Federated Fine-tuning of SAM-Med3D for MRI-based Dementia Classification
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866911211690393600 |
|---|---|
| author | Mouheb, Kaouther Elbatel, Marawan Papma, Janne Biessels, Geert Jan Claassen, Jurgen Middelkoop, Huub van Munster, Barbara van der Flier, Wiesje Ramakers, Inez Klein, Stefan Bron, Esther E. |
| author_facet | Mouheb, Kaouther Elbatel, Marawan Papma, Janne Biessels, Geert Jan Claassen, Jurgen Middelkoop, Huub van Munster, Barbara van der Flier, Wiesje Ramakers, Inez Klein, Stefan Bron, Esther E. |
| contents | While foundation models (FMs) offer strong potential for AI-based dementia diagnosis, their integration into federated learning (FL) systems remains underexplored. In this benchmarking study, we systematically evaluate the impact of key design choices: classification head architecture, fine-tuning strategy, and aggregation method, on the performance and efficiency of federated FM tuning using brain MRI data. Using a large multi-cohort dataset, we find that the architecture of the classification head substantially influences performance, freezing the FM encoder achieves comparable results to full fine-tuning, and advanced aggregation methods outperform standard federated averaging. Our results offer practical insights for deploying FMs in decentralized clinical settings and highlight trade-offs that should guide future method development. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2508_21458 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Federated Fine-tuning of SAM-Med3D for MRI-based Dementia Classification Mouheb, Kaouther Elbatel, Marawan Papma, Janne Biessels, Geert Jan Claassen, Jurgen Middelkoop, Huub van Munster, Barbara van der Flier, Wiesje Ramakers, Inez Klein, Stefan Bron, Esther E. Computer Vision and Pattern Recognition While foundation models (FMs) offer strong potential for AI-based dementia diagnosis, their integration into federated learning (FL) systems remains underexplored. In this benchmarking study, we systematically evaluate the impact of key design choices: classification head architecture, fine-tuning strategy, and aggregation method, on the performance and efficiency of federated FM tuning using brain MRI data. Using a large multi-cohort dataset, we find that the architecture of the classification head substantially influences performance, freezing the FM encoder achieves comparable results to full fine-tuning, and advanced aggregation methods outperform standard federated averaging. Our results offer practical insights for deploying FMs in decentralized clinical settings and highlight trade-offs that should guide future method development. |
| title | Federated Fine-tuning of SAM-Med3D for MRI-based Dementia Classification |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2508.21458 |