Related Experiment Video
Updated: Aug 6, 2025
09:33
Automated Slide Scanning and Segmentation in Fluorescently-labeled Tissues Using a Widefield High-content Analysis System
Published on: May 3, 2018
8.2K
An automatic entropy method to efficiently mask histology whole-slide images
Yipei Song1,2, Francesco Cisternino3, Joost M Mekke4
1Department of Computer Engineering, University of Virginia, Charlottesville, VA, 22908, USA.
Scientific Reports
|March 16, 2023
Summary
EntropyMasker, a novel image entropy method, accurately segments porous tissues in whole-slide images (WSI). This approach improves digital pathology for disease diagnosis and research, outperforming existing segmentation techniques.
Area of Science:
- Digital pathology
- Computational imaging
- Biomedical image analysis
Background:
- Accurate tissue segmentation in whole-slide images (WSI) is crucial for disease diagnosis and research.
- Segmenting porous and heterogeneous tissues, like atherosclerotic plaques, presents significant challenges.
Purpose of the Study:
- To develop and evaluate EntropyMasker, a novel image entropy-based method for foreground-background segmentation in histology WSI.
- To address the limitations of existing methods in segmenting challenging tissue structures.
Main Methods:
- Developed EntropyMasker, a unique approach utilizing image entropy for WSI segmentation.
- Evaluated EntropyMasker on 97 high-resolution WSI of human carotid atherosclerotic plaques.
- Compared EntropyMasker against Otsu's method, Adaptive mean, Adaptive Gaussian, and slideMask using multiple benchmarking metrics.
Main Results:
- EntropyMasker demonstrated superior performance compared to four widely used segmentation methods.
- Achieved the highest sensitivity and Jaccard similarity index in segmentation accuracy.
- Successfully segmented porous and heterogeneous tissue structures in various staining types.
Conclusions:
- EntropyMasker offers a robust solution for WSI preprocessing and image analysis.
- The method has the potential to enhance machine learning pipelines in digital pathology.
- Enables advanced disease phenotyping beyond atherosclerosis research.
