From 100,000+ images to winning the first brain MRI foundation model challenges: Sharing lessons and models

Gordaliza, Pedro;Banus, Jaume;Gérin, Benoît;Wynen, Maxence;Cuadra, Meritxell;et.al.
(2026) , 4 pages

Files

2601.13166v1.pdf
  • Open Access
  • Adobe PDF
  • 781.71 KB
Download

Details

Authors
  • Gordaliza, PedroDepartment of Pharmaceutical Chemistry, Pharmacy Faculty, Salamanca University, Campus Miguel de Unamuno, 37007 Salamanca, Spain
    Author
  • Banus, Jaume
    Author
  • Gérin, Benoît
    Author
  • Wynen, Maxence
    Author
  • Cuadra, Meritxell
    Author
Show more
Abstract
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× smaller than competing transformer-based approaches.
Affiliations

Citations

  • APA
  • Chicago
  • FWB

Gordaliza, P., Banus, J., Gérin, B., Wynen, M., Molchanova, N., Richiardi, J., & Cuadra, M. (2026). From 100,000+ images to winning the first brain MRI foundation model challenges: Sharing lessons and models. https://hdl.handle.net/2078.5/281607