StainCUT: Stain Normalization with Contrastive Learning
In recent years, numerous deep-learning approaches have been developed for the analysis of histopathology Whole Slide Images (WSI). A recurrent issue is the lack of generalization ability of a model that has been trained with images of one laboratory and then used to analyze images of a different la...
Main Authors: | , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2022-07-01
|
Series: | Journal of Imaging |
Subjects: | |
Online Access: | https://www.mdpi.com/2313-433X/8/7/202 |
_version_ | 1797433453220724736 |
---|---|
author | José Carlos Gutiérrez Pérez Daniel Otero Baguer Peter Maass |
author_facet | José Carlos Gutiérrez Pérez Daniel Otero Baguer Peter Maass |
author_sort | José Carlos Gutiérrez Pérez |
collection | DOAJ |
description | In recent years, numerous deep-learning approaches have been developed for the analysis of histopathology Whole Slide Images (WSI). A recurrent issue is the lack of generalization ability of a model that has been trained with images of one laboratory and then used to analyze images of a different laboratory. This occurs mainly due to the use of different scanners, laboratory procedures, and staining variations. This can produce strong color differences, which change not only the characteristics of the image, such as the contrast, brightness, and saturation, but also create more complex style variations. In this paper, we present a deep-learning solution based on contrastive learning to transfer from one staining style to another: StainCUT. This method eliminates the need to choose a reference frame and does not need paired images with different staining to learn the mapping between the stain distributions. Additionally, it does not rely on the CycleGAN approach, which makes the method efficient in terms of memory consumption and running time. We evaluate the model using two datasets that consist of the same specimens digitized with two different scanners. We also apply it as a preprocessing step for the semantic segmentation of metastases in lymph nodes. The model was trained on data from one of the laboratories and evaluated on data from another. The results validate the hypothesis that stain normalization indeed improves the performance of the model. Finally, we also investigate and compare the application of the stain normalization step during the training of the model and at inference. |
first_indexed | 2024-03-09T10:16:17Z |
format | Article |
id | doaj.art-54b8e27fa5e9433cac9bc5df14b3531f |
institution | Directory Open Access Journal |
issn | 2313-433X |
language | English |
last_indexed | 2024-03-09T10:16:17Z |
publishDate | 2022-07-01 |
publisher | MDPI AG |
record_format | Article |
series | Journal of Imaging |
spelling | doaj.art-54b8e27fa5e9433cac9bc5df14b3531f2023-12-01T22:19:00ZengMDPI AGJournal of Imaging2313-433X2022-07-018720210.3390/jimaging8070202StainCUT: Stain Normalization with Contrastive LearningJosé Carlos Gutiérrez Pérez0Daniel Otero Baguer1Peter Maass2Center for Industrial Mathematics, University of Bremen, 28359 Bremen, GermanyCenter for Industrial Mathematics, University of Bremen, 28359 Bremen, GermanyCenter for Industrial Mathematics, University of Bremen, 28359 Bremen, GermanyIn recent years, numerous deep-learning approaches have been developed for the analysis of histopathology Whole Slide Images (WSI). A recurrent issue is the lack of generalization ability of a model that has been trained with images of one laboratory and then used to analyze images of a different laboratory. This occurs mainly due to the use of different scanners, laboratory procedures, and staining variations. This can produce strong color differences, which change not only the characteristics of the image, such as the contrast, brightness, and saturation, but also create more complex style variations. In this paper, we present a deep-learning solution based on contrastive learning to transfer from one staining style to another: StainCUT. This method eliminates the need to choose a reference frame and does not need paired images with different staining to learn the mapping between the stain distributions. Additionally, it does not rely on the CycleGAN approach, which makes the method efficient in terms of memory consumption and running time. We evaluate the model using two datasets that consist of the same specimens digitized with two different scanners. We also apply it as a preprocessing step for the semantic segmentation of metastases in lymph nodes. The model was trained on data from one of the laboratories and evaluated on data from another. The results validate the hypothesis that stain normalization indeed improves the performance of the model. Finally, we also investigate and compare the application of the stain normalization step during the training of the model and at inference.https://www.mdpi.com/2313-433X/8/7/202stain normalizationgenerative adversarial networkcontrastive learningdigital pathology |
spellingShingle | José Carlos Gutiérrez Pérez Daniel Otero Baguer Peter Maass StainCUT: Stain Normalization with Contrastive Learning Journal of Imaging stain normalization generative adversarial network contrastive learning digital pathology |
title | StainCUT: Stain Normalization with Contrastive Learning |
title_full | StainCUT: Stain Normalization with Contrastive Learning |
title_fullStr | StainCUT: Stain Normalization with Contrastive Learning |
title_full_unstemmed | StainCUT: Stain Normalization with Contrastive Learning |
title_short | StainCUT: Stain Normalization with Contrastive Learning |
title_sort | staincut stain normalization with contrastive learning |
topic | stain normalization generative adversarial network contrastive learning digital pathology |
url | https://www.mdpi.com/2313-433X/8/7/202 |
work_keys_str_mv | AT josecarlosgutierrezperez staincutstainnormalizationwithcontrastivelearning AT danieloterobaguer staincutstainnormalizationwithcontrastivelearning AT petermaass staincutstainnormalizationwithcontrastivelearning |