Research map / Vision-language and multimodal / CLIP and zero-shot
fMRI-LM: Towards a Universal Foundation Model for Language-Aligned fMRI Understanding
Yuxiang Wei, Yanteng Zhang, Xi Xiao, Chengxuan Qian, Tianyang Wang, Vince D. Calhoun
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 |
|---|---|---|---|
| add imaging-based descriptors (contrastive loss and F2T objective) in pretraining (ablation) | fMRI-LM · sex classification | accuracy | Improved |
| add imaging-based descriptors in pretraining (ablation) | fMRI-LM · ADNI-AD | accuracy | Hurt |
| use LoRA instead of full fine-tuning (ablation) | fMRI-LM · HCP-Sex and ADHD200 | accuracy | Mixed |
| increase pretraining data (UKB/ABCD fraction) (ablation) | fMRI-LM · HCP sex and ADNI AD | accuracy | Improved |
| remove ABCD pretraining data vs remove UKB (ablation) | fMRI-LM · HCP sex and ADNI AD | accuracy | No clear effect |
| joint multi-dataset training for open-ended questions (ablation) | fMRI-LM · UKB/HCP-A/ADNI open-ended | accuracy | Mixed |
| multi-question instead of single-question paradigm | fMRI-LM · UKB/HCP-A/ADNI | accuracy | Mixed |
| use Qwen3-0.6B instead of GPT-2 backbone | fMRI-LM-B · ADNI-AD | accuracy: 77.92 -> 85.27 | Improved |
| use fMRI-LM-B(Q) instead of Brain-JEPA | fMRI-LM · ADNI-AD | AUC: 84.05 -> 81.02 | Hurt |
| use fMRI-LM-B(Q) instead of BrainLM | fMRI-LM · ABIDE2-ASD | accuracy: 65.22 -> 76.56 | Improved |
| few-shot (2 labeled samples) instead of zero-shot | fMRI-LM · new tasks/datasets | performance | Improved |
Closest papers by meaning
- MindLLM: A Subject-Agnostic and Versatile Model for fMRI-to-text Decoding (ICML 2025 · similarity 0.91)
- fMRI predictors based on language models of increasing complexity recover brain left lateralization (NeurIPS 2024 · similarity 0.90)
- Neuro-Vision to Language: Enhancing Brain Recording-based Visual Reconstruction and Language Interaction (NeurIPS 2024 · similarity 0.89)
- BrainLM: A foundation model for brain activity recordings (ICLR 2024 · similarity 0.89)
- Brain-OF: An Omnifunctional Foundation Model for fMRI, EEG and MEG (NeurIPS 2026 · similarity 0.89)
- Bridging the Semantic Latent Space between Brain and Machine: Similarity Is All You Need (AAAI 2024 · similarity 0.89)
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.
