Research map / Point clouds / Segmentation and detection
PointScatter: Point Set Representation for Tubular Structure Extraction
Dong Wang, Zhao Zhang, Ziwei Zhao, Yuhang Liu, Yihong Chen, Liwei Wang
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 mask segmentation (softDice) with PointScatter | UNet · DeepGlobe centerline extraction | Dice (%): 56.58 -> 62.78 | Improved |
| replace mask segmentation (softDice) with PointScatter | UNet · DRIVE centerline extraction | Dice (%): 73.41 -> 81.92 | Improved |
| replace mask segmentation (softDice) with PointScatter | UNet · DeepGlobe segmentation | Dice (%): 74.71 -> 78.07 | Improved |
| replace mask segmentation (softDice) with PointScatter | UNet · DRIVE segmentation | Dice (%): 81.09 -> 81.63 | Improved |
| replace mask segmentation with PointScatter | UNet · DRIVE segmentation | topology error beta_0 (lower better): 1.072 -> 1.317 | Hurt |
| add PSAUX auxiliary centerline point branch to softDice | UNet · DeepGlobe segmentation | clDice (%): 80.08 -> 83.96 | Improved |
| add PSAUX auxiliary branch to softDice | UNet · DRIVE segmentation | beta_0 error (lower better): 1.504 -> 1.004 | Improved |
| add PS auxiliary branch to softDice | UNet · DRIVE centerline extraction | Dice (%): 73.41 -> 75.92 | Improved |
| increase points per scatter region N from 8 to 16 (ablation) | PointScatter · DRIVE segmentation | Dice (%): 64.80 -> 81.63 | Improved |
| increase N from 16 to 32 (ablation) | PointScatter · DRIVE segmentation | Dice (%): 81.63 -> 78.73 | Hurt |
| increase N from 16 to 64 (ablation) | PointScatter · MassRoads centerline extraction | Dice (%): 69.63 -> 70.38 | No clear effect |
| lower downsample rate D from 4 to 2 (ablation) | PointScatter · DRIVE centerline extraction | Dice (%): 81.92 -> 82.70 | Improved |
Closest papers by meaning
- Segmentation of Tubular Structures Using Iterative Training with Tailored Samples (ICCV 2023 · similarity 0.89)
- NETracer: A Topology-Aware Iterative Tracing Approach for Tubular Structure Extraction (ICCV 2025 · similarity 0.88)
- GraphMorph: Tubular Structure Extraction by Morphing Predicted Graphs (NeurIPS 2024 · similarity 0.87)
- Skeleton Recall Loss for Connectivity Conserving and Resource Efficient Segmentation of Thin Tubular Structures (ECCV 2024 · similarity 0.87)
- Dynamic Snake Convolution Based on Topological Geometric Constraints for Tubular Structure Segmentation (ICCV 2023 · similarity 0.86)
- HarmonySeg: Tubular Structure Segmentation with Deep-Shallow Feature Fusion and Growth-Suppression Balanced Loss (ICCV 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.
