Normalization of HE-stained histological images using cycle consistent generative adversarial networks
Uloženo v:
| Název: | Normalization of HE-stained histological images using cycle consistent generative adversarial networks |
|---|---|
| Autoři: | Runz, Marlen, Rusche, Daniel, Schmidt, Stefan, Weihrauch, Martin R., Hesser, Jürgen, Weis, Cleo-Aron |
| Zdroj: | http://lobid.org/resources/99370677502106441#!, 16(1):71. |
| Rok vydání: | 2021 |
| Sbírka: | Publisso (ZB MED-Publikationsportal Lebenswissenschaften) |
| Témata: | Coloring Agents [MeSH], Female [MeSH], Humans [MeSH], Staining and Labeling/methods [MeSH], Digital pathology, Adenocarcinoma, Follicular/pathology [MeSH], Histology stain normalization, Thyroid Neoplasms/pathology [MeSH], Style transfer, Image Processing, Computer-Assisted/methods [MeSH], Unpaired image-to-image translation, Generative adversarial networks, Staining and Labeling/standards [MeSH], Reproducibility of Results [MeSH], Research, Eosine Yellowish-(YS) [MeSH], Hematoxylin [MeSH], Models, Statistical [MeSH], Deep learning, Color [MeSH], Breast Neoplasms/pathology [MeSH], HE-stain |
| Popis: | Background!#!Histological images show strong variance (e.g. illumination, color, staining quality) due to differences in image acquisition, tissue processing, staining, etc. This can impede downstream image analysis such as staining intensity evaluation or classification. Methods to reduce these variances are called image normalization techniques.!##!Methods!#!In this paper, we investigate the potential of CycleGAN (cycle consistent Generative Adversarial Network) for color normalization in hematoxylin-eosin stained histological images using daily clinical data with consideration of the variability of internal staining protocol variations. The network consists of a generator network G!##!Results!#!Qualitative results of the images generated by our network are compared to original color distributions. Our evaluation indicates that by mapping images to a target domain, the similarity training images from that domain improves up to 96%. We also achieve a high cycle consistency for the generator networks by obtaining similarity indices greater than 0.9. When applying the CycleGAN normalization to HE-stain images from our institute the kappa-value of the ResNet-model that is only trained on Camelyon16 data is increased more than 50%.!##!Conclusions!#!CycleGANs have proven to efficiently normalize HE-stained images. The approach compensates for deviations resulting from image acquisition (e.g. different scanning devices) as well as from tissue staining (e.g. different staining protocols), and thus overcomes the staining variations in images from various institutions.The code is publicly available at https://github.com/m4ln/stainTransfer_CycleGAN_pytorch . The data set supporting the solutions is available at https://doi.org/10.11588/data/8LKEZF . |
| Druh dokumentu: | article in journal/newspaper |
| Jazyk: | English |
| Relation: | https://repository.publisso.de/resource/frl:6465902; https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8349020/ |
| DOI: | 10.1186/s13000-021-01126-y |
| Dostupnost: | https://repository.publisso.de/resource/frl:6465902 https://doi.org/10.1186/s13000-021-01126-y https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8349020/ |
| Rights: | https://creativecommons.org/licenses/by/4.0/ |
| Přístupové číslo: | edsbas.E280BAD0 |
| Databáze: | BASE |
Nájsť tento článok vo Web of Science | Abstrakt: | Background!#!Histological images show strong variance (e.g. illumination, color, staining quality) due to differences in image acquisition, tissue processing, staining, etc. This can impede downstream image analysis such as staining intensity evaluation or classification. Methods to reduce these variances are called image normalization techniques.!##!Methods!#!In this paper, we investigate the potential of CycleGAN (cycle consistent Generative Adversarial Network) for color normalization in hematoxylin-eosin stained histological images using daily clinical data with consideration of the variability of internal staining protocol variations. The network consists of a generator network G!##!Results!#!Qualitative results of the images generated by our network are compared to original color distributions. Our evaluation indicates that by mapping images to a target domain, the similarity training images from that domain improves up to 96%. We also achieve a high cycle consistency for the generator networks by obtaining similarity indices greater than 0.9. When applying the CycleGAN normalization to HE-stain images from our institute the kappa-value of the ResNet-model that is only trained on Camelyon16 data is increased more than 50%.!##!Conclusions!#!CycleGANs have proven to efficiently normalize HE-stained images. The approach compensates for deviations resulting from image acquisition (e.g. different scanning devices) as well as from tissue staining (e.g. different staining protocols), and thus overcomes the staining variations in images from various institutions.The code is publicly available at https://github.com/m4ln/stainTransfer_CycleGAN_pytorch . The data set supporting the solutions is available at https://doi.org/10.11588/data/8LKEZF . |
|---|---|
| DOI: | 10.1186/s13000-021-01126-y |