FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W4309761346> ?p ?o ?g. }

• W4309761346 endingPage "100258" @default.
• W4309761346 startingPage "100258" @default.
• W4309761346 abstract "PurposeRare disease diagnosis is challenging in medical image-based artificial intelligence due to a natural class imbalance in datasets, leading to biased prediction models. Inherited retinal diseases (IRDs) are a research domain that particularly faces this issue. This study investigates the applicability of synthetic data in improving AI-enabled diagnosis of IRDs using Generative Adversarial Networks (GANs).DesignDiagnostic study of gene-labeled fundus autofluorescence (FAF) IRD images using Deep Learning (DL).ParticipantsMoorfields eye hospital (MEH) dataset of 15,692 FAF images obtained from 1800 patients with confirmed genetic diagnosis of one of 36 IRD genes.MethodsA StyleGAN2 model is trained on the IRD dataset to generate 512x512 resolution images. Convolutional neural networks (CNNs) are trained for classification using different synthetically augmented datasets, including the real IRD images plus 1800 and 3600 synthetic images, and a fully rebalanced dataset. We also perform an experiment only with synthetic data. All models are compared against a baseline CNN trained only on real data.Main Outcome MeasuresWe evaluated synthetic data quality using a Visual Turing Test (VTT) conducted with four ophthalmologists from MEH. Synthetic and real images were compared in feature space using feature space visualization, and similarity analysis to detect memorized images, and BRISQUE score for no-reference-based quality evaluation. CNN diagnostic performance was determined on a held-out test set using the area-under-receiver operating-characteristic-curve (AUROC) and Cohen’s Kappa (k).ResultsAn average true recognition rate of 63% and fake recognition rate of 47% was obtained from the VTT. Thus, a considerable proportion of the synthetic images were classified as real by clinical experts. Similarity analysis no copied real images, meaning the GAN was able to generalize. Comparing the rebalanced model with the baseline, we found no significant change in the average AUROC and k (R-AUROC=0.86(0.85-88), RB-AUROC=0.88(0.86-0.89), R-k =0.51(0.49-0.53), RB-k =0.52(0.50-0.54)). The synthetic data trained model achieves similar performance as the baseline (S-AUROC=0.86(0.85-87), S-k=0.48(0.46-0.50)).ConclusionsSynthetic generation of realistic IRD FAF images is feasible. Synthetic data augmentation does not deliver improvements in classification performance. However, synthetic data alone delivers a similar performance as real data, hence may be useful as a proxy to real data." @default.
• W4309761346 created "2022-11-29" @default.
• W4309761346 creator A5019091666 @default.
• W4309761346 creator A5021469260 @default.
• W4309761346 creator A5023795910 @default.
• W4309761346 creator A5028939056 @default.
• W4309761346 creator A5041000511 @default.
• W4309761346 creator A5042102084 @default.
• W4309761346 creator A5049231383 @default.
• W4309761346 creator A5049860771 @default.
• W4309761346 creator A5056054539 @default.
• W4309761346 creator A5057960653 @default.
• W4309761346 creator A5067795081 @default.
• W4309761346 creator A5072064334 @default.
• W4309761346 creator A5077508604 @default.
• W4309761346 creator A5083496383 @default.
• W4309761346 creator A5083903120 @default.
• W4309761346 creator A5087735317 @default.
• W4309761346 creator A5090291315 @default.
• W4309761346 date "2023-06-01" @default.
• W4309761346 modified "2023-09-30" @default.
• W4309761346 title "SynthEye: Investigating the Impact of Synthetic Data on Artificial Intelligence-assisted Gene Diagnosis of Inherited Retinal Disease" @default.
• W4309761346 cites W1982471090 @default.
• W4309761346 cites W2381926744 @default.
• W4309761346 cites W2512877629 @default.
• W4309761346 cites W2761299546 @default.
• W4309761346 cites W2810924486 @default.
• W4309761346 cites W2939662580 @default.
• W4309761346 cites W2943735103 @default.
• W4309761346 cites W2985996852 @default.
• W4309761346 cites W2995233992 @default.
• W4309761346 cites W2995939557 @default.
• W4309761346 cites W3017060591 @default.
• W4309761346 cites W3113214702 @default.
• W4309761346 cites W3117085616 @default.
• W4309761346 cites W3138423443 @default.
• W4309761346 cites W3138851647 @default.
• W4309761346 cites W3153758115 @default.
• W4309761346 cites W3202960634 @default.
• W4309761346 cites W3205310854 @default.
• W4309761346 cites W4200331648 @default.
• W4309761346 cites W4212958758 @default.
• W4309761346 cites W4226322299 @default.
• W4309761346 cites W4226348722 @default.
• W4309761346 cites W4297183115 @default.
• W4309761346 doi "https://doi.org/10.1016/j.xops.2022.100258" @default.
• W4309761346 hasPubMedId "https://pubmed.ncbi.nlm.nih.gov/36685715" @default.
• W4309761346 hasPublicationYear "2023" @default.
• W4309761346 type Work @default.
• W4309761346 citedByCount "3" @default.
• W4309761346 countsByYear W43097613462023 @default.
• W4309761346 crossrefType "journal-article" @default.
• W4309761346 hasAuthorship W4309761346A5019091666 @default.
• W4309761346 hasAuthorship W4309761346A5021469260 @default.
• W4309761346 hasAuthorship W4309761346A5023795910 @default.
• W4309761346 hasAuthorship W4309761346A5028939056 @default.
• W4309761346 hasAuthorship W4309761346A5041000511 @default.
• W4309761346 hasAuthorship W4309761346A5042102084 @default.
• W4309761346 hasAuthorship W4309761346A5049231383 @default.
• W4309761346 hasAuthorship W4309761346A5049860771 @default.
• W4309761346 hasAuthorship W4309761346A5056054539 @default.
• W4309761346 hasAuthorship W4309761346A5057960653 @default.
• W4309761346 hasAuthorship W4309761346A5067795081 @default.
• W4309761346 hasAuthorship W4309761346A5072064334 @default.
• W4309761346 hasAuthorship W4309761346A5077508604 @default.
• W4309761346 hasAuthorship W4309761346A5083496383 @default.
• W4309761346 hasAuthorship W4309761346A5083903120 @default.
• W4309761346 hasAuthorship W4309761346A5087735317 @default.
• W4309761346 hasAuthorship W4309761346A5090291315 @default.
• W4309761346 hasBestOaLocation W43097613461 @default.
• W4309761346 hasConcept C118487528 @default.
• W4309761346 hasConcept C119857082 @default.
• W4309761346 hasConcept C138885662 @default.
• W4309761346 hasConcept C148524875 @default.
• W4309761346 hasConcept C153180895 @default.
• W4309761346 hasConcept C154945302 @default.
• W4309761346 hasConcept C160920958 @default.
• W4309761346 hasConcept C16910744 @default.
• W4309761346 hasConcept C169903167 @default.
• W4309761346 hasConcept C199360897 @default.
• W4309761346 hasConcept C22019652 @default.
• W4309761346 hasConcept C2776391266 @default.
• W4309761346 hasConcept C2776401178 @default.
• W4309761346 hasConcept C41008148 @default.
• W4309761346 hasConcept C41895202 @default.
• W4309761346 hasConcept C50644808 @default.
• W4309761346 hasConcept C58471807 @default.
• W4309761346 hasConcept C58489278 @default.
• W4309761346 hasConcept C71924100 @default.
• W4309761346 hasConcept C81363708 @default.
• W4309761346 hasConceptScore W4309761346C118487528 @default.
• W4309761346 hasConceptScore W4309761346C119857082 @default.
• W4309761346 hasConceptScore W4309761346C138885662 @default.
• W4309761346 hasConceptScore W4309761346C148524875 @default.
• W4309761346 hasConceptScore W4309761346C153180895 @default.
• W4309761346 hasConceptScore W4309761346C154945302 @default.
• W4309761346 hasConceptScore W4309761346C160920958 @default.
• W4309761346 hasConceptScore W4309761346C16910744 @default.

• next