| name | brain-dit-fmri-foundation-model-v6 |
| description | Brain-DiT v6 universal multi-state fMRI foundation model with metadata-conditioned pretraining across 24 datasets covering resting, task, naturalistic, disease, and sleep states. Use when working with fMRI foundation models, brain state decoding, or multi-state neuroimaging. |
| version | 1.0.0 |
| metadata | {"hermes":{"tags":["fmri","foundation-model","diffusion-transformer","brain-decoding","multi-state","neuroimaging"],"source_paper":"Brain-DiT: A Universal Multi-state fMRI Foundation Model with Metadata-Conditioned Pretraining (arXiv:2604.12683)"}} |
Brain-DiT: A Universal Multi-state fMRI Foundation Model with Metadata-Conditioned Pretraining
Source
- arXiv: 2604.12683
- Authors: Junfeng Xia, Wenhao Ye, Xuanye Pan
- Published: 2026-04-14
- Categories: cs.CV, q-bio.NC
Abstract
Current fMRI foundation models primarily rely on a limited range of brain states and mismatched pretraining tasks, restricting their ability to learn generalized representations across diverse brain states. We present \textit{Brain-DiT}, a universal multi-state fMRI foundation model pretrained on 349,898 sessions from 24 datasets spanning resting, task, naturalistic, disease, and sleep states. Unlike prior fMRI foundation models that rely on masked reconstruction in the raw-signal space or a lat
Key Concepts
- Identify core methodology from the paper
- Understand the mathematical framework
- Note the evaluation protocol and datasets used
Implementation Notes
- Refer to the original paper for detailed equations and architecture
- Check for any available code repositories linked in the paper
- Consider computational requirements for reproduction
Activation Keywords
- fmri, foundation-model, diffusion-transformer, brain-decoding
