Research map / Architectures and efficiency / CNN design
Fourier-Net: Fast Image Registration with Band-Limited Deformation
Xi Jia, Joseph Bartlett, Wei Chen, Siyang Song, Tianyang Zhang, Xinxing Cheng, Wenqi Lu, Zhaowen Qiu, Jinming Duan
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 |
|---|---|---|---|
| Use band-limited Fourier-Net-S instead of TransMorph | 3D IXI registration · 3D IXI | Dice: .754±.124 -> .759±.132 | Improved |
| Use band-limited Fourier-Net-S instead of TransMorph | 3D IXI registration · 3D IXI | CPU inference time (s): 22.035 -> 1.919 | Improved |
| Use Fourier-Net-L instead of TransMorph-Large | 3D OASIS registration · 3D OASIS | HD95: 1.431 -> 1.375 | Improved |
| Increase band-limited patch size from 20×24 to 40×48 (ablation) | Fourier-Net, 2D OASIS · 2D OASIS | Dice: .735±.037 -> .756±.037 | Improved |
| Add the encoder DFT layer (ablation) | Fourier-Net, 2D OASIS, patch 20×24 · 2D OASIS | Dice: .664±.040 -> .732±.042 | Improved |
| Add the encoder DFT layer (ablation) | Fourier-Net, 2D OASIS, patch 40×48 · 2D OASIS | Dice: .675±.038 -> .756±.039 | Improved |
| Add squaring and scaling to the 20×24 Fourier-Net (ablation) | Fourier-Net, 2D OASIS · 2D OASIS | negative Jacobian percentage: .434±.355 -> 0.0±0.0 | Improved |
| Add squaring and scaling to the 40×48 Fourier-Net (ablation) | Fourier-Net, 2D OASIS · 2D OASIS | negative Jacobian percentage: .753±.407 -> <0.0001 | Improved |
| Add squaring and scaling to the 20×24 Fourier-Net (ablation) | Fourier-Net, 2D OASIS · 2D OASIS | Dice: .732±.042 -> .735±.037 | No clear effect |
| Add squaring and scaling to the 40×48 Fourier-Net (ablation) | Fourier-Net, 2D OASIS · 2D OASIS | Dice: .756±.039 -> .756±.037 | No clear effect |
| Increase Fourier-Net model size from small to large (ablation) | Fourier-Net, 2D OASIS · 2D OASIS | Dice: .748±.039 -> .759±.040 | Improved |
| Increase Fourier-Net model size from small to large (ablation) | Fourier-Net, 2D OASIS · 2D OASIS | CPU inference time (s): .007 -> .037 | Hurt |
Closest papers by meaning
- Learning-Based Image Registration With Meta-Regularization (CVPR 2021 · similarity 0.87)
- EDFFDNet: Towards Accurate and Efficient Unsupervised Multi-Grid Image Registration (ICCV 2025 · similarity 0.86)
- Learning Spatially Collaged Fourier Bases for Implicit Neural Representation (AAAI 2024 · similarity 0.85)
- CARL: A Framework for Equivariant Image Registration (CVPR 2025 · similarity 0.85)
- A Plug-and-Play Image Registration Network (ICLR 2024 · similarity 0.84)
- How Learnable Grids Recover Fine Detail in Low Dimensions: A Neural Tangent Kernel Analysis of Multigrid Parametric Encodings (ICLR 2025 · similarity 0.84)
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.
