From 100,000+ images to winning the first brain MRI foundation model challenges: Sharing lessons and models
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866909995246813184 |
|---|---|
| author | Gordaliza, Pedro M. Banus, Jaume Gérin, Benoît Wynen, Maxence Molchanova, Nataliia Richiardi, Jonas Cuadra, Meritxell Bach |
| author_facet | Gordaliza, Pedro M. Banus, Jaume Gérin, Benoît Wynen, Maxence Molchanova, Nataliia Richiardi, Jonas Cuadra, Meritxell Bach |
| contents | Developing Foundation Models for medical image analysis is essential to overcome the unique challenges of radiological tasks. The first challenges of this kind for 3D brain MRI, SSL3D and FOMO25, were held at MICCAI 2025. Our solution ranked first in tracks of both contests. It relies on a U-Net CNN architecture combined with strategies leveraging anatomical priors and neuroimaging domain knowledge. Notably, our models trained 1-2 orders of magnitude faster and were 10 times smaller than competing transformer-based approaches. Models are available here: https://github.com/jbanusco/BrainFM4Challenges. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_13166 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | From 100,000+ images to winning the first brain MRI foundation model challenges: Sharing lessons and models Gordaliza, Pedro M. Banus, Jaume Gérin, Benoît Wynen, Maxence Molchanova, Nataliia Richiardi, Jonas Cuadra, Meritxell Bach Computer Vision and Pattern Recognition Artificial Intelligence Machine Learning Developing Foundation Models for medical image analysis is essential to overcome the unique challenges of radiological tasks. The first challenges of this kind for 3D brain MRI, SSL3D and FOMO25, were held at MICCAI 2025. Our solution ranked first in tracks of both contests. It relies on a U-Net CNN architecture combined with strategies leveraging anatomical priors and neuroimaging domain knowledge. Notably, our models trained 1-2 orders of magnitude faster and were 10 times smaller than competing transformer-based approaches. Models are available here: https://github.com/jbanusco/BrainFM4Challenges. |
| title | From 100,000+ images to winning the first brain MRI foundation model challenges: Sharing lessons and models |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2601.13166 |