Real-world federated learning for brain imaging scientists

1

Federated learning (FL) has significant potential for neuroimaging research, particularly in predicting cognitive status in multiple sclerosis patients.
2

FLightcase, a new FL toolbox, was developed to facilitate real-world applications of federated learning in brain research.
3

Federated training outperformed centralized training in predicting age from brain MRI, achieving a mean absolute error of 6.08.
4

Deep transfer learning yielded better performance in predicting cognitive tests compared to shallow transfer learning in the study.
5

The study demonstrates the feasibility of real-world FL in neuroimaging, promoting access to large datasets without compromising data privacy.

Frontiers In Digital Health

by Stijn Denissen, Jorne Laton, Matthias Grothe, Manuela Vaneckova, Tomáš Uher, Matěj Kudrna, Dana Horáková, Johan Baijot, Iris-Katharina Penner, Michael Kirsch, Jiří Motýl, Maarten De Vos, Oliver Y. Chén, Jeroen Van Schependom, Diana Maria Sima, Guy Nagels
March 13, 2026

1