Research map / Image generation / Fast diffusion sampling
Diffusion MRI Transformer with a Diffusion Space Rotary Positional Embedding (D-RoPE)
Gustavo Chau Loo Kung, Mohammad Abbasi, Camila Blank, Juze Zhang, Alan Q. Wang, Sophie Ostmeier, Akshay Chaudhari, Kilian Pohl, Ehsan Adeli
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 D-RoPE (ablation) | MAE with spatial masking · HCP-YA recon (b=2000) | FID: 13.49 -> 9.61 | Improved |
| add D-RoPE (ablation) | MAE with diffusion masking · HCP-YA recon (b=2000) | PSNR: 10.07 -> 15.31 | Improved |
| add D-RoPE (ablation) | MAE with alternating masking · HCP-YA recon (b=2000) | FID: 10.81 -> 8.05 | Improved |
| use alternating instead of spatial masking (with D-RoPE) (ablation) | MAE w/ D-RoPE · HCP-YA recon (b=2000) | FID: 9.61 -> 8.05 | Improved |
| use alternating instead of spatial masking (no D-RoPE) (ablation) | MAE · HCP-YA recon (b=2000) | FID: 13.49 -> 10.81 | Improved |
| use diffusion-only masking without D-RoPE (ablation) | MAE · HCP-YA recon (b=2000) | PSNR: 14.91 -> 10.07 | Hurt |
| add D-RoPE (alternating masking) (ablation) | MAE · HCP-YA FA error | FA error: 0.080 -> 0.055 | Improved |
| add D-RoPE (alternating masking) (ablation) | MAE · HCP-YA MD error | MD error x10^3: 0.109 -> 0.103 | No clear effect |
| pretrained frozen latent with MLP head vs ViT w/RoPE on raw dMRI | Ours · CN vs MCI classification (ADNI) | Balanced accuracy (%): 47.40 -> 64.70 | Improved |
| pretrained frozen latent with MLP head vs ViT w/RoPE on raw dMRI | Ours · ADAS-Cog prediction (ADNI) | Correlation rho: -0.05 -> 0.38 | Improved |
| linear vs MLP head on frozen latent (ablation) | Ours · Sex classification (HCP-A) | Accuracy (%): 74.60 vs 59.30 | Mixed |
| finetune last layer vs frozen latent MLP head (ablation) | Ours · Age prediction | Correlation rho: 0.90 -> 0.94 | Improved |
Closest papers by meaning
- A Steerable Deep Network for Model-Free Diffusion MRI Registration (NeurIPS 2026 · similarity 0.87)
- Self-Supervised Diffusion MRI Denoising via Iterative and Stable Refinement (ICLR 2025 · similarity 0.86)
- Equivariant spatio-hemispherical networks for diffusion MRI deconvolution (NeurIPS 2024 · similarity 0.85)
- Patch2Self2: Self-supervised Denoising on Coresets via Matrix Sketching (CVPR 2024 · similarity 0.85)
- Rethinking Diffusion Model for Multi-Contrast MRI Super-Resolution (CVPR 2024 · similarity 0.84)
- DDMI: Domain-agnostic Latent Diffusion Models for Synthesizing High-Quality Implicit Neural Representations (ICLR 2024 · similarity 0.84)
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.
