Research map / 3D vision / Gaussian splatting
FaCT-GS: Fast and Scalable CT Reconstruction with Gaussian Splatting
Pawel Pieta, Rasmus Juul Pedersen, Sina Borgi, Jakob S. Jørgensen, Jens Wenzel Andreasen, Vedrana Dahl
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 FaCT-GS instead of R2-Gaussian at same iterations | GS CT reconstruction · R2-Gaussian dataset, 75-view | time: 9m -> 1m52s | Improved |
| use FaCT-GS vs R2-Gaussian | GS CT reconstruction · R2-Gaussian dataset, 75-view | PSNR: 39.04 -> 39.05 | No clear effect |
| use FaCT-GS vs R2-Gaussian | GS CT reconstruction · R2-Gaussian dataset, 50-view | PSNR: 38.09 -> 37.97 | Hurt |
| add accelerated rasterizer (per-Gaussian backward pass) (ablation) | FaCT-GS · 75-view, RTX 4070 | runtime: +359 s when reverted | Improved |
| add fused SSIM loss (ablation) | FaCT-GS · 2k coral, LS40 | runtime: +187 s when reverted | Improved |
| add fused TV loss (ablation) | FaCT-GS · 75-view, RTX 4070 | runtime: +10 s when reverted | Improved |
| stop FaCT-GS at 150 iterations instead of 400 (ablation) | FaCT-GS · 75-view | PSNR: 39.05 -> 38.33 | Hurt |
| initialize Gaussians by FDK gradient instead of intensity (ablation) | FaCT-GS · 4 volume pairs | 3D SSIM | Improved |
| warm start with fitted volume prior (ablation) | FaCT-GS · RPLHR-CT, 5 pairs | PSNR: 34.51 -> 34.74 | Improved |
| warm start with fitted volume prior (ablation) | FaCT-GS · RPLHR-CT, 5 pairs | iterations to match cold start: 62.7% (PSNR), 61.2% (SSIM) of iterations | Improved |
| increase initial Gaussian count beyond ~50,000 (ablation) | FaCT-GS · R2-Gaussian dataset | relative SSIM | No clear effect |
Closest papers by meaning
- R$^2$-Gaussian: Rectifying Radiative Gaussian Splatting for Tomographic Reconstruction (NeurIPS 2024 · similarity 0.92)
- Volumetrically Consistent 3D Gaussian Rasterization (CVPR 2025 · similarity 0.88)
- Speedy-Splat: Fast 3D Gaussian Splatting with Sparse Pixels and Sparse Primitives (CVPR 2025 · similarity 0.87)
- DDGS-CT: Direction-Disentangled Gaussian Splatting for Realistic Volume Rendering (NeurIPS 2024 · similarity 0.87)
- Superpoint Gaussian Splatting for Real-Time High-Fidelity Dynamic Scene Reconstruction (ICML 2024 · similarity 0.86)
- GS-2DGS: Geometrically Supervised 2DGS for Reflective Object Reconstruction (CVPR 2025 · 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.
