Research map / Image generation / Autoregressive and latent generation
Translating and Segmenting Multimodal Medical Volumes With Cycle- and Shape-Consistency Generative Adversarial Network
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use coupled generators and segmentors with online synthetic data instead of real data alone | Cardiovascular CT segmentation · Cardiovascular CT, limited real training data | Dice score (%): 67.8 -> 74.4 | Helped |
| Use coupled generators and segmentors with online synthetic data instead of real data alone | Cardiovascular MRI segmentation · Cardiovascular MRI, limited real training data | Dice score (%): 70.3 -> 73.2 | Helped |
| Use offline synthetic data augmentation instead of real data alone | Cardiovascular CT segmentation · Cardiovascular CT, limited real training data | Dice score (%): 67.8 -> 66.0 | Hurt |
| Use offline synthetic data augmentation instead of real data alone | Cardiovascular MRI segmentation · Cardiovascular MRI, limited real training data | Dice score (%): 70.3 -> 71.0 | Helped |
| Add shape-consistency loss (ablation) | Unpaired CT-to-MRI and MRI-to-CT generators · Synthetic cardiovascular CT | S-score (%): 66.8 -> 69.2 | Helped |
| Add shape-consistency loss (ablation) | Unpaired CT-to-MRI and MRI-to-CT generators · Synthetic cardiovascular MRI | S-score (%): 67.5 -> 69.6 | Helped |
| Use online coupled training rather than offline synthetic-data augmentation | Cardiovascular CT segmentation · Cardiovascular CT, limited real training data | Dice score (%): 66.0 -> 74.4 | Helped |
| Use online coupled training rather than offline synthetic-data augmentation | Cardiovascular MRI segmentation · Cardiovascular MRI, limited real training data | Dice score (%): 71.0 -> 73.2 | Helped |
| Use 23% synthetic data in the proposed online method rather than 100% synthetic data in ADA | MRI segmentor trained with additional CT-derived synthetic data · Cardiovascular MRI, S2 | Dice score: 23% synthetic data; 100% synthetic data | Helped |
| Use the proposed method rather than ADA to reduce the real–synthetic data gap | MRI segmentation · Cardiovascular MRI, 14% real data | Gap reduction (%): 61% | Helped |
| Use the proposed method rather than ADA to reduce the real–synthetic data gap | MRI segmentation · Cardiovascular MRI, 85% real data | Gap reduction (%): 20.9% | Helped |
Closest papers by meaning
- Learning General-purpose Biomedical Volume Representations using Randomized Synthesis (ICLR 2025 · similarity 0.85)
- TCAM-Diff: Triplane-Aware Cross-Attention Medical Diffusion Model (AAAI 2025 · similarity 0.84)
- CycleINR: Cycle Implicit Neural Representation for Arbitrary-Scale Volumetric Super-Resolution of Medical Data (CVPR 2024 · similarity 0.84)
- Generative Medical Segmentation (AAAI 2025 · similarity 0.84)
- Knowledge Transfer Scaling Laws for 3D Medical Imaging (NeurIPS 2026 · similarity 0.84)
- Better Tokens for Better 3D: Advancing Vision-Language Modeling in 3D Medical Imaging (NeurIPS 2025 · similarity 0.83)
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.