Related Experiment Video
Updated: Jan 18, 2026
09:11
Automation of the Micronucleus Assay Using Imaging Flow Cytometry and Artificial Intelligence
Published on: January 27, 2023
2.7K
Artificial Intelligence-Based Automatic Mitosis Scoring in Breast Cancer Improves Inter-Observer Concordance and
Chien-Hui Wu1, Min-Hsiang Chang2, Hui-Juan Chen3
1Department of Pathology, Taiwan Adventist Hospital, Songshan District, Taipei City, Taiwan; Li Jen Pathology Clinic, Neihu Dist., Taipei City, Taiwan.
Clinical Breast Cancer
|May 24, 2025
Summary
An artificial intelligence system for automatic mitosis scoring in breast cancer images significantly reduces pathologist reading time and improves inter-observer agreement. This AI-driven approach enhances diagnostic efficiency and supports precision medicine in digital pathology.
Area of Science:
- Digital Pathology
- Computational Pathology
- Breast Cancer Diagnostics
Background:
- Manual assessment of mitotic activity in breast cancer is subjective and prone to inter-observer variability.
- Accurate mitotic counting is critical for treatment decisions and prognosis.
- Artificial intelligence (AI) offers a potential solution for objective and efficient mitosis scoring.
Purpose of the Study:
- To develop and evaluate an AI-based workflow for automated mitosis scoring in breast cancer whole-slide images.
- To assess the impact of the AI system on diagnostic efficiency and inter-pathologist concordance.
Main Methods:
- A mitosis detection model was trained using a dataset of 97 whole-slide images, guided by phosphohistone-H3 immunohistochemistry.
- An automated scoring framework incorporated image partitioning, epithelial segmentation, mitosis detection, hotspot analysis, and score classification.
- Clinical utility was validated by three pathologists assessing 20 slide images.
Main Results:
- The AI mitosis model achieved an F1-score of 0.75, comparable to state-of-the-art algorithms.
- The automated workflow improved inter-pathologist concordance by analyzing entire slides for hotspots.
- Reading time per case decreased significantly from 452s to 52s (P < .01), despite a slight underestimation of the mitosis score.
Conclusions:
- An integrated automatic mitosis scoring system enhances efficiency and concordance for pathology end-users.
- This AI-driven approach provides a foundation for precision medicine in breast cancer.
- Novel methodologies are crucial for assessing mitotic activity in the digital pathology era.
