Research map / Image generation / Fast diffusion sampling
NeuroPictor: Refining fMRI-to-Image Reconstruction via Multi-individual Pretraining and Multi-level Modulation
Jingyang Huo, Yikai Wang, Yanwei Fu, Xuelin Qian, Chong Li, Yun Wang, Jianfeng Feng
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 multi-subject pretraining (8 subjects) before single-subject finetuning (ablation) | NeuroPictor · NSD Subject-1 | PixCorr: .265 -> .277 | Improved |
| add single-subject finetuning after multi-subject pretraining (ablation) | NeuroPictor · NSD Subject-1 | PixCorr: .167 -> .277 | Improved |
| pretrain fMRI encoder on UKB instead of random init (ablation) | NeuroPictor · NSD Subject-1 | CLIP: 86.0% -> 93.4% | Improved |
| fine-tune pretrained fMRI encoder rather than freeze it (ablation) | NeuroPictor · NSD Subject-1 | PixCorr: .220 -> .265 | Improved |
| use full high-dimensional encoder output instead of [CLS] token (ablation) | NeuroPictor · NSD Subject-1 | PixCorr: .139 -> .265 | Improved |
| add Low-Level Manipulation Network (ablation) | NeuroPictor · NSD Subject-1 | AlexNet(2): 75.4% -> 98.8% | Improved |
| add auxiliary semantic encoder (ablation) | NeuroPictor · NSD Subject-1 | SSIM: .276 -> .385 | Improved |
| add semantic loss L_sem (ablation) | NeuroPictor · NSD Subject-1 | CLIP: 93.0% -> 94.5% | Improved |
| compare NeuroPictor to prior methods on pixel correlation | NeuroPictor · NSD within-subject | PixCorr: .229 vs .353 (Brain-Cap) | Hurt |
| NeuroPictor vs MindEye on CLIP two-way accuracy | NeuroPictor · NSD within-subject | CLIP: 93.3% vs 94.1% | Hurt |
Later papers that compared with this method
1 later accepted paper reports NeuroPictor as a baseline in their results tables; 0 of them report a better number on most of the results they share with it.
- Cross-Subject Mind Decoding from Inaccurate Representations (ICCV 2025 · better on 0 of 1 shared result)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- MindEye2: Shared-Subject Models Enable fMRI-To-Image With 1 Hour of Data (ICML 2024 · similarity 0.88)
- Moving Beyond Diffusion: Hierarchy-to-Hierarchy Autoregression for fMRI-to-Image Reconstruction (ICLR 2026 · similarity 0.88)
- Brain-IT: Image Reconstruction from fMRI via Brain-Interaction Transformer (ICLR 2026 · similarity 0.88)
- NeuroClips: Towards High-fidelity and Smooth fMRI-to-Video Reconstruction (NeurIPS 2024 · similarity 0.86)
- See Through Their Minds: Learning Transferable Brain Decoding Models from Cross-Subject fMRI (AAAI 2025 · similarity 0.86)
- NSD-Imagery: A Benchmark Dataset for Extending fMRI Vision Decoding Methods to Mental Imagery (CVPR 2025 · 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.
