Research map / Image generation / Fast diffusion sampling
VGD: Value-Guided Diffusion Toward High-Utility Medical Image Segmentation
Hongyu Zhang, Haipeng Chen, Chengxin Yang, Yingda Lyu
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 value-guided sampling to SiameseDiff | SiameseDiff · ACDC (nnU-Net) | DSC: 88.9 -> 91.4 | Improved |
| add value-guided sampling to SegDiff | SegDiff · Polyps (nnU-Net) | DSC: 78.6 -> 78.8 | Improved |
| add VGD to FairDiff on Polyps (nnU-Net) | FairDiff · Polyps (nnU-Net) | DSC: 80.9 -> 81.5 | No clear effect |
| use VGD instead of random sampling | SD1.5+ControlNet · ACDC (nnU-Net) | DSC: 89.0 -> 90.5 | Improved |
| use VGD instead of unguided TFG sampling | SD1.5+ControlNet · ACDC (nnU-Net) | DSC: 78.5 -> 90.5 | Improved |
| add VGD at the cost of fidelity (FID) vs random sampling | SD1.5+ControlNet · fidelity (avg) | FID: 119.8 -> 122.3 | Hurt |
| add cross-entropy term to value score (ablation) | VGD · ACDC (nnU-Net) | DSC: 89.8 -> 90.5 | Improved |
| add margin-based uncertainty term to value score (ablation) | VGD · ACDC (nnU-Net) | DSC: 89.4 -> 90.5 | Improved |
| add value focus mask (ablation) | VGD · ACDC (nnU-Net) | DSC: 88.6 -> 90.5 | Improved |
| add time-travel resampling (ablation) | VGD · ACDC (nnU-Net) | DSC: 89.9 -> 90.5 | Improved |
| add fidelity constraint (trust region) (ablation) | VGD · ACDC | sampling stability | Improved |
| raise trust coefficient k to 0.06 or more (ablation) | VGD · ACDC (nnU-Net) | DSC | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- SegDiff — SegDiff: Segmented Trajectory Diffusion for Consistent and Adaptive Robot Manipulation (ECCV 2026) · better on 1 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
- Noise-Consistent Siamese-Diffusion for Medical Image Synthesis and Segmentation (CVPR 2025 · similarity 0.89)
- MedSegDiff-V2: Diffusion-Based Medical Image Segmentation with Transformer (AAAI 2024 · similarity 0.87)
- MedFlowSeg: Flow Matching for Medical Image Segmentation with Frequency-Aware Attention (NeurIPS 2026 · similarity 0.86)
- Factorized Diffusion Architectures for Unsupervised Image Generation and Segmentation (NeurIPS 2024 · similarity 0.86)
- LangDAug: Langevin Data Augmentation for Multi-Source Domain Generalization in Medical Image Segmentation (ICML 2025 · similarity 0.86)
- L-Diffusion: Laplace Diffusion for Efficient Pathology Image Segmentation (ICML 2025 · similarity 0.86)
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.
