Research map / Detection and segmentation / Open-set and few-shot detection
Recurrent Mask Refinement for Few-Shot Medical Image Segmentation
Hao Tang, Xingwei Liu, Shanlin Sun, Xiangyi Yan, Xiaohui Xie
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 CRE instead of plain mask concatenation (ablation) | Affine + Concat · ABD-110 CT | mean DSC: 58.66 -> 62.48 | Improved |
| add CRE to affine alignment (ablation) | Affine · ABD-110 CT | mean DSC: 55.62 -> 62.48 | Improved |
| add recurrent mask refinement to concat model (ablation) | Affine + Concat · ABD-110 CT | mean DSC: 58.66 -> 66.50 | Improved |
| add recurrent mask refinement to CRE model (ablation) | Affine + CRE · ABD-110 CT | mean DSC: 62.48 -> 81.91 | Improved |
| use CRE with recurrence instead of Grabcut refinement | Affine + Grabcut · ABD-110 CT | mean DSC: 62.91 -> 81.91 | Improved |
| use U-Net instead of VGG16 backbone (ablation) | RP-Net · ABD-110 CT | mean DSC: 67.48 -> 81.91 | Improved |
| use U-Net instead of Res18 backbone (ablation) | RP-Net · ABD-110 CT | mean DSC: 78.50 -> 81.91 | Improved |
| use correlation radius 5 instead of 0 (ablation) | RP-Net · ABD-110 CT | mean DSC: 81.58 -> 81.91 | No clear effect |
| use correlation radius 5 instead of 7 (ablation) | RP-Net · ABD-110 CT | mean DSC: 79.95 -> 81.91 | Improved |
| initialize with Demons instead of affine (ablation) | RP-Net · ABD-110 CT | mean DSC: 81.91 -> 82.93 | Improved |
| replace SSL-ALPNet with RP-Net | SSL-ALPNet · ABD-110 CT | mean DSC: 65.59 -> 81.91 | Improved |
| replace PANet with RP-Net | PANet · ABD-110 CT | mean DSC: 42.50 -> 81.91 | Improved |
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.
- PANet — PANet: Few-Shot Image Semantic Segmentation With Prototype Alignment (ICCV 2019) · better on 3 of 3 shared results
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
- CRNet: Cross-Reference Networks for Few-Shot Segmentation (CVPR 2020 · similarity 0.89)
- CRCNet: Few-Shot Segmentation with Cross-Reference and Region–Global Conditional Networks (IJCV 2022 · similarity 0.89)
- Bidirectional RNN-based Few Shot Learning for 3D Medical Image Segmentation (AAAI 2021 · similarity 0.88)
- Recurrent Saliency Transformation Network: Incorporating Multi-Stage Visual Cues for Small Organ Segmentation (CVPR 2018 · similarity 0.88)
- DeFSS: Image-to-Mask Denoising Learning for Few-shot Segmentation (ICCV 2025 · similarity 0.87)
- Unlocking the Potential of Pre-trained Vision Transformers for Few-Shot Semantic Segmentation through Relationship Descriptors (CVPR 2024 · 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.
