Research map / Point clouds / Shape and surface completion
UBone3D: Physics-Rectified Conditional Flow Matching for Anatomical 3D Shape Completion from Ultrasound
Weiying Chen, Yuchong Gao, Siyuan Li, Marek Reformat, Rui Zheng, Edmond Lou
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 USimNet physics rectification (ablation) | UBone3D (BoneFM) · Spine1K-PC sim (trained with P_phys-full) | L1 CD x1e3: 49.196 -> 28.161 | Improved |
| add USimNet physics rectification (ablation) | UBone3D (BoneFM) · Spine1K-PC sim (trained with P_phys-full) | EMD x1e3: 96.673 -> 74.298 | Improved |
| add USimNet physics rectification (ablation) | UBone3D (BoneFM) · Spine1K-PC sim (trained with P_phys-full) | F-score@1%: 0.064 -> 0.079 | Improved |
| train USimNet on full instead of simple physics (ablation) | UBone3D · Spine1K-PC sim (trained with P_phys-full) | L1 CD x1e3: 44.287 -> 28.161 | Improved |
| train USimNet on full instead of simple physics (ablation) | UBone3D · Spine1K-PC sim (trained with P_phys-full) | EMD x1e3: 87.271 -> 74.298 | Improved |
| use simple physics proxy over none (ablation) | UBone3D · Spine1K-PC sim (trained with P_phys-full) | L1 CD x1e3: 49.196 -> 44.287 | Improved |
| use flow-matching BoneFM instead of DDIM prior BoneDiff (ablation) | UBone3D · Spine1K-PC sim (trained with P_phys-full) | L1 CD x1e3: 42.597 -> 28.161 | Improved |
| use flow-matching BoneFM instead of DDIM prior BoneDiff (ablation) | UBone3D · Spine1K-PC sim (trained with P_geo) | L1 CD x1e3: 59.274 -> 56.294 | Improved |
| use flow-matching BoneFM instead of DDIM prior BoneDiff (ablation) | UBone3D · Spine1K-PC sim (trained with P_geo) | F-score@1%: 0.068 -> 0.065 | Hurt |
| vary physics guidance weight lambda (ablation) | UBone3D · Spine1K-PC sim | CD: lambda = 0.6 best | Mixed |
| increase ODE steps toward 35 (ablation) | UBone3D · Spine1K-PC sim | CD: 25 to 35 steps largest gains | Improved |
| train on full physics simulation instead of geometric cut | UBone3D · Spine1K-PC sim (trained with P_phys-full) | L1 CD x1e3: 56.294 -> 28.161 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- PoinTr — PoinTr: Diverse Point Cloud Completion With Geometry-Aware Transformers (ICCV 2021) · better on 5 of 5 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Similarity Registration Problems for 2D/3D Ultrasound Calibration (ECCV 2016 · similarity 0.84)
- SonoGym: High Performance Simulation for Challenging Surgical Tasks with Robotic Ultrasound (NeurIPS 2025 · similarity 0.83)
- PolyPose: Deformable 2D/3D Registration via Polyrigid Transformations (NeurIPS 2025 · similarity 0.83)
- Towards a Comprehensive, Efficient and Promptable Anatomic Structure Segmentation Model Using 3D Whole-Body CT Scans (AAAI 2025 · similarity 0.82)
- VISTA3D: A Unified Segmentation Foundation Model For 3D Medical Imaging (CVPR 2025 · similarity 0.82)
- LiFT: Lifted Inter-slice Feature Trajectories for 3D Image Generation from 2D Generators (NeurIPS 2026 · similarity 0.82)
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.
