AI May Improve Breast Mitotic Scoring - Summary - MDSpire

AI May Improve Breast Mitotic Scoring

  • By

  • Andrea Surnit

  • May 13, 2026

  • 4 min

Share

Objective:

To evaluate the effectiveness of an AI tool in enhancing mitotic score accuracy and reproducibility among junior pathologists assessing breast carcinoma whole-slide images.

Key Findings:
  • Mitotic score accuracy improved from 62% to 76% and from 64% to 78% with AI assistance for the two junior pathologists.
  • Weighted Cohen’s kappa increased significantly, indicating improved agreement with expert consensus.
  • The AI tool enhanced consistency in hotspot selection, with the proportion of AI-suggested hotspots intersecting selected counting zones increasing from 46% to 80%.
Interpretation:

AI assistance notably improved mitotic score accuracy, particularly in diagnostically challenging cases, while maintaining pathologists' autonomy in scoring.

Limitations:
  • The study involved only two junior pathologists at a single institution.
  • Specimens were retrospectively collected and evaluated in a structured setting, not during active clinical workflow.
Conclusion:

The AI tool demonstrated promise in improving mitotic scoring accuracy and consistency, warranting further investigation in broader clinical settings.

Original Source(s)

Related Content