Related Experiment Video
Updated: Mar 15, 2026
06:17
Analysis of Multidimensional Microscopy Data Using Cell-ACDC
Published on: November 7, 2025
668
A Subphase-Labeled Mitotic Dataset for AI-powered Cell Division Analysis
Zsanett Zsofia Ivan1,2,3, Dominik Hirling4,5, Istvan Grexa4,5
1HUN-REN Biological Research Centre, Szeged, Hungary. ivan.zsanett@brc.hu.
Scientific Data
|March 14, 2026
Summary
Automating mitosis detection in digital pathology is crucial for accurate tumor grading. This study enhances mitosis detection by introducing subphase information and an atypical class, improving model performance and aiding biological research.
Area of Science:
- Digital pathology
- Computational biology
- Cancer research
Background:
- Mitosis detection is vital for tumor grading and prognosis.
- Manual mitosis counting is time-consuming and prone to variability.
- Advancements in automated mitosis detection are driven by challenges like MIDOG.
Purpose of the Study:
- To improve automated mitosis detection by incorporating subphase information.
- To enhance domain generalization of mitosis detection models.
- To facilitate molecular analysis of mitotic subphases.
Main Methods:
- Developed a new mitotic figure dataset with subphase annotations.
- Expanded the dataset with a previously unrepresented tumor domain.
- Utilized subtyping information and an atypical mitotic class for model enhancement.
Main Results:
- Demonstrated improved model performance through mitosis subtyping.
- Enhanced the diversity and applicability of mitosis detection models.
- Provided a foundation for further biological investigations into mitotic mechanisms.
Conclusions:
- Mitosis subtyping offers valuable insights for automated detection in digital pathology.
- This approach opens new avenues for molecular characterization of mitotic subphases.
- The enhanced dataset and methodology advance both clinical and research applications.
