Scalable and Efficient Deep Learning-Based Pipeline for Mitotic Detection and Analysis in Pathology Images
1
Laboratory of Cancer Biology and Genetics, Center for Cancer Research, National Cancer Institute, National Institutes of Health, Bethesda, MD 20892, USA
2
Department of Radiation Oncology, Duke University Medical Center, Durham, NC 27710, USA
3
University of Hawaii Cancer Center, University of Hawaii at Manoa, Honolulu, HI 96813, USA
4
Molecular Imaging Branch, Center for Cancer Research, National Cancer Institute, National Institutes of Health, Bethesda, MD 20892, USA
*
Authors to whom correspondence should be addressed.
Cancers 2026, 18(11), 1807; https://doi.org/10.3390/cancers18111807
Submission received: 14 April 2026
/
Revised: 21 May 2026
/
Accepted: 27 May 2026
/
Published: 1 June 2026
(This article belongs to the Section Cancer Pathophysiology)
Simple Summary
Mitotic figures are important markers of tumor growth, but counting them manually in whole-slide pathology images is time-consuming and subject to observer variability. We developed an efficient artificial intelligence pipeline that detects mitotic figures, suppresses likely false-positive candidates, and further classifies mitotic figures as atypical or normal. The method was evaluated on public mitosis detection and atypical mitosis classification benchmarks and applied to breast cancer whole-slide images from the TCGA-BRCA cohort. Our results show that the pipeline achieves strong detection and classification performance while processing gigapixel slides within minutes on a single GPU. In an exploratory survival analysis of early-stage TCGA-BRCA cases, mitosis-derived features showed modest additional prognostic information beyond clinical and nuclei morphology features. These findings suggest that efficient automated mitosis analysis may support large-scale pathology studies and motivate further validation in independent clinical cohorts.
Abstract
Background: Accurate and efficient analysis of mitotic figures in whole-slide images (WSIs) is essential for tumor grading and prognosis. Methods: In this work, we present a three-stage pipeline for WSI-scale mitosis analysis that balances accuracy with clinical throughput: (1) a YOLOv11-based detector to propose mitosis candidates; (2) an ultra-lightweight classifier to refine detections and suppress false positives; and (3) a downstream classifier to distinguish atypical from normal mitoses for deeper biological insight. Results: In benchmark datasets, the two-stage detector improves F1 over detection-only baselines, while the atypical/normal module achieves strong accuracy, demonstrating cross-domain generalization. We further perform a proof-of-concept survival analysis on early-stage (I–II) cases from the TCGA-BRCA cohort, suggesting that mitosis-derived features may provide modest incremental prognostic information beyond the clinical baseline and nuclei features. Conclusions: Overall, the method delivers accurate detection, robust atypical mitosis classification, and high efficiency, processing gigapixel WSIs in minutes on a single GPU, positioning it for large-scale translational studies and future clinical workflow validation.
Keywords:
mitosis; atypical mitosis; detection; classification; pathology; deep learning; survival analysis; efficiency
