Parameter-efficient fine-tuning (PEFT) of Vision Foundation Models for Atypical Mitotic Figure Classification
This work addresses the problem of detecting rare abnormal cell divisions for pathologists, but it is incremental as it applies existing parameter-efficient fine-tuning methods to a new medical imaging dataset.
The study tackled the classification of atypical mitotic figures, a challenging task due to subtle cues and class imbalance, by fine-tuning vision foundation models with LoRA, achieving a balanced accuracy of 88.37% on a test set and ranking 9th in a challenge.
Atypical mitotic figures (AMFs) are rare abnormal cell divisions associated with tumor aggressiveness and poor prognosis. Their detection remains a significant challenge due to subtle morphological cues, class imbalance, and inter-observer variability among pathologists. The MIDOG 2025 challenge introduced a dedicated track for atypical mitosis classification, enabling systematic evaluation of deep learning methods. In this study, we investigated the use of large vision foundation models, including Virchow, Virchow2, and UNI, with Low-Rank Adaptation (LoRA) for parameter-efficient fine-tuning. We conducted extensive experiments with different LoRA ranks, as well as random and group-based data splits, to analyze robustness under varied conditions. Our best approach, Virchow with LoRA rank 8 and ensemble of three-fold cross-validation, achieved a balanced accuracy of 88.37% on the preliminary test set, ranking joint 9th in the challenge leaderboard. These results highlight the promise of foundation models with efficient adaptation strategies for the classification of atypical mitosis, while underscoring the need for improvements in specificity and domain generalization.