Related Experiment Video
Updated: May 17, 2025
09:31
High-Speed Ultraviolet Photoacoustic Microscopy for Histological Imaging with Virtual-Staining assisted by Deep Learning
Published on: April 28, 2022
2.9K
Stain Normalization of Histopathological Images Based on Deep Learning: A Review
Chuanyun Xu1, Yisha Sun1, Yang Zhang1
1School of Computer & Information Science, Chongqing Normal University, Chongqing 401331, China.
Diagnostics (Basel, Switzerland)
|May 1, 2025
Summary
Deep learning stain normalization methods standardize H&E stained histopathology images, overcoming color variations for improved cancer diagnosis algorithms. This review covers recent advancements and future directions in the field.
Area of Science:
- Digital Pathology
- Computational Biology
- Medical Imaging
Background:
- Hematoxylin and eosin (H&E) staining is vital for cancer diagnosis.
- Color variations in H&E images hinder downstream algorithmic performance.
- Stain normalization is essential for standardizing histopathological image data.
Purpose of the Study:
- To review the latest advancements in deep learning-based stain normalization.
- To provide a comprehensive overview of evaluation metrics and datasets.
- To analyze supervised, unsupervised, and self-supervised deep learning approaches.
Main Methods:
- Systematic review of 115 publications on deep learning stain normalization.
- Categorization of methods by core technologies.
- Analysis of contributions and limitations of reviewed approaches.
Main Results:
- Deep learning methods offer robust and template-independent stain normalization.
- Various supervised, unsupervised, and self-supervised techniques show promise.
- Established evaluation metrics and datasets facilitate method comparison.
Conclusions:
- Deep learning stain normalization is crucial for reliable cancer diagnosis AI.
- Further research is needed to address current challenges and advance the field.
- This review aids researchers in understanding and developing intelligent cancer diagnosis tools.
