Research map / Image generation / Image restoration
Preserving Tumor Volumes for Unsupervised Medical Image Registration
Qihua Dong, Hao Du, Ying Song, Yan Xu, Jing Liao
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 adaptive volume-preserving loss to VTN | VTN · LiTS17 | STSR: 2.330 -> 1.260 | Improved |
| add adaptive volume-preserving loss to VoxelMorph | VoxelMorph · LiTS17 | STSR: 1.738 -> 1.241 | Improved |
| add adaptive volume-preserving loss to LapIRN | LapIRN · BraTS20 | STSR: 1.889 -> 1.224 | Improved |
| add volume-preserving loss to VTN | VTN · LiTS17 | Dice: 0.912 -> 0.908 | No clear effect |
| add volume-preserving loss to LapIRN | LapIRN · LiTS17 | Dice: 0.8933 -> 0.8893 | No clear effect |
| add volume-preserving loss to VTN | VTN · LiTS17 | Std. Jacobian determinant: 2.048 -> 1.124 | Improved |
| use adaptive estimated tumor mask instead of whole-organ mask (ablation) | VTN volume-preserving loss · LiTS17 | STSR: 1.640 -> 1.260 | Improved |
| use adaptive estimated tumor mask instead of whole-organ mask (ablation) | TransMorph volume-preserving loss · LiTS17 | STSR: 1.454 -> 1.302 | Improved |
| use ground-truth binary tumor mask instead of estimated soft mask (ablation) | VTN volume-preserving loss · LiTS17 | STSR: 1.260 -> 1.205 | Mixed |
| use soft mask (sigmoid) instead of hard threshold mask (ablation) | VTN volume-preserving loss · LiTS17 | STSR: 1.46 -> 1.26 (threshold 1.5), 1.86 -> 1.26 (threshold 2) | Improved |
| replace sigmoid with sin transformation for soft mask (ablation) | VTN volume-preserving loss · LiTS17 | STSR: 1.26 -> 1.29 | No clear effect |
| use better-registering network (higher Dice) | regular similarity-based registration · LiTS17 | STSR: VTN Dice 0.912 with STSR 2.330 vs VXM Dice 0.860 with STSR 1.738 | Hurt |
Closest papers by meaning
- An Unsupervised Learning Model for Deformable Medical Image Registration (CVPR 2018 · similarity 0.85)
- Metric Learning for Image Registration (CVPR 2019 · similarity 0.84)
- Deep Learning in Medical Image Registration: Magic or Mirage? (NeurIPS 2024 · similarity 0.83)
- Correlation-aware Coarse-to-fine MLPs for Deformable Medical Image Registration (CVPR 2024 · similarity 0.82)
- MAPSeg: Unified Unsupervised Domain Adaptation for Heterogeneous Medical Image Segmentation Based on 3D Masked Autoencoding and Pseudo-Labeling (CVPR 2024 · similarity 0.82)
- Modality-Agnostic Structural Image Representation Learning for Deformable Multi-Modality Medical Image Registration (CVPR 2024 · similarity 0.82)
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.
