Foundation Model-Driven Classification of Atypical Mitotic Figures with Domain-Aware Training Strategies
This addresses a specific pathology image analysis problem, but appears incremental as it applies existing techniques to a new dataset.
The paper tackled binary classification of normal versus atypical mitotic figures in pathology images using a foundation model with domain-aware training strategies, achieving reasonable performance in preliminary evaluation.
We present a solution for the MIDOG 2025 Challenge Track~2, addressing binary classification of normal mitotic figures (NMFs) versus atypical mitotic figures (AMFs). The approach leverages pathology-specific foundation model H-optimus-0, selected based on recent cross-domain generalization benchmarks and our empirical testing, with Low-Rank Adaptation (LoRA) fine-tuning and MixUp augmentation. Implementation includes soft labels based on multi-expert consensus, hard negative mining, and adaptive focal loss, metric learning and domain adaptation. The method demonstrates both the promise and challenges of applying foundation models to this complex classification task, achieving reasonable performance in the preliminary evaluation phase.