Research map / Label-efficient and robust learning / Domain adaptation and generalization
JSSR: A Joint Synthesis, Segmentation, and Registration System for 3D Multi-Modal Image Alignment of Large-scale Pathological CT Scans
Fengze Liu, Jinzheng Cai, Yuankai Huo, Chi-Tung Cheng, Ashwin Raju, Dakai Jin, Jing Xiao, Alan Yuille, Le Lu, ChienHung Liao, Adam P. Harrison
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 synthesis generator (JSynR) to registration, over VoxelMorph (ablation) | VoxelMorph · CGMH liver registration (Arterial to venous) | Dice (%): 94.28 -> 94.81 | Improved |
| add segmentation module to JSynR (full JSSR) (ablation) | JSynR-Reg · CGMH liver registration (Arterial to venous) | Dice (%): 94.81 -> 95.56 | Improved |
| add synthesis generator to JSegR (full JSSR) (ablation) | JSegR-Reg · CGMH liver registration (Non-Contrast to venous) | Dice (%): 95.37 -> 95.41 | No clear effect |
| add segmentation module to registration without generator (JSegR) over JSynR (ablation) | JSynR-Reg · CGMH liver registration (Non-Contrast to venous) | Dice (%): 94.57 -> 95.37 | Improved |
| use JSSR-Reg instead of Deeds | Deeds · CGMH liver registration | Dice (%): 94.73 -> 95.56 | Improved |
| use JSSR-Reg instead of VoxelMorph | VoxelMorph · CGMH liver registration | Dice (%): exceeded by 1.28% | Improved |
| joint training for segmentation (JSSR-Seg vs VNet) with No-Synthesis | VNet · CGMH liver segmentation | Dice (%): 1.86% | Improved |
| apply Pix2Pix synthesis before segmentation (ablation) | JSSR-Seg · CGMH liver segmentation (Arterial) | Dice (%): 91.25 -> 85.30 | Hurt |
| apply Pix2Pix synthesis before segmentation (ablation) | VNet · CGMH liver segmentation (Non-Contrast) | Dice (%): 89.38 -> 67.48 | Hurt |
| use JSSR-Syn or JSynR-Syn generator instead of no synthesis (ablation) | JSSR-Seg · CGMH liver segmentation (Arterial) | Dice (%): 91.25 -> 91.39 | No clear effect |
| use JSynR-Syn generator instead of Pix2Pix (ablation) | VNet · CGMH liver segmentation (Non-Contrast) | Dice (%): 67.48 -> 90.15 | Improved |
| add segmentation consistency loss without synthesis in registration (ablation) | registration across phases · CGMH liver registration | HD95 (mm): 4.93 -> 4.44 | Improved |
Closest papers by meaning
- Foundation-Guided Representation Alignment for Multimodal Medical Image Registration (ECCV 2026 · similarity 0.87)
- Deep Complementary Joint Model for Complex Scene Registration and Few-shot Segmentation on Medical Images (ECCV 2020 · similarity 0.87)
- Modality-Agnostic Structural Image Representation Learning for Deformable Multi-Modality Medical Image Registration (CVPR 2024 · similarity 0.86)
- Better Tokens for Better 3D: Advancing Vision-Language Modeling in 3D Medical Imaging (NeurIPS 2025 · similarity 0.85)
- Autoregressive Sequence Modeling for 3D Medical Image Representation (AAAI 2025 · similarity 0.85)
- CAT: Coordinating Anatomical-Textual Prompts for Multi-Organ and Tumor Segmentation (NeurIPS 2024 · 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.
