Cargando…

Synthetic whole-slide image tile generation with gene expression profile-infused deep generative models

In this work, we propose an approach to generate whole-slide image (WSI) tiles by using deep generative models infused with matched gene expression profiles. First, we train a variational autoencoder (VAE) that learns a latent, lower-dimensional representation of multi-tissue gene expression profile...

Descripción completa

Detalles Bibliográficos
Autores principales: Carrillo-Perez, Francisco, Pizurica, Marija, Ozawa, Michael G., Vogel, Hannes, West, Robert B., Kong, Christina S., Herrera, Luis Javier, Shen, Jeanne, Gevaert, Olivier
Formato: Online Artículo Texto
Lenguaje:English
Publicado: Elsevier 2023
Materias:
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10475789/
https://www.ncbi.nlm.nih.gov/pubmed/37671024
http://dx.doi.org/10.1016/j.crmeth.2023.100534
_version_ 1785100792222449664
author Carrillo-Perez, Francisco
Pizurica, Marija
Ozawa, Michael G.
Vogel, Hannes
West, Robert B.
Kong, Christina S.
Herrera, Luis Javier
Shen, Jeanne
Gevaert, Olivier
author_facet Carrillo-Perez, Francisco
Pizurica, Marija
Ozawa, Michael G.
Vogel, Hannes
West, Robert B.
Kong, Christina S.
Herrera, Luis Javier
Shen, Jeanne
Gevaert, Olivier
author_sort Carrillo-Perez, Francisco
collection PubMed
description In this work, we propose an approach to generate whole-slide image (WSI) tiles by using deep generative models infused with matched gene expression profiles. First, we train a variational autoencoder (VAE) that learns a latent, lower-dimensional representation of multi-tissue gene expression profiles. Then, we use this representation to infuse generative adversarial networks (GANs) that generate lung and brain cortex tissue tiles, resulting in a new model that we call RNA-GAN. Tiles generated by RNA-GAN were preferred by expert pathologists compared with tiles generated using traditional GANs, and in addition, RNA-GAN needs fewer training epochs to generate high-quality tiles. Finally, RNA-GAN was able to generalize to gene expression profiles outside of the training set, showing imputation capabilities. A web-based quiz is available for users to play a game distinguishing real and synthetic tiles: https://rna-gan.stanford.edu/, and the code for RNA-GAN is available here: https://github.com/gevaertlab/RNA-GAN.
format Online
Article
Text
id pubmed-10475789
institution National Center for Biotechnology Information
language English
publishDate 2023
publisher Elsevier
record_format MEDLINE/PubMed
spelling pubmed-104757892023-09-05 Synthetic whole-slide image tile generation with gene expression profile-infused deep generative models Carrillo-Perez, Francisco Pizurica, Marija Ozawa, Michael G. Vogel, Hannes West, Robert B. Kong, Christina S. Herrera, Luis Javier Shen, Jeanne Gevaert, Olivier Cell Rep Methods Article In this work, we propose an approach to generate whole-slide image (WSI) tiles by using deep generative models infused with matched gene expression profiles. First, we train a variational autoencoder (VAE) that learns a latent, lower-dimensional representation of multi-tissue gene expression profiles. Then, we use this representation to infuse generative adversarial networks (GANs) that generate lung and brain cortex tissue tiles, resulting in a new model that we call RNA-GAN. Tiles generated by RNA-GAN were preferred by expert pathologists compared with tiles generated using traditional GANs, and in addition, RNA-GAN needs fewer training epochs to generate high-quality tiles. Finally, RNA-GAN was able to generalize to gene expression profiles outside of the training set, showing imputation capabilities. A web-based quiz is available for users to play a game distinguishing real and synthetic tiles: https://rna-gan.stanford.edu/, and the code for RNA-GAN is available here: https://github.com/gevaertlab/RNA-GAN. Elsevier 2023-07-19 /pmc/articles/PMC10475789/ /pubmed/37671024 http://dx.doi.org/10.1016/j.crmeth.2023.100534 Text en © 2023 The Author(s) https://creativecommons.org/licenses/by-nc-nd/4.0/This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
spellingShingle Article
Carrillo-Perez, Francisco
Pizurica, Marija
Ozawa, Michael G.
Vogel, Hannes
West, Robert B.
Kong, Christina S.
Herrera, Luis Javier
Shen, Jeanne
Gevaert, Olivier
Synthetic whole-slide image tile generation with gene expression profile-infused deep generative models
title Synthetic whole-slide image tile generation with gene expression profile-infused deep generative models
title_full Synthetic whole-slide image tile generation with gene expression profile-infused deep generative models
title_fullStr Synthetic whole-slide image tile generation with gene expression profile-infused deep generative models
title_full_unstemmed Synthetic whole-slide image tile generation with gene expression profile-infused deep generative models
title_short Synthetic whole-slide image tile generation with gene expression profile-infused deep generative models
title_sort synthetic whole-slide image tile generation with gene expression profile-infused deep generative models
topic Article
url https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10475789/
https://www.ncbi.nlm.nih.gov/pubmed/37671024
http://dx.doi.org/10.1016/j.crmeth.2023.100534
work_keys_str_mv AT carrilloperezfrancisco syntheticwholeslideimagetilegenerationwithgeneexpressionprofileinfuseddeepgenerativemodels
AT pizuricamarija syntheticwholeslideimagetilegenerationwithgeneexpressionprofileinfuseddeepgenerativemodels
AT ozawamichaelg syntheticwholeslideimagetilegenerationwithgeneexpressionprofileinfuseddeepgenerativemodels
AT vogelhannes syntheticwholeslideimagetilegenerationwithgeneexpressionprofileinfuseddeepgenerativemodels
AT westrobertb syntheticwholeslideimagetilegenerationwithgeneexpressionprofileinfuseddeepgenerativemodels
AT kongchristinas syntheticwholeslideimagetilegenerationwithgeneexpressionprofileinfuseddeepgenerativemodels
AT herreraluisjavier syntheticwholeslideimagetilegenerationwithgeneexpressionprofileinfuseddeepgenerativemodels
AT shenjeanne syntheticwholeslideimagetilegenerationwithgeneexpressionprofileinfuseddeepgenerativemodels
AT gevaertolivier syntheticwholeslideimagetilegenerationwithgeneexpressionprofileinfuseddeepgenerativemodels