Foundation Model-Driven Classification of Atypical Mitotic Figures with Domain-Aware Training Strategies
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866915559137869824 |
|---|---|
| author | Giedziun, Piotr Sołtysik, Jan Górczany, Mateusz Ropiak, Norbert Przymus, Marcin Krajewski, Piotr Kwiecień, Jarosław Bartczak, Artur Wasiak, Izabela Maniewski, Mateusz |
| author_facet | Giedziun, Piotr Sołtysik, Jan Górczany, Mateusz Ropiak, Norbert Przymus, Marcin Krajewski, Piotr Kwiecień, Jarosław Bartczak, Artur Wasiak, Izabela Maniewski, Mateusz |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_02601 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Foundation Model-Driven Classification of Atypical Mitotic Figures with Domain-Aware Training Strategies Giedziun, Piotr Sołtysik, Jan Górczany, Mateusz Ropiak, Norbert Przymus, Marcin Krajewski, Piotr Kwiecień, Jarosław Bartczak, Artur Wasiak, Izabela Maniewski, Mateusz Image and Video Processing Computer Vision and Pattern Recognition 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. |
| title | Foundation Model-Driven Classification of Atypical Mitotic Figures with Domain-Aware Training Strategies |
| topic | Image and Video Processing Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2509.02601 |