the association of single nucleotide polymorphisms (SNPs) in VEGF genes and their receptors (VEGFR) with the response rate to ranibizumab in 366 patients with neovascular AMD. DESIGN: Case series study. PARTICIPANTS: A total of 366 eyes of 366 patients with neovascular AMD. METHODS: Visual acuity (VA) was determined at baseline, after 3 monthly ranibizumab injections, and after 1 year of treatment. Genotyping of 126 SNPs in the genes encoding VEGF family members VEGFA, VEGFB, VEGFC, VEGFD (FIGF), and placental growth factor (PGF); VEGF receptors VEGFR1 (FLT1), VEGFR2 (KDR), and VEGFR3 (FLT4); and the gene encoding pigment epithelium-derived factor (PEDF) (SERPINF1) was performed. MAIN OUTCOME MEASURES: The changes in VA after 3 injections and after 1 year of treatment and their association with VEGF and VEGFR genotypes. RESULTS: Univariate analyses of variance (ANOVAs) revealed a significant effect of SNP rs4576072 in the VEGFR2 gene on VA change after 12 months (F[1,235] = 14.05; P = 0.02). A stepwise linear regression analysis returned a model (P = 0.01) with SNPs rs4576072 and rs6828477 in the VEGFR2 gene as independent predictors for VA change after 12 months, with a mean increase in VA of 0.26 on the logarithm of the minimum angle of resolution (logMAR) scale in patients with 3 contributing minor alleles compared with a loss of 0.03 logMAR in patients with no minor allele. CONCLUSIONS: Polymorphisms in the VEGFR2/KDR gene significantly influence visual outcome in patients receiving ranibizumab treatment for neovascular AMD. This study shows that genetic variation partially explains the wide range of response to ranibizumab treatment, which in the future might help clinicians tailoring medical interventions to individual needs. FINANCIAL DISCLOSURE(S): The author(s) have no proprietary or commercial interest in any materials discussed in this article.
AU  - Hermann, M.M.*
AU  - van Asten, F.*
AU  - Muether, P.S.*
AU  - Smailhodzic, D.*
AU  - Lichtner, P.
AU  - Hoyng, C.B.*
AU  - Kirchhof, B.*
AU  - Grefkes, C.*
AU  - den Hollander, A.I.*
AU  - Fauser, S.*
C1  - 29085
C2  - 33628
CY  - New York
SP  - 905-910
TI  - Polymorphisms in vascular endothelial growth factor receptor 2 are associated with better response rates to ranibizumab treatment in age-related macular degeneration.
JO  - Ophthalmology
VL  - 121
IS  - 4
PB  - Elsevier Science Inc
PY  - 2014
SN  - 0161-6420
ER  -