Research map / Image generation / Text-to-image diffusion
GuideGen: A Text-Guided Framework for Paired Full-torso Anatomy and CT Volume Generation
Linrui Dai, Rongzhao Zhang, Yongrui Yu, Xiaofan Zhang
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 TCSS-generated semantic masks (ablation) | GuideGen image generator · GuideGen (TCIA/RJ val) | FID: 54.3 -> 20.2 | Improved |
| use TCSS masks instead of LDM masks (ablation) | GuideGen image generator · GuideGen (TCIA/RJ val) | FID: 38.4 -> 20.2 | Improved |
| add knowledge injection (ablation) | GuideGen · GuideGen (TCIA/RJ val) | avg accuracy: 0.57 -> 0.69 | Improved |
| add knowledge injection (ablation) | GuideGen · GuideGen (TCIA/RJ val) | avg DSC (tumor seg): 0.25 -> 0.52 | Improved |
| add knowledge injection (ablation) | GuideGen · GuideGen (TCIA/RJ val) | FID: 21.7 -> 20.2 | No clear effect |
| add anatomy preservation in autoencoder (ablation) | GuideGen · GuideGen (TCIA/RJ val) | FID: 32.4 -> 20.2 | Improved |
| add HDR accommodation (ablation) | GuideGen · GuideGen (TCIA/RJ val) | FID: 40.9 -> 20.2 | Improved |
| use categorical diffusion mask synthesizer instead of LDM | mask synthesizer · Full-torso anatomy masks | FID: 98.6 -> 7.1 | Improved |
| use categorical diffusion mask synthesizer instead of MedGen3D on tumor-only masks | mask synthesizer · Tumor-only masks | LPIPS: 0.29 -> 0.29 | No clear effect |
| train segmenter on GuideGen synthetic data instead of MAISI | nnU-Net · BTCV multi-organ | avg DSC: 0.69 -> 0.79 | Improved |
| train segmenter on GuideGen synthetic data instead of real data | nnU-Net · AMOS multi-organ | avg DSC: 0.84 -> 0.78 | Hurt |
| train segmenter on GuideGen synthetic data instead of real data | nnU-Net · MSD lung tumor | DSC: 0.69 -> 0.71 | Improved |
Closest papers by meaning
- Sketch2CT: Multimodal Diffusion for Structure-Aware 3D Medical Volume Generation (CVPR 2026 · similarity 0.90)
- Flexible Control of 3D CT Generation via Text and Semantically-Defined Segmentation Prompts (ECCV 2026 · similarity 0.89)
- CAT: Coordinating Anatomical-Textual Prompts for Multi-Organ and Tumor Segmentation (NeurIPS 2024 · similarity 0.88)
- Better Tokens for Better 3D: Advancing Vision-Language Modeling in 3D Medical Imaging (NeurIPS 2025 · similarity 0.87)
- Towards a Comprehensive, Efficient and Promptable Anatomic Structure Segmentation Model Using 3D Whole-Body CT Scans (AAAI 2025 · similarity 0.87)
- LiFT: Lifted Inter-slice Feature Trajectories for 3D Image Generation from 2D Generators (NeurIPS 2026 · similarity 0.85)
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.
