Mitotic figure analysis improves with new task-specific AI training
MiTHras: Task-specific Hierarchical Semi-supervised Contrastive Masked Autoencoder for Mitotic Figure Analysis
Computer Vision and Pattern RecognitionMachine Learning
Summary
Detecting mitotic figures—cells in the process of division—is important for assessing tumor severity and patient prognosis. The authors created MiTHras, a special type of AI training that combines several techniques to better recognize these cells across different tissue types and imaging methods. They also built a large dataset of cell images from many cancer types to help train the model. Tests show MiTHras performs better than general AI models, particularly in classifying mitotic figures and their subtypes, making it a useful tool for automated tumor analysis.
What this means in practice
- •For pathology analysis teams: Improve automated tumor cell division detection and classification for more consistent grading across diverse cancer tissue images.
- •For medical imaging software developers: Integrate the MiTHras pretraining framework to enhance AI-based mitotic figure classification within pathology imaging tools.
Authors
Trinh T. L. Vuong, Simon Graham, Quoc Dang Vu, Phat T. H. Ho, Jeewoo Lim, Mostafa Jahanifar, Nasir Rajpoot, Jin T. Kwak
Abstract
Mitotic figure (MF) analysis supports tumor grading and prognostic assessment, but automated models remain sensitive to differences in tissue type and image acquisition. We present MiTHras, a task-specific pretraining framework that combines pseudo-label-guided image- and token-level contrastive learning with masked reconstruction. We construct TCGA-MF-Pseudo, a corpus of 1.8 million cell-centered images from 14 TCGA cohorts spanning 11 organ sites. Comprehensive evaluation on MF classification, detection, count-based survival prediction, and subtype classification demonstrates the efficacy of MiTHras. It achieves the highest mean F1 on all three MF classification benchmarks and both subtype benchmarks. MiTHras also outperforms general-purpose and pathology foundation encoders by a larger margin under frozen-encoder linear probing than under full fine-tuning. Although detection gains are modest due to a shared candidate-detection stage, ablations confirm that token-level supervision improves typical-versus-atypical classification and linear probing. These findings establish that MiTHras yields robust, transferable representations for automated mitotic activity assessment.