Research map / Detection and segmentation / Weakly supervised detection and segmentation
CycleMix: A Holistic Strategy for Medical Image Segmentation From Scribble Supervision
Ke Zhang, Xiahai Zhuang
Check 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 |
|---|---|---|---|
| Use CycleMix instead of the strongest scribble baseline, Cutout | UNet+ on ACDC · ACDC | Dice (average): .800 -> .848 | Improved |
| Use CycleMix instead of the strongest scribble baseline, CutMix | UNet+ on MSCMRseg · MSCMRseg | Dice (average): .654 -> .800 | Improved |
| Use scribble-supervised CycleMix instead of the fully supervised UNet+ F | UNet+ on ACDC · ACDC | Dice (average): .820 -> .848 | Improved |
| Use scribble-supervised CycleMix instead of the fully supervised UNet+ F | UNet+ on MSCMRseg · MSCMRseg | Dice (average): .755 -> .800 | Improved |
| Add mixed-image scribble loss Lmix (ablation) | CycleMix on ACDC · ACDC | Dice (average): .752 -> .809 | Improved |
| Add global consistency loss Lcon-g (ablation) | CycleMix on ACDC · ACDC | Dice (average): .809 -> .830 | Improved |
| Add random occlusion to the mixed images (ablation) | CycleMix on ACDC · ACDC | Dice (average): .830 -> .843 | Improved |
| Add local consistency loss Lcon-l (ablation) | CycleMix on ACDC · ACDC | Dice (average): .843 -> .848 | Improved |
| Replace scribble-only training with a mix of 56 scribble and 14 fully labeled images | CycleMix on ACDC · ACDC | Dice (average): .848 -> .872 | Improved |
| Replace the scribble-only training set with fully annotated masks | CycleMix on ACDC · ACDC | Dice (average): .848 -> .886 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- CutMix — CutMix: Regularization Strategy to Train Strong Classifiers With Localizable Features (ICCV 2019) · 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
- Scribble-Supervised Semantic Segmentation with Prototype-based Feature Augmentation (ICML 2024 · similarity 0.87)
- Scribbles for All: Benchmarking Scribble Supervised Segmentation Across Datasets (NeurIPS 2024 · similarity 0.87)
- LoMix: Learnable Weighted Multi-Scale Logits Mixing for Medical Image Segmentation (NeurIPS 2025 · similarity 0.85)
- Interactive Medical Image Segmentation: A Benchmark Dataset and Baseline (CVPR 2025 · similarity 0.84)
- ConDSeg: A General Medical Image Segmentation Framework via Contrast-Driven Feature Enhancement (AAAI 2025 · similarity 0.84)
- Scribble Hides Class: Promoting Scribble-Based Weakly-Supervised Semantic Segmentation with Its Class Label (AAAI 2024 · 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.
