Related Experiment Video
Updated: Jul 28, 2025
09:31
Automated Quantification of Hematopoietic Cell – Stromal Cell Interactions in Histological Images of Undecalcified Bone
Published on: April 8, 2015
11.7K
CW-NET for multitype cell detection and classification in bone marrow examination and mitotic figure examination
Ching-Wei Wang1,2, Sheng-Chuan Huang3,4,5, Muhammad-Adil Khalil2
1Graduate Institute of Biomedical Engineering, National Taiwan University of Science and Technology, Taipei City, 106335, Taiwan.
Bioinformatics (Oxford, England)
|May 30, 2023
Summary
This study introduces CW-Net, an automated system for bone marrow and mitotic figure examination from whole-slide images. CW-Net overcomes challenges in manual analysis, offering robust and generalizable performance for hematologic disorder diagnosis and cancer assessment.
Area of Science:
- Digital pathology
- Computational hematology
- Artificial intelligence in medicine
Background:
- Bone marrow (BM) examination is crucial for diagnosing hematologic disorders, typically requiring manual microscopy.
- Mitotic figure detection is vital for cancer diagnosis, grading, and predicting treatment outcomes.
- Current manual methods face challenges including cell diversity, overlapping cells, stain variations, and laborious, variable annotations.
Purpose of the Study:
- To develop an efficient and fully automated approach for bone marrow and mitotic figure examination using whole-slide images (WSIs).
- To address the limitations of manual microscopic analysis and sparse data annotation in AI model training.
- To demonstrate the performance and generalizability of the proposed method on large-scale datasets.
Main Methods:
- Development of a novel, efficient, and fully automatic Convolutional Wavelet Network (CW-Net).
- Application of CW-Net to whole-slide images for both bone marrow cell type classification and mitotic figure detection.
- Validation on a large BM WSI dataset (16,456 cells, 19 types) and a mitotic figure WSI dataset (262,481 cells, 5 types).
Main Results:
- CW-Net demonstrated superior performance in automated bone marrow examination.
- The approach achieved high accuracy in mitotic figure detection and assessment.
- The method showed robustness and generalizability across large and diverse WSI datasets.
Conclusions:
- The proposed CW-Net offers an efficient and automated solution for complex digital pathology tasks.
- Automated analysis of bone marrow and mitotic figures can overcome the limitations of manual methods.
- The developed system has significant potential to improve diagnostic accuracy and efficiency in hematology and oncology.
