Omni-fMRI: A Universal Atlas-Free fMRI Foundation Model

TL;DR

Omni-fMRI introduces an atlas-free, voxel-level foundation model for fMRI that leverages dynamic patching for scalable pretraining and outperforms existing models across diverse datasets and tasks.

Contribution

It presents a novel atlas-free fMRI foundation model with a dynamic patching mechanism for scalable pretraining and a comprehensive benchmark suite for fair evaluation.

Findings

Outperforms existing models across multiple datasets

Supports both resting-state and task-based fMRI tasks

Provides a scalable, reproducible framework for brain representation learning

Abstract

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.