Research map / Label-efficient and robust learning / Self-supervised and contrastive
Self-Supervised Vessel Segmentation via Adversarial Learning
Yuxin Ma, Yang Hua, Hanming Deng, Tao Song, Hao Wang, Zhengui Xue, Heng Cao, Ruhui Ma, Haibing Guan
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 attention-guided generator (ablation) | CycleGAN base with fractal maps · XCAD ablation | Jaccard: 0.229 -> 0.305 | Improved |
| add attention-guided generator (ablation) | CycleGAN base with fractal maps · XCAD ablation | Sensitivity: 0.320 -> 0.487 | Improved |
| add segmentation loss (ablation) | CycleGAN base with fractal maps · XCAD ablation | Jaccard: 0.229 -> 0.346 | Improved |
| add segmentation loss (ablation) | CycleGAN base with fractal maps · XCAD ablation | Dice: 0.363 -> 0.511 | Improved |
| add segmentation loss on top of attention (ablation) | Base + Attention · XCAD ablation | Jaccard: 0.305 -> 0.389 | Improved |
| add attention-guided generator on top of segmentation loss (ablation) | Base + SegLoss · XCAD ablation | Jaccard: 0.346 -> 0.389 | Improved |
| combine attention and segmentation loss (ablation) | Base · XCAD ablation | Accuracy: 0.850 -> 0.945 | Improved |
| self-supervised training vs domain adaptation from DRIVE | UNet/MMD/YNet · XCAD test (126 angiograms) | Jaccard: 0.287 -> 0.389 | Improved |
| self-supervised vs unsupervised ReDO | ReDO · XCAD test (126 angiograms) | Jaccard: 0.151 -> 0.389 | Improved |
| full supervision with annotations vs self-supervised | Ours · XCAD test (126 angiograms) | Jaccard: 0.571 -> 0.389 | Hurt |
| self-supervised XCAD-pretrained model vs traditional on DRIVE | Khan · DRIVE | AUC: 0.885 -> 0.888 | No clear effect |
| self-supervised vs best supervised Soomro | Soomro · DRIVE | AUC: 0.948 -> 0.888 | Hurt |
Closest papers by meaning
- Diffusion Adversarial Representation Learning for Self-supervised Vessel Segmentation (ICLR 2023 · similarity 0.93)
- DeNVeR: Deformable Neural Vessel Representations for Unsupervised Video Vessel Segmentation (CVPR 2025 · similarity 0.89)
- vesselFM: A Foundation Model for Universal 3D Blood Vessel Segmentation (CVPR 2025 · similarity 0.87)
- Learning Dynamic Collaborative Network for Semi-supervised 3D Vessel Segmentation (CVPR 2025 · similarity 0.85)
- VasTSD: Learning 3D Vascular Tree-state Space Diffusion Model for Angiography Synthesis (CVPR 2025 · similarity 0.82)
- Generative Medical Segmentation (AAAI 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.
