Project description:ImportanceDeep learning (DL) used for discriminative tasks in ophthalmology, such as diagnosing diabetic retinopathy or age-related macular degeneration (AMD), requires large image data sets graded by human experts to train deep convolutional neural networks (DCNNs). In contrast, generative DL techniques could synthesize large new data sets of artificial retina images with different stages of AMD. Such images could enhance existing data sets of common and rare ophthalmic diseases without concern for personally identifying information to assist medical education of students, residents, and retinal specialists, as well as for training new DL diagnostic models for which extensive data sets from large clinical trials of expertly graded images may not exist.ObjectiveTo develop DL techniques for synthesizing high-resolution realistic fundus images serving as proxy data sets for use by retinal specialists and DL machines.Design, setting, and participantsGenerative adversarial networks were trained on 133 821 color fundus images from 4613 study participants from the Age-Related Eye Disease Study (AREDS), generating synthetic fundus images with and without AMD. We compared retinal specialists' ability to diagnose AMD on both real and synthetic images, asking them to assess image gradability and testing their ability to discern real from synthetic images. The performance of AMD diagnostic DCNNs (referable vs not referable AMD) trained on either all-real vs all-synthetic data sets was compared.Main outcomes and measuresAccuracy of 2 retinal specialists (T.Y.A.L. and K.D.P.) for diagnosing and distinguishing AMD on real vs synthetic images and diagnostic performance (area under the curve) of DL algorithms trained on synthetic vs real images.ResultsThe diagnostic accuracy of 2 retinal specialists on real vs synthetic images was similar. The accuracy of diagnosis as referable vs nonreferable AMD compared with certified human graders for retinal specialist 1 was 84.54% (error margin, 4.06%) on real images vs 84.12% (error margin, 4.16%) on synthetic images and for retinal specialist 2 was 89.47% (error margin, 3.45%) on real images vs 89.19% (error margin, 3.54%) on synthetic images. Retinal specialists could not distinguish real from synthetic images, with an accuracy of 59.50% (error margin, 3.93%) for retinal specialist 1 and 53.67% (error margin, 3.99%) for retinal specialist 2. The DCNNs trained on real data showed an area under the curve of 0.9706 (error margin, 0.0029), and those trained on synthetic data showed an area under the curve of 0.9235 (error margin, 0.0045).Conclusions and relevanceDeep learning-synthesized images appeared to be realistic to retinal specialists, and DCNNs achieved diagnostic performance on synthetic data close to that for real images, suggesting that DL generative techniques hold promise for training humans and machines.

Project description:The first goal of the study was to find the genes that explain the difference in progression of age-related retinal degeneration (ageRD) between the inbred strains C57BL/6J-c2J (B6a) and BALB/cByJ (C). The specific goal was to find the candidate genes that explain the chromosome 6 and chromosome 10 quantitative trait loci (QTL) found in a previous study. Genes differentially expressed in retina between both strains and that map to the QTL are candidate genes. By integrating different approaches: microarray study, knowledge-based candidate gene search and screening of genetic variants, we found 35 candidate genes for chromosome 6 QTL and 14 candidate genes for chromosome 10 QTL. Five genes for each QTL were selected for sequencing and qPCR analysis. One gene, Tnfaip3 was selected for phenotypic testing and was excluded as the quantitative trait gene for chr10 QTL. The second goal of the study was to find biological processes and pathways associated with ageRD by Gene Ontology analysis of the microarray data. We concentrated on GO terms overrepresented in either strain at 8 months; the age when there is a difference in the phenotype (outer nuclear layer thickness). The GO analysis revealed that transcripts associated with inflammation, mitochondrial envelope, DNA repairing and oxidative stress were increased in the ageRD sensitive C strain with age (8 months). In the ageRD resistant B6a strain transcripts associated with antioxidant response, regulation of blood vessel size and growth factor activity were increased with age. Transcripts associated with cell remodeling and lipid metabolism were increased in either strain with age. This study aims to identify gene expression differences in posterior eyecups between the strains C and B6a at 4 months and at 8 months. Twelve animals were used per strain and time point for a total of 48 mice. Total RNA for 3 animals (6 eyes) were pooled giving a total of 4 biological replicates per strain and time point. The 4-month C samples labeled with Cy5 were hybridized with the 4-month B6a samples labeled with Cy3 (4 biological replicates); the 8-month C samples labeled with Cy5 were hybridized with the 8-month B6a samples labeled with Cy3, for 8 arrays (4 biological replicates). A second set of eight arrays were used to analyzed the same RNA samples but with the dye orientation reversed, yielding a total of 16 arrays in the experiment. Set of arrays organized by shared biological context, such as organism, tumors types, processes, etc.

Project description:The first goal of the study was to find the genes that explain the difference in progression of age-related retinal degeneration (ageRD) between the inbred strains C57BL/6J-c2J (B6a) and BALB/cByJ (C). The specific goal was to find the candidate genes that explain the chromosome 6 and chromosome 10 quantitative trait loci (QTL) found in a previous study. Genes differentially expressed in retina between both strains and that map to the QTL are candidate genes. By integrating different approaches: microarray study, knowledge-based candidate gene search and screening of genetic variants, we found 35 candidate genes for chromosome 6 QTL and 14 candidate genes for chromosome 10 QTL. Five genes for each QTL were selected for sequencing and qPCR analysis. One gene, Tnfaip3 was selected for phenotypic testing and was excluded as the quantitative trait gene for chr10 QTL. The second goal of the study was to find biological processes and pathways associated with ageRD by Gene Ontology analysis of the microarray data. We concentrated on GO terms overrepresented in either strain at 8 months; the age when there is a difference in the phenotype (outer nuclear layer thickness). The GO analysis revealed that transcripts associated with inflammation, mitochondrial envelope, DNA repairing and oxidative stress were increased in the ageRD sensitive C strain with age (8 months). In the ageRD resistant B6a strain transcripts associated with antioxidant response, regulation of blood vessel size and growth factor activity were increased with age. Transcripts associated with cell remodeling and lipid metabolism were increased in either strain with age. This study aims to identify gene expression differences in posterior eyecups between the strains C and B6a at 4 months and at 8 months. Twelve animals were used per strain and time point for a total of 48 mice. Total RNA for 3 animals (6 eyes) were pooled giving a total of 4 biological replicates per strain and time point. The 4-month C samples labeled with Cy5 were hybridized with the 4-month B6a samples labeled with Cy3 (4 biological replicates); the 8-month C samples labeled with Cy5 were hybridized with the 8-month B6a samples labeled with Cy3, for 8 arrays (4 biological replicates). A second set of eight arrays were used to analyzed the same RNA samples but with the dye orientation reversed, yielding a total of 16 arrays in the experiment.

Project description:PurposeTo describe the value of integrating phenotype/genotype data, disease staging, and evaluation of functional vision in patient-centered management of retinal dystrophies.Methods(1) Cross-sectional structure-function and retrospective longitudinal studies to assess the correlations between standard fundus autofluorescence (FAF), optical coherence tomography, visual acuity (VA), and perimetry (visual field [VF]) examinations to evaluate photoreceptor functional loss in a cohort of patients with rod-cone dystrophy (RCD); (2) flood-illumination adaptive optics (FIAO) imaging focusing on photoreceptor misalignment and orientation of outer segments; and (3) evaluation of the impact of visual impairment in daily life activities, based on functional (visual and mobility) vision assessment in a naturalistic environment in visually impaired subjects with RCD and subjects treated with LuxturnaⓇ for RPE65-related Leber congenital amaurosis before and after therapy.ResultsThe results of the cross-sectional transversal study showed that (1) VA and macular sensitivity were weakly correlated with the structural variables; and (2) functional impairment (VF) was correlated with reduction of anatomical markers of photoreceptor structure and increased width of autofluorescent ring. The dimensions of the ring of increased FAF evolved faster. Other criteria that differed among groups were the lengths of the ellipsoid zone, the external limiting membrane, and the foveal thickness. FIAO revealed a variety of phenotypes: paradoxical visibility of foveal cones; heterogeneous brightness of cones; dim, inner segment-like, and RPE-like mosaic. Directional illumination by varying orientation of incident light (Stiles-Crawford effect) and the amount of side illumination (gaze-dependent imaging) affected photoreceptor visibility. Mobility assessment under different lighting conditions showed correlation with VF, VA, contrast sensitivity (CS), and dark adaptation, with different predictive values depending on mobility study paradigms and illumination level. At high illumination level (235 lux), VF was a predictor for all mobility performance models. Under low illumination (1 and 2 lux), VF was the most significant predictor of mobility performance variables, while CS best explained the number of collisions and segments. In subjects treated with LuxturnaⓇ, a very favorable impact on travel speed and reduction in the number of collisions, especially at low luminance, was observable 6 months following injection, in both children and adults.ConclusionsOur results suggest the benefit of development and implementation of quantitative and reproducible tools to evaluate the status of photoreceptors and the impact of both visual impairment and novel therapies in real-life conditions. NOTE: Publication of this article is sponsored by the American Ophthalmological Society.

Dataset Information

High-resolution Imaging in Male Germ Cell-Associated Kinase (MAK)-related Retinal Degeneration.

Publications

High-resolution Imaging in Male Germ Cell-Associated Kinase (MAK)-related Retinal Degeneration.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets