Research map / Detection and segmentation / Open-set and few-shot detection
SegAnyPET: Universal Promptable Segmentation from Positron Emission Tomography Images
Yichi Zhang, Le Xue, Wenbo Zhang, Lanlan Li, Yuchen Liu, Chen Jiang, Yuan Cheng, Yuan Qi
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 cross prompting consistency regularization on LQ images (ablation) | SegAnyPET · PETS-5k seen organs, 1 point | DSC (%): 87.61 -> 89.28 | Improved |
| add cross prompting consistency regularization on LQ images (ablation) | SegAnyPET · PETS-5k unseen organs, 1 point | DSC (%): 62.74 -> 67.52 | Improved |
| add uncertainty-guided self-rectification on top of consistency (full CPCL) (ablation) | SegAnyPET · PETS-5k seen organs, 1 point | DSC (%): 89.28 -> 90.49 | Improved |
| add uncertainty-guided self-rectification on top of consistency (full CPCL) (ablation) | SegAnyPET · PETS-5k unseen organs, 1 point | DSC (%): 67.52 -> 73.96 | Improved |
| add uncertainty-guided self-rectification on top of consistency (full CPCL) (ablation) | SegAnyPET · PETS-5k unseen organs, 5 points | DSC (%): 70.97 -> 77.87 | Improved |
| increase prompt points from 1 to 5 | SegAnyPET · PETS-5k seen organs | DSC (%): 90.49 -> 91.05 | Improved |
| increase prompt points from 1 to 5 | SAM-Med3D-turbo · PETS-5k seen organs | DSC (%): 72.09 -> 78.35 | Improved |
| use 3D volumetric model instead of slice-by-slice 2D SAM | SegAnyPET vs SAM · PETS-5k seen organs, 1 point | DSC (%): 13.15 -> 90.49 | Improved |
| train on PET with PETS-5k instead of using CT-based SAM-Med3D-turbo | SegAnyPET · organ segmentation, 1 point | DSC (%): 66.59 -> 90.49 | Improved |
| use SegAnyPET promptable model instead of automatic nnUNet | SegAnyPET · PETS-5k seen organs | DSC (%): 89.83 -> 90.49 | Improved |
| apply 3D SegAnyPET to scattered tumor lesions | SegAnyPET vs SAM · AutoPET tumor, 5 points | DSC (%): 40.91 -> 24.93 | Hurt |
| evaluate SegAnyPET on unseen AutoPET-Organ lung and aorta organs | SegAnyPET vs SAM-Med3D · AutoPET-Organ Lung-UL, 1 point | DSC (%): 51.69 -> 10.80 | Hurt |
Closest papers by meaning
- PET2Rep: Towards Vision-Language Model-Drived Automated Radiology Report Generation for Positron Emission Tomography (AAAI 2026 · similarity 0.86)
- PETAR: Localized Findings Generation with Mask-Aware Vision-Language Modeling for PET Automated Reporting (CVPR 2026 · similarity 0.84)
- UniverSeg: Universal Medical Image Segmentation (ICCV 2023 · similarity 0.83)
- VISTA3D: A Unified Segmentation Foundation Model For 3D Medical Imaging (CVPR 2025 · similarity 0.82)
- Training Like a Medical Resident: Context-Prior Learning Toward Universal Medical Image Segmentation (CVPR 2024 · similarity 0.82)
- Cross-Modal Interactive Perception Network with Mamba for Lung Tumor Segmentation in PET-CT Images (CVPR 2025 · similarity 0.81)
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.
