Cargando…
Structure preserving adversarial generation of labeled training samples for single-cell segmentation
We introduce a generative data augmentation strategy to improve the accuracy of instance segmentation of microscopy data for complex tissue structures. Our pipeline uses regular and conditional generative adversarial networks (GANs) for image-to-image translation to construct synthetic microscopy im...
Autores principales: | , , |
---|---|
Formato: | Online Artículo Texto |
Lenguaje: | English |
Publicado: |
Elsevier
2023
|
Materias: | |
Acceso en línea: | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10545934/ https://www.ncbi.nlm.nih.gov/pubmed/37725984 http://dx.doi.org/10.1016/j.crmeth.2023.100592 |
_version_ | 1785114770161008640 |
---|---|
author | Tasnadi, Ervin Sliz-Nagy, Alex Horvath, Peter |
author_facet | Tasnadi, Ervin Sliz-Nagy, Alex Horvath, Peter |
author_sort | Tasnadi, Ervin |
collection | PubMed |
description | We introduce a generative data augmentation strategy to improve the accuracy of instance segmentation of microscopy data for complex tissue structures. Our pipeline uses regular and conditional generative adversarial networks (GANs) for image-to-image translation to construct synthetic microscopy images along with their corresponding masks to simulate the distribution and shape of the objects and their appearance. The synthetic samples are then used for training an instance segmentation network (for example, StarDist or Cellpose). We show on two single-cell-resolution tissue datasets that our method improves the accuracy of downstream instance segmentation tasks compared with traditional training strategies using either the raw data or basic augmentations. We also compare the quality of the object masks with those generated by a traditional cell population simulation method, finding that our synthesized masks are closer to the ground truth considering Fréchet inception distances. |
format | Online Article Text |
id | pubmed-10545934 |
institution | National Center for Biotechnology Information |
language | English |
publishDate | 2023 |
publisher | Elsevier |
record_format | MEDLINE/PubMed |
spelling | pubmed-105459342023-10-04 Structure preserving adversarial generation of labeled training samples for single-cell segmentation Tasnadi, Ervin Sliz-Nagy, Alex Horvath, Peter Cell Rep Methods Article We introduce a generative data augmentation strategy to improve the accuracy of instance segmentation of microscopy data for complex tissue structures. Our pipeline uses regular and conditional generative adversarial networks (GANs) for image-to-image translation to construct synthetic microscopy images along with their corresponding masks to simulate the distribution and shape of the objects and their appearance. The synthetic samples are then used for training an instance segmentation network (for example, StarDist or Cellpose). We show on two single-cell-resolution tissue datasets that our method improves the accuracy of downstream instance segmentation tasks compared with traditional training strategies using either the raw data or basic augmentations. We also compare the quality of the object masks with those generated by a traditional cell population simulation method, finding that our synthesized masks are closer to the ground truth considering Fréchet inception distances. Elsevier 2023-09-18 /pmc/articles/PMC10545934/ /pubmed/37725984 http://dx.doi.org/10.1016/j.crmeth.2023.100592 Text en © 2023 The Authors https://creativecommons.org/licenses/by/4.0/This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). |
spellingShingle | Article Tasnadi, Ervin Sliz-Nagy, Alex Horvath, Peter Structure preserving adversarial generation of labeled training samples for single-cell segmentation |
title | Structure preserving adversarial generation of labeled training samples for single-cell segmentation |
title_full | Structure preserving adversarial generation of labeled training samples for single-cell segmentation |
title_fullStr | Structure preserving adversarial generation of labeled training samples for single-cell segmentation |
title_full_unstemmed | Structure preserving adversarial generation of labeled training samples for single-cell segmentation |
title_short | Structure preserving adversarial generation of labeled training samples for single-cell segmentation |
title_sort | structure preserving adversarial generation of labeled training samples for single-cell segmentation |
topic | Article |
url | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10545934/ https://www.ncbi.nlm.nih.gov/pubmed/37725984 http://dx.doi.org/10.1016/j.crmeth.2023.100592 |
work_keys_str_mv | AT tasnadiervin structurepreservingadversarialgenerationoflabeledtrainingsamplesforsinglecellsegmentation AT sliznagyalex structurepreservingadversarialgenerationoflabeledtrainingsamplesforsinglecellsegmentation AT horvathpeter structurepreservingadversarialgenerationoflabeledtrainingsamplesforsinglecellsegmentation |