Research map / Image generation / Super-resolution
Resolution-Agnostic Neural Operators for Multi-Rate Sparse-View CT
Aujasvit Datta, Jiayun Wang, Asad Aali, Anima Anandkumar
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 |
|---|---|---|---|
| replace regular convolutions with DISCO neural operators in unrolled network (ablation) | CTO · AAPM abdomen CT, 18-view | PSNR: 29.14 -> 31.57 | Improved |
| train one model on all sampling rates instead of rate-specific training (ablation) | LEARN · AAPM abdomen CT, 18-view | PSNR: 4.16 -> 25.51 | Improved |
| train one model on all sampling rates instead of 72-view only (ablation) | LEARN · AAPM abdomen CT, 72-view | PSNR: 32.32 -> 29.90 | Hurt |
| add flipped circular padding along view angle (rotation equivariance) (ablation) | CTO · 24-view reconstruction | PSNR: 32.17 -> 33.00 | Improved |
| add sinogram-space operator N_s (ablation) | CTO · sparse-view CT averaged over view counts | PSNR: 4.82 dB drop when removed | Improved |
| add image-space operator N_i (ablation) | CTO · sparse-view CT averaged over view counts | PSNR: 3.86 dB drop when removed | Improved |
| add spatial branch in N_s (ablation) | CTO · sparse-view CT | PSNR: 3.1 dB drop when removed | Improved |
| add frequency branch in N_s (ablation) | CTO · sparse-view CT | PSNR: 2.7 dB drop when removed | Improved |
| apply Fourier transform along detector axis in N_s,freq vs no transform (ablation) | CTO · sparse-view CT | PSNR: 2.7 dB gain | Improved |
| use Morlet or Zernike DISCO basis instead of piecewise linear (ablation) | CTO · DISCO hyperparameter search | validation PSNR | Hurt |
| CTO vs diffusion DPS on inference speed | CTO · AAPM abdomen CT | inference time: 51.58 s -> 0.065 s | Improved |
| zero-shot 2x image resolution test | CTO · AAPM 18-view 256 to 512 | PSNR: +3 dB over CNN baseline | Improved |
Closest papers by meaning
- QN-Mixer: A Quasi-Newton MLP-Mixer Model for Sparse-View CT Reconstruction (CVPR 2024 · similarity 0.86)
- Unsupervised Self-Prior Embedding Neural Representation for Iterative Sparse-View CT Reconstruction (AAAI 2025 · similarity 0.86)
- C^2RV: Cross-Regional and Cross-View Learning for Sparse-View CBCT Reconstruction (CVPR 2024 · similarity 0.85)
- A Unified Model for Compressed Sensing MRI Across Undersampling Patterns (CVPR 2025 · similarity 0.84)
- Learning Image Priors Through Patch-Based Diffusion Models for Solving Inverse Problems (NeurIPS 2024 · similarity 0.84)
- Hierarchical Neural Operator Transformer with Learnable Frequency-aware Loss Prior for Arbitrary-scale Super-resolution (ICML 2024 · similarity 0.83)
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.
