Deepfakes in Ophthalmology

Purpose: Generative adversarial networks (GANs) are deep learning (DL) models that can create and modify realistic-appearing synthetic images, or deepfakes, from real images. The purpose of our study was to evaluate the ability of experts to discern synthesized retinal fundus images from real fundus...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Jimmy S. Chen, MD, Aaron S. Coyner, PhD, R.V. Paul Chan, MD, M. Elizabeth Hartnett, MD, Darius M. Moshfeghi, MD, Leah A. Owen, MD, PhD, Jayashree Kalpathy-Cramer, PhD, Michael F. Chiang, MD, MA, J. Peter Campbell, MD, MPH
Formato:	article
Lenguaje:	EN
Publicado:	Elsevier 2021
Materias:	Deep learning Generative adversarial networks Ophthalmology Synthetic images RE1-994
Acceso en línea:	https://doaj.org/article/724242de3b5f4588ac9b5f2db684fce5
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:724242de3b5f4588ac9b5f2db684fce5
record_format	dspace
spelling	oai:doaj.org-article:724242de3b5f4588ac9b5f2db684fce52021-11-28T04:39:33ZDeepfakes in Ophthalmology2666-914510.1016/j.xops.2021.100079https://doaj.org/article/724242de3b5f4588ac9b5f2db684fce52021-12-01T00:00:00Zhttp://www.sciencedirect.com/science/article/pii/S2666914521000816https://doaj.org/toc/2666-9145Purpose: Generative adversarial networks (GANs) are deep learning (DL) models that can create and modify realistic-appearing synthetic images, or deepfakes, from real images. The purpose of our study was to evaluate the ability of experts to discern synthesized retinal fundus images from real fundus images and to review the current uses and limitations of GANs in ophthalmology. Design: Development and expert evaluation of a GAN and an informal review of the literature. Participants: A total of 4282 image pairs of fundus images and retinal vessel maps acquired from a multicenter ROP screening program. Methods: Pix2Pix HD, a high-resolution GAN, was first trained and validated on fundus and vessel map image pairs and subsequently used to generate 880 images from a held-out test set. Fifty synthetic images from this test set and 50 different real images were presented to 4 expert ROP ophthalmologists using a custom online system for evaluation of whether the images were real or synthetic. Literature was reviewed on PubMed and Google Scholars using combinations of the terms ophthalmology, GANs, generative adversarial networks, ophthalmology, images, deepfakes, and synthetic. Ancestor search was performed to broaden results. Main Outcome Measures: Expert ability to discern real versus synthetic images was evaluated using percent accuracy. Statistical significance was evaluated using a Fisher exact test, with P values ≤ 0.05 thresholded for significance. Results: The expert majority correctly identified 59% of images as being real or synthetic (P = 0.1). Experts 1 to 4 correctly identified 54%, 58%, 49%, and 61% of images (P = 0.505, 0.158, 1.000, and 0.043, respectively). These results suggest that the majority of experts could not discern between real and synthetic images. Additionally, we identified 20 implementations of GANs in the ophthalmology literature, with applications in a variety of imaging modalities and ophthalmic diseases. Conclusions: Generative adversarial networks can create synthetic fundus images that are indiscernible from real fundus images by expert ROP ophthalmologists. Synthetic images may improve dataset augmentation for DL, may be used in trainee education, and may have implications for patient privacy.Jimmy S. Chen, MDAaron S. Coyner, PhDR.V. Paul Chan, MDM. Elizabeth Hartnett, MDDarius M. Moshfeghi, MDLeah A. Owen, MD, PhDJayashree Kalpathy-Cramer, PhDMichael F. Chiang, MD, MAJ. Peter Campbell, MD, MPHElsevierarticleDeep learningGenerative adversarial networksOphthalmologySynthetic imagesOphthalmologyRE1-994ENOphthalmology Science, Vol 1, Iss 4, Pp 100079- (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	Deep learning Generative adversarial networks Ophthalmology Synthetic images Ophthalmology RE1-994
spellingShingle	Deep learning Generative adversarial networks Ophthalmology Synthetic images Ophthalmology RE1-994 Jimmy S. Chen, MD Aaron S. Coyner, PhD R.V. Paul Chan, MD M. Elizabeth Hartnett, MD Darius M. Moshfeghi, MD Leah A. Owen, MD, PhD Jayashree Kalpathy-Cramer, PhD Michael F. Chiang, MD, MA J. Peter Campbell, MD, MPH Deepfakes in Ophthalmology
description	Purpose: Generative adversarial networks (GANs) are deep learning (DL) models that can create and modify realistic-appearing synthetic images, or deepfakes, from real images. The purpose of our study was to evaluate the ability of experts to discern synthesized retinal fundus images from real fundus images and to review the current uses and limitations of GANs in ophthalmology. Design: Development and expert evaluation of a GAN and an informal review of the literature. Participants: A total of 4282 image pairs of fundus images and retinal vessel maps acquired from a multicenter ROP screening program. Methods: Pix2Pix HD, a high-resolution GAN, was first trained and validated on fundus and vessel map image pairs and subsequently used to generate 880 images from a held-out test set. Fifty synthetic images from this test set and 50 different real images were presented to 4 expert ROP ophthalmologists using a custom online system for evaluation of whether the images were real or synthetic. Literature was reviewed on PubMed and Google Scholars using combinations of the terms ophthalmology, GANs, generative adversarial networks, ophthalmology, images, deepfakes, and synthetic. Ancestor search was performed to broaden results. Main Outcome Measures: Expert ability to discern real versus synthetic images was evaluated using percent accuracy. Statistical significance was evaluated using a Fisher exact test, with P values ≤ 0.05 thresholded for significance. Results: The expert majority correctly identified 59% of images as being real or synthetic (P = 0.1). Experts 1 to 4 correctly identified 54%, 58%, 49%, and 61% of images (P = 0.505, 0.158, 1.000, and 0.043, respectively). These results suggest that the majority of experts could not discern between real and synthetic images. Additionally, we identified 20 implementations of GANs in the ophthalmology literature, with applications in a variety of imaging modalities and ophthalmic diseases. Conclusions: Generative adversarial networks can create synthetic fundus images that are indiscernible from real fundus images by expert ROP ophthalmologists. Synthetic images may improve dataset augmentation for DL, may be used in trainee education, and may have implications for patient privacy.
format	article
author	Jimmy S. Chen, MD Aaron S. Coyner, PhD R.V. Paul Chan, MD M. Elizabeth Hartnett, MD Darius M. Moshfeghi, MD Leah A. Owen, MD, PhD Jayashree Kalpathy-Cramer, PhD Michael F. Chiang, MD, MA J. Peter Campbell, MD, MPH
author_facet	Jimmy S. Chen, MD Aaron S. Coyner, PhD R.V. Paul Chan, MD M. Elizabeth Hartnett, MD Darius M. Moshfeghi, MD Leah A. Owen, MD, PhD Jayashree Kalpathy-Cramer, PhD Michael F. Chiang, MD, MA J. Peter Campbell, MD, MPH
author_sort	Jimmy S. Chen, MD
title	Deepfakes in Ophthalmology
title_short	Deepfakes in Ophthalmology
title_full	Deepfakes in Ophthalmology
title_fullStr	Deepfakes in Ophthalmology
title_full_unstemmed	Deepfakes in Ophthalmology
title_sort	deepfakes in ophthalmology
publisher	Elsevier
publishDate	2021
url	https://doaj.org/article/724242de3b5f4588ac9b5f2db684fce5
work_keys_str_mv	AT jimmyschenmd deepfakesinophthalmology AT aaronscoynerphd deepfakesinophthalmology AT rvpaulchanmd deepfakesinophthalmology AT melizabethhartnettmd deepfakesinophthalmology AT dariusmmoshfeghimd deepfakesinophthalmology AT leahaowenmdphd deepfakesinophthalmology AT jayashreekalpathycramerphd deepfakesinophthalmology AT michaelfchiangmdma deepfakesinophthalmology AT jpetercampbellmdmph deepfakesinophthalmology
_version_	1718408266566860800

Deepfakes in Ophthalmology

Ejemplares similares