Self-supervised fMRI foundation models have shown promising transfer performance, yet most rely on predefined region-level parcellations that discard fine-grained voxel information and introduce atlas-dependent biases. We propose Omni-fMRI, an atlas-free foundation model that operates directly on voxel-level signals. To enable scalable pretraining on 49,497 fMRI sessions across nine datasets, Omni-fMRI introduces a dynamic patching mechanism that substantially reduces computational cost while preserving informative spatial structure. To support reproducibility and fair comparison, we establish a comprehensive benchmark suite spanning 11 datasets and a diverse set of resting-state and task-based fMRI tasks. Experimental results demonstrate that Omni-fMRI consistently outperforms existing foundation models, providing a scalable and reproducible framework for atlas-free brain representation learning. Code and logs are available.
@article{arxiv.2601.23090,
title = {Omni-fMRI: A Universal Atlas-Free fMRI Foundation Model},
author = {Mo Wang and Wenhao Ye and Junfeng Xia and Junxiang Zhang and Xuanye Pan and Minghao Xu and Haotian Deng and Hongkai Wen and Quanying Liu},
journal= {arXiv preprint arXiv:2601.23090},
year = {2026}
}