Research map / Label-efficient and robust learning / Self-supervised and contrastive
Omni-fMRI: A Universal Atlas-Free fMRI Foundation Model
Mo Wang, Wenhao Ye, Junfeng Xia, Junxiang Zhang, Xuanye Pan, Minghao Xu, Haotian Deng, Hongkai Wen, Quanying Liu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use intensity variance instead of Shannon entropy for patch complexity (ablation) | Omni-fMRI · HCP | accuracy: 78.43 -> 85.18 | Improved |
| use intensity variance instead of Laplacian response for patch complexity (ablation) | Omni-fMRI · ADNI-AD | accuracy: 68.40 -> 75.69 | Improved |
| use intensity variance instead of MSE complexity (ablation) | Omni-fMRI · HCP | accuracy: 78.37 -> 85.18 | Improved |
| omit per-patch normalization in MAE pretraining (ablation) | Omni-fMRI · ADNI-AD | accuracy: 80.21 -> 84.72 | Improved |
| omit per-patch normalization in MAE pretraining (ablation) | Omni-fMRI · HCP | accuracy: 86.61 -> 87.05 | Improved |
| raise patch subdivision threshold from 0.25 to 0.3 (ablation) | Omni-fMRI · ABCD | accuracy: 67.29 -> 64.58 | Hurt |
| raise patch subdivision threshold from 0.25 to 0.3 (ablation) | Omni-fMRI · HCP | accuracy: 85.10 -> 85.10 | No clear effect |
| lower threshold from 0.25 to 0.1 (more tokens) (ablation) | Omni-fMRI · HCP | accuracy: 85.10 -> 85.77 | No clear effect |
| use no dynamic patching (threshold 0, 14K tokens) (ablation) | Omni-fMRI · ABCD/HCP | memory | Hurt |
| scale model size and pretraining data (ablation) | Omni-fMRI · ADNI (AD) and 3 other tasks | accuracy | Improved |
| replace NeuroSTORM with Omni-fMRI under linear probing | Omni-fMRI · ADNI (AD) | accuracy drop from fine-tuning: 3% versus 13% | Improved |
| use Omni-fMRI instead of NeuroSTORM | Omni-fMRI · NSD 100-way retrieval | Top-1 accuracy (%): 3.00 -> 6.93 | Improved |
Closest papers by meaning
- Brain-OF: An Omnifunctional Foundation Model for fMRI, EEG and MEG (NeurIPS 2026 · similarity 0.89)
- fMRI-LM: Towards a Universal Foundation Model for Language-Aligned fMRI Understanding (CVPR 2026 · similarity 0.88)
- FlatClip: Reusing Image Foundation Models for fMRI Representation Learning via Cortical Flatmaps (NeurIPS 2026 · similarity 0.87)
- MIRAGE: Adaptive Multimodal Gating for Whole-Brain fMRI Encoding (NeurIPS 2026 · similarity 0.87)
- MindEye2: Shared-Subject Models Enable fMRI-To-Image With 1 Hour of Data (ICML 2024 · similarity 0.86)
- BrainLM: A foundation model for brain activity recordings (ICLR 2024 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
