Research map / Label-efficient and robust learning / Domain adaptation and generalization
Wavelet-Driven Cross-Domain Consistency for Mixed-Supervised 3D Tumor Segmentation
Tianzhong Lan, Weili Jiang, Yisong Liu, Yi Zhou, Junqi Bai, Si Yong Yeo, Xulei Yang, Min Zhu
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 all four components (KernelWave, KL, domain head, detection branch) (ablation) | 3D UNet baseline · LIDC-IDRI | DSC: 63.17 -> 80.32 | Improved |
| add the auxiliary detection branch using box labels (ablation) | WCC4MS · LIDC-IDRI | DSC | Improved |
| add KernelWave (ablation) | WCC4MS · LIDC-IDRI | DSC: +2.04% DSC | Improved |
| add KL consistency and domain head to AD-only model (ablation) | WCC4MS · LIDC-IDRI | DSC std: 0.31 to 0.12 | Improved |
| add domain-aware head (ablation) | WCC4MS · cross-domain | DSC: 1.6% | Improved |
| use KernelWave instead of Gaussian blur (ablation) | WCC4MS · LIDC-IDRI | DSC: 79.48 -> 80.32 | Improved |
| use KernelWave instead of Laplacian (ablation) | WCC4MS · LIDC-IDRI | DSC: 76.98 -> 80.32 | Improved |
| use KernelWave instead of Fourier augmentation (ablation) | WCC4MS · LIDC-IDRI | DSC: 77.56 -> 80.32 | Improved |
| raise wavelet gain from 2 to 4 (ablation) | WCC4MS · LIDC-IDRI | DSC: 79.26 -> 80.32 | Improved |
| raise wavelet gain from 4 to 5 (ablation) | WCC4MS · LIDC-IDRI | DSC: 80.32 -> 79.94 | Hurt |
| increase weak box ratio from 1:2 to 1:4 and beyond (ablation) | WCC4MS · LIDC-IDRI | DSC | Hurt |
| add 20% box noise | WCC4MS · LIDC-IDRI | DSC: 77.6% vs 72.9% | Hurt |
Closest papers by meaning
- A Unified Degradation-Robust Approach to SSL and UDA for 3D Medical Images (AAAI 2025 · similarity 0.88)
- CSC-PA: Cross-image Semantic Correlation via Prototype Attentions for Single-network Semi-supervised Breast Tumor Segmentation (CVPR 2025 · similarity 0.86)
- DyCON: Dynamic Uncertainty-aware Consistency and Contrastive Learning for Semi-supervised Medical Image Segmentation (CVPR 2025 · similarity 0.85)
- Why Invariance is Not Enough for Biomedical Domain Generalization and How to Fix It (NeurIPS 2026 · similarity 0.85)
- Knowledge Transfer Scaling Laws for 3D Medical Imaging (NeurIPS 2026 · similarity 0.85)
- Constructing and Exploring Intermediate Domains in Mixed Domain Semi-supervised Medical Image Segmentation (CVPR 2024 · similarity 0.85)
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.
