Disentangled Autoencoder for Cross-Stain Feature Extraction in Pathology Image Analysis
A novel deep autoencoder architecture is proposed for the analysis of histopathology images. Its purpose is to produce a disentangled latent representation in which the structure and colour information are confined to different subspaces so that stain-independent models may be learned. For this, we...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2020-09-01
|
| Series: | Applied Sciences |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2076-3417/10/18/6427 |
| _version_ | 1797553642498162688 |
|---|---|
| author | Helge Hecht Mhd Hasan Sarhan Vlad Popovici |
| author_facet | Helge Hecht Mhd Hasan Sarhan Vlad Popovici |
| author_sort | Helge Hecht |
| collection | DOAJ |
| description | A novel deep autoencoder architecture is proposed for the analysis of histopathology images. Its purpose is to produce a disentangled latent representation in which the structure and colour information are confined to different subspaces so that stain-independent models may be learned. For this, we introduce two constraints on the representation which are implemented as a classifier and an adversarial discriminator. We show how they can be used for learning a latent representation across haematoxylin-eosin and a number of immune stains. Finally, we demonstrate the utility of the proposed representation in the context of matching image patches for registration applications and for learning a bag of visual words for whole slide image summarization. |
| first_indexed | 2024-03-10T16:18:29Z |
| format | Article |
| id | doaj.art-529bc7c2f48b458b94ea0ea55d9953a2 |
| institution | Directory Open Access Journal |
| issn | 2076-3417 |
| language | English |
| last_indexed | 2024-03-10T16:18:29Z |
| publishDate | 2020-09-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Applied Sciences |
| spelling | doaj.art-529bc7c2f48b458b94ea0ea55d9953a22023-11-20T13:49:47ZengMDPI AGApplied Sciences2076-34172020-09-011018642710.3390/app10186427Disentangled Autoencoder for Cross-Stain Feature Extraction in Pathology Image AnalysisHelge Hecht0Mhd Hasan Sarhan1Vlad Popovici2RECETOX, Masaryk University, 62500 Brno, Czech RepublicChair for Computer Aided Medical Procedures, Technical University of Munich, 80333 Munich, GermanyRECETOX, Masaryk University, 62500 Brno, Czech RepublicA novel deep autoencoder architecture is proposed for the analysis of histopathology images. Its purpose is to produce a disentangled latent representation in which the structure and colour information are confined to different subspaces so that stain-independent models may be learned. For this, we introduce two constraints on the representation which are implemented as a classifier and an adversarial discriminator. We show how they can be used for learning a latent representation across haematoxylin-eosin and a number of immune stains. Finally, we demonstrate the utility of the proposed representation in the context of matching image patches for registration applications and for learning a bag of visual words for whole slide image summarization.https://www.mdpi.com/2076-3417/10/18/6427digital pathologyimage registrationdeep learningdisentangled autoencoder |
| spellingShingle | Helge Hecht Mhd Hasan Sarhan Vlad Popovici Disentangled Autoencoder for Cross-Stain Feature Extraction in Pathology Image Analysis Applied Sciences digital pathology image registration deep learning disentangled autoencoder |
| title | Disentangled Autoencoder for Cross-Stain Feature Extraction in Pathology Image Analysis |
| title_full | Disentangled Autoencoder for Cross-Stain Feature Extraction in Pathology Image Analysis |
| title_fullStr | Disentangled Autoencoder for Cross-Stain Feature Extraction in Pathology Image Analysis |
| title_full_unstemmed | Disentangled Autoencoder for Cross-Stain Feature Extraction in Pathology Image Analysis |
| title_short | Disentangled Autoencoder for Cross-Stain Feature Extraction in Pathology Image Analysis |
| title_sort | disentangled autoencoder for cross stain feature extraction in pathology image analysis |
| topic | digital pathology image registration deep learning disentangled autoencoder |
| url | https://www.mdpi.com/2076-3417/10/18/6427 |
| work_keys_str_mv | AT helgehecht disentangledautoencoderforcrossstainfeatureextractioninpathologyimageanalysis AT mhdhasansarhan disentangledautoencoderforcrossstainfeatureextractioninpathologyimageanalysis AT vladpopovici disentangledautoencoderforcrossstainfeatureextractioninpathologyimageanalysis |