Research map / Label-efficient and robust learning / Semi-supervised learning
Gradient-Aware for Class-Imbalanced Semi-supervised Medical Image Segmentation
Wenbo Qi, Jiafei Wu, S. C. Chan
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 |
|---|---|---|---|
| apply GA loss to MT | MT · Synapse (20% labeled) | DSC: 45.58 -> 57.67 | Improved |
| apply GA loss to CPS | CPS · Synapse (20% labeled) | DSC: 48.49 -> 66.39 | Improved |
| apply GA loss to MagicNet | MagicNet · AMOS (5% labeled) | DSC: 54.08 -> 63.51 | Improved |
| apply GA loss to BCP | BCP · ACDC (5% labeled) | DSC: 87.59 -> 88.24 | Improved |
| apply GA loss to SS-Net | SS-Net · ACDC (5% labeled) | DSC: 65.83 -> 73.39 | Improved |
| add dynamic weight equilibrium (DWE) to baseline (ablation) | MagicNet (17k iters) · Synapse (20% labeled) | DSC: 42.33 -> 66.96 | Improved |
| add hard instance mining class equilibrium (ablation) | MagicNet (17k iters) · Synapse (20% labeled) | DSC: 42.33 -> 56.46 | Improved |
| add GADice loss (ablation) | MagicNet (17k iters) · Synapse (20% labeled) | DSC: 42.33 -> 44.98 | Improved |
| add GADice on top of GACE (ablation) | GA-MagicNet · Synapse (20% labeled) | DSC: 67.99 -> 68.43 | Improved |
| set gamma to 0.5 versus 1 or 0 (ablation) | GACE · Synapse (20% labeled) | DSC: 59.40 -> 68.43 (gamma 1 to 0.5); 52.22 at gamma 0 | Improved |
| restrict hard instance mining to worst 10% instead of all (ablation) | GACE · Synapse (20% labeled) | DSC: 67.18 -> 68.43 | Improved |
| apply GA loss on both labeled and unlabeled data rather than one (ablation) | CPS · Synapse (20% labeled) | DSC: 59.12 -> 66.39 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- MagicNet — MagicNet: Semi-Supervised Multi-Organ Segmentation via Magic-Cube Partition and Recovery (CVPR 2023) · better on 2 of 2 shared results
- CReST — CReST: A Class-Rebalancing Self-Training Framework for Imbalanced Semi-Supervised Learning (CVPR 2021) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- A Unified Loss for Handling Inter-Class and Intra-Class Imbalance in Medical Image Segmentation (AAAI 2025 · similarity 0.89)
- LCGC: Learning from Consistency Gradient Conflicting for Class-Imbalanced Semi-Supervised Debiasing (AAAI 2025 · similarity 0.86)
- Towards Realistic Semi-supervised Medical Image Classification (AAAI 2025 · similarity 0.86)
- DyCON: Dynamic Uncertainty-aware Consistency and Contrastive Learning for Semi-supervised Medical Image Segmentation (CVPR 2025 · similarity 0.86)
- Combinatorial CNN-Transformer Learning with Manifold Constraints for Semi-supervised Medical Image Segmentation (AAAI 2024 · similarity 0.85)
- Gradient Alignment Improves Test-Time Adaptation for Medical Image Segmentation (AAAI 2025 · 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.
