Research map / Image generation / Image restoration
Plug-and-Play Diffusion Meets ADMM: Dual-Variable Coupling for Robust Medical Image Reconstruction
Chenhe Du, Xuanyu Tian, Qing Wu, Muyu Liu, Jingyi Yu, Hongjiang Wei, Yuyao Zhang
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 dual variable coupling (ADMM) to PnP-HQS (ablation) | DC-PnPDP · LACT-90 component ablation | PSNR (dB): 31.36 -> 35.91 | Improved |
| add spectral homogenization without dual variable (ablation) | PnP-HQS (DiffPIR) · LACT-90 component ablation | PSNR (dB): 31.36 -> 31.51 | Improved |
| add spectral homogenization on top of dual coupling (ablation) | DC-PnPDP · LACT-90 component ablation | PSNR (dB): 35.91 -> 37.02 | Improved |
| add dual coupling and spectral homogenization (ablation) | PnP-HQS (DiffPIR) · LACT-90 component ablation | LPIPS: 0.023 -> 0.011 | Improved |
| replace DiffPIR with DC-PnPDP (100 NFE) | DiffPIR · LACT-90 | PSNR (dB): 34.70 -> 39.46 | Improved |
| replace DiffPIR with DC-PnPDP (100 NFE) | DiffPIR · SVCT-20 | PSNR (dB): 37.86 -> 40.55 | Improved |
| replace SITCOM with DC-PnPDP (100 NFE) | SITCOM · fastMRI brain (AF=6) | PSNR (dB): 35.58 -> 36.43 | Improved |
| replace SITCOM with DC-PnPDP (100 NFE) | SITCOM · fastMRI brain (AF=10) | PSNR (dB): 28.67 -> 30.91 | Improved |
| reduce DC-PnPDP budget from 100 to 50 NFE | DC-PnPDP · LACT-90 | PSNR (dB): 39.46 -> 37.53 | Hurt |
| use DC-PnPDP at matched 50 NFE instead of DiffPIR | DiffPIR · LACT-90 (50 NFE, appendix) | PSNR (dB): 31.03 -> 36.98 | Improved |
| use DC-PnPDP at matched 50 NFE instead of DiffPIR | DiffPIR · fastMRI knee AF=10 (50 NFE, appendix) | PSNR (dB): 25.32 -> 26.21 | Improved |
| add dual update and SH step to DiffPIR | DiffPIR · SVCT-20 runtime | time per iteration (s): 0.1951 -> 0.1958 | No clear effect |
Closest papers by meaning
- Provably Robust Score-Based Diffusion Posterior Sampling for Plug-and-Play Image Reconstruction (NeurIPS 2024 · similarity 0.92)
- Improving Diffusion Posterior Samplers with Lagged Temporal Corrections for Image Restoration (NeurIPS 2026 · similarity 0.90)
- InverseBench: Benchmarking Plug-and-Play Diffusion Priors for Inverse Problems in Physical Sciences (ICLR 2025 · similarity 0.89)
- Prior Mismatch and Adaptation in PnP-ADMM with a Nonconvex Convergence Analysis (ICML 2024 · similarity 0.89)
- DEFT: Efficient Fine-tuning of Diffusion Models by Learning the Generalised $h$-transform (NeurIPS 2024 · similarity 0.89)
- A Variational Perspective on Solving Inverse Problems with Diffusion Models (ICLR 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.
