Related Experiment Video
Updated: Aug 9, 2026
08:40
Quantitation of Protein Expression and Co-localization Using Multiplexed Immuno-histochemical Staining and Multispectral Imaging
Published on: April 8, 2016
Classification of Histologic Images Using a Single Staining: Experiments with Deep Learning on Deconvolved Images
Vincenzo Della Mea1, David Pilutti1
1Dept. Of Mathematics, Computer Science and Physics, University of Udine, Italy.
Studies in Health Technology and Informatics
|June 24, 2020
Summary
Deep learning can identify tumor areas on immunohistochemistry slides using only the hematoxylin stain. This approach simplifies analysis by eliminating the need for separate Hematoxylin-Eosin stained slides.
Area of Science:
- Digital pathology
- Computational pathology
- Machine learning in histopathology
Background:
- Automated analysis of immunohistochemistry (IHC) slides is complex.
- Accurate analysis requires focusing on tumor regions.
- Tumor areas are typically identified on separate Hematoxylin-Eosin (H-E) stained slides.
Purpose of the Study:
- To assess deep learning performance on the hematoxylin component of H-E staining.
- To explore the potential for classifying tumor areas on IHC slides using only hematoxylin.
- To develop a more efficient IHC analysis workflow.
Main Methods:
- Utilized H-E color deconvolution to isolate single stain images.
- Applied deep learning methods to analyze the hematoxylin component.
- Conducted a preliminary experiment to evaluate classification accuracy.
Main Results:
- Single stain images were generated for hematoxylin and eosin.
- Preliminary experiments showed promising accuracy for hematoxylin (0.808) and eosin (0.812) components.
- These results indicate feasibility for deep learning-based tumor area recognition.
Conclusions:
- Deep learning models show potential for analyzing hematoxylin-only images.
- This method could enable tumor area identification directly on IHC slides.
- Simplifying the workflow may improve efficiency in digital pathology analysis.
