Published July 3, 2022
| Version 1.0.0
Dataset
Open
Multi Stain Deep Learning
Creators
- Foersch, Sebastian1
- Glasner, Christina1
- Woerl, Ann-Christin2
- Eckstein, Markus3
- Wagner, Daniel-Christoph1
- Schulz, Stefan1
- Kellers, Franziska1
- Fernandez, Aurélie1
- Tserea, Konstantina1
- Kloth, Michael1
- Hartmann, Arndt3
- Heintz, Achim4
- Weichert, Wilko5
- Roth, Wilfried1
- Geppert, Carol3
- Kather, Jakob Nikolas6
- Jesinghaus, Moritz7
- 1. Institute of Pathology, University Medical Center Mainz, Mainz, Germany
- 2. Institute of Computer Science, Johannes Gutenberg University Mainz, Mainz, Germany
- 3. Institute of Pathology, University Clinic Erlangen, Erlangen, Germany
- 4. Department of General, Visceral and Vascular Surgery, Marien Hospital Mainz, Mainz, Germany
- 5. Institute of Pathology, Technical University Munich, Munich, Germany
- 6. Else Kroener Fresenius Center for Digital Health, Medical Faculty Carl Gustav Carus, Technical University Dresden, Dresden, Germany
- 7. Institute of Pathology, University Hospital Marburg, Marburg, Germany
Description
This repository contains surrogate data for our code repository on multistain deep learning. For more information, please refer to the link and associated publications.
All data provided is intended for research purposes only. We, the authors, assume no guarantee, liability or warranty. There is no claim to correctness.
- data.zip:
This archive contains histopathological image tiles of healthy colonic mucosa and colorectal carcinoma. The folders postfixed with "_noise" contain images of uniformly distributed random noise. The images in quarter_crop(_noise) have been generated by center cropping the images in half_crop(_noise) to half their original resolution. The original whole slides are available for research purposes from the TCGA COAD cohort. Please consider the original TCGA licenses and all other limitations found within the GDC data portal.
- saved.zip:
This archive contains checkpoint files of unimodal and multimodal models trained on a subset of the data in data.zip. Furthermore, the "evaluation" subfolders contain model evaluation results on validation data.