Research map / 3D vision / Gaussian splatting
Proto-Gaussian: MRI Modality Translation Based on Learnable Structural Prototypes and 2D Gaussian Splatting
Zizheng Li, RenDong Xie, Huadeng Wang, Zhifen He, Bin Liu, Bo Li, Xiaonan Luo
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 |
|---|---|---|---|
| represent images with explicit 2D Gaussians plus label-adaptive refinement instead of implicit methods | Proto-Gaussian vs SelfRDB · IXI T1->T2 | PSNR: 29.24 -> 30.57 | Improved |
| Proto-Gaussian instead of SelfRDB | Proto-Gaussian · BraTS T1->T2 | PSNR: 27.49 -> 28.52 | Improved |
| Proto-Gaussian vs MaskGAN on BraTS T2->T1 FID | Proto-Gaussian · BraTS T2->T1 | FID: 43.14 -> 43.68 | Hurt |
| weight pixel loss 0.90 with small structure/texture terms instead of pure structural loss (ablation) | Proto-Gaussian · IXI T1->T2 | PSNR: 28.23 -> 30.57 | Improved |
| use (0.05,0.05,0.90) weights instead of equal weights (ablation) | Proto-Gaussian · IXI T1->T2 | PSNR: 29.04 -> 30.57 | Improved |
| use pure pixel loss instead of mixed weights (ablation) | Proto-Gaussian · IXI T1->T2 | SSIM: 0.9512 -> 0.9290 | Hurt |
| increase Gaussian count from 256 to 1024 (ablation) | Proto-Gaussian · IXI T1->T2 | PSNR: 23.63 -> 30.57 | Improved |
| increase Gaussian count from 1024 to 1536 (ablation) | Proto-Gaussian · IXI T1->T2 | PSNR: 30.57 -> 30.10 | Hurt |
Closest papers by meaning
- Adaptive Anisotropic Gaussian Splatting for Multi-contrast MRI Arbitrary-Scale Super-Resolution with Anatomy Guidance (CVPR 2026 · similarity 0.88)
- Anatomy-Preserving Unpaired Medical Image Translation via Shared Latent Anchoring (NeurIPS 2026 · similarity 0.88)
- GaussianBlock: Building Part-Aware Compositional and Editable 3D Scene by Primitives and Gaussians (ICLR 2025 · similarity 0.85)
- Modality-Agnostic Structural Image Representation Learning for Deformable Multi-Modality Medical Image Registration (CVPR 2024 · similarity 0.85)
- Multi-modal Vision Pre-training for Medical Image Analysis (CVPR 2025 · similarity 0.85)
- Structure Consistent Gaussian Splatting with Matching Prior for Few-shot Novel View Synthesis (NeurIPS 2024 · 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.
