Evaluation of Generative Adversarial Networks for High-Resolution Synthetic Image Generation of Circumpapillary Optical Coherence Tomography Images for Glaucoma

Page view(s)
9
Checked on Nov 25, 2024
Evaluation of Generative Adversarial Networks for High-Resolution Synthetic Image Generation of Circumpapillary Optical Coherence Tomography Images for Glaucoma
Title:
Evaluation of Generative Adversarial Networks for High-Resolution Synthetic Image Generation of Circumpapillary Optical Coherence Tomography Images for Glaucoma
Journal Title:
JAMA Ophthalmology
Keywords:
Publication Date:
01 September 2022
Citation:
Sreejith Kumar, A. J., Chong, R. S., Crowston, J. G., Chua, J., Bujor, I., Husain, R., Vithana, E. N., Girard, M. J. A., Ting, D. S. W., Cheng, C.-Y., Aung, T., Popa-Cherecheanu, A., Schmetterer, L., & Wong, D. (2022). Evaluation of Generative Adversarial Networks for High-Resolution Synthetic Image Generation of Circumpapillary Optical Coherence Tomography Images for Glaucoma. JAMA Ophthalmology, 140(10), 974. https://doi.org/10.1001/jamaophthalmol.2022.3375
Abstract:
ImportanceDeep learning (DL) networks require large data sets for training, which can be challenging to collect clinically. Generative models could be used to generate large numbers of synthetic optical coherence tomography (OCT) images to train such DL networks for glaucoma detection.ObjectiveTo assess whether generative models can synthesize circumpapillary optic nerve head OCT images of normal and glaucomatous eyes and determine the usability of synthetic images for training DL models for glaucoma detection.Design, Setting, and ParticipantsProgressively growing generative adversarial network models were trained to generate circumpapillary OCT scans. Image gradeability and authenticity were evaluated on a clinical set of 100 real and 100 synthetic images by 2 clinical experts. DL networks for glaucoma detection were trained with real or synthetic images and evaluated on independent internal and external test data sets of 140 and 300 real images, respectively.Main Outcomes and MeasuresEvaluations of the clinical set between the experts were compared. Glaucoma detection performance of the DL networks was assessed using area under the curve (AUC) analysis. Class activation maps provided visualizations of the regions contributing to the respective classifications.ResultsA total of 990 normal and 862 glaucomatous eyes were analyzed. Evaluations of the clinical set were similar for gradeability (expert 1: 92.0%; expert 2: 93.0%) and authenticity (expert 1: 51.8%; expert 2: 51.3%). The best-performing DL network trained on synthetic images had AUC scores of 0.97 (95% CI, 0.95-0.99) on the internal test data set and 0.90 (95% CI, 0.87-0.93) on the external test data set, compared with AUCs of 0.96 (95% CI, 0.94-0.99) on the internal test data set and 0.84 (95% CI, 0.80-0.87) on the external test data set for the network trained with real images. An increase in the AUC for the synthetic DL network was observed with the use of larger synthetic data set sizes. Class activation maps showed that the regions of the synthetic images contributing to glaucoma detection were generally similar to that of real images.Conclusions and RelevanceDL networks trained with synthetic OCT images for glaucoma detection were comparable with networks trained with real images. These results suggest potential use of generative models in the training of DL networks and as a means of data sharing across institutions without patient information confidentiality issues.
License type:
Attribution 4.0 International (CC BY 4.0)
Funding Info:
This research / project is supported by the National Medical Research Council - NA
Grant Reference no. : CG/C010A/2017_SERI, OFLCG/004c/2018-00, MOH-000249-00, MOH-000647-00, MOH-001001-00, MOH-001015-00, MOH-000500-00, and MOH-000707-00

This research / project is supported by the National Research Foundation Singapore - Retinal Analytics via Machine Learning Aiding Physics
Grant Reference no. : NRF2019-THE002-0006

This research / project is supported by the National Research Foundation Singapore - Competitive Research Programme
Grant Reference no. : NRF-CRP24-2020-0001

This research / project is supported by the Singapore Eye Research Institute & Nanyang Technological University - SERI-NTU Advanced Ocular Engineering [STANCE] Program
Grant Reference no. : NA

This research / project is supported by the SERI-Lee Foundation - NA
Grant Reference no. : LF1019-1

This research / project is supported by the A*STAR - MARIO : Multimodal AI-Driven Decision Making for Ophthalmology
Grant Reference no. : A20H4b0141

This research / project is supported by the National Medical Research Council - Open Fund - Large Collaborative
Grant Reference no. : OFLCG/004c/2018-00

This research / project is supported by the Ministry of Education - Improved Detection of Glaucomatous Structural Damage using Wide-Field Optical Coherence Tomography
Grant Reference no. : MOH-000249-00

This research / project is supported by the Ministry of Education - Revisiting the structure/function relationship in glaucoma for improved patient care
Grant Reference no. : MOH-000647-00

This research / project is supported by the Ministry of Education - SAMURAI (Singapore Advanced Multi-subspecialty Unified Research and Innovation Centre in Ophthalmology)
Grant Reference no. : MOH-001001-00

This research / project is supported by the Ministry of Education - INTEgRating bRain, Eye And Cardiac Research (INTER-REACH): How the Scarecrow found a Brain, the Lion Vision to locate Courage and the Tin Man a Heart
Grant Reference no. : MOH-001015-00

This research / project is supported by the Ministry of Education - Prevention of and Biomarkers for Vascular Cognitive Impairment
Grant Reference no. : MOH-000500-00

This research / project is supported by the Ministry of Education - Validating Existing and Emerging Multi-modal Biomarkers of Vascular Cognitive Impairment
Grant Reference no. : MOH-000707-00
Description:
ISSN:
2168-6165
Files uploaded: