Research map / Label-efficient and robust learning / Domain adaptation and generalization
Taxonomy Adaptive Cross-Domain Adaptation in Medical Imaging via Optimization Trajectory Distillation
Jianan Fan, Dongnan Liu, Hang Chang, Heng Huang, Mei Chen, Weidong Cai
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 cross-domain distillation (ablation) | Optimization trajectory distillation · Skin lesion diagnosis (10-shot) | mAcc*: 44.34 -> 52.14 | Improved |
| add cross-class distillation (ablation) | Optimization trajectory distillation · Cancer tissue phenotyping (10-shot) | mAcc: 48.90 -> 55.86 | Improved |
| add historical self-distillation (ablation) | Optimization trajectory distillation · Cancer tissue phenotyping (10-shot) | mAcc: 50.17 -> 55.86 | Improved |
| add gradient projection (ablation) | Optimization trajectory distillation · Nuclei PanNuke to Lizard (10-shot) | mF1: 40.06 -> 43.88 | Improved |
| add cross-domain distillation on cancer task (ablation) | Optimization trajectory distillation · Cancer tissue phenotyping (10-shot) | mAcc: 52.81 -> 55.86 | Improved |
| apply competing UDA/new-class methods such as DANN instead of Sup-only on nuclei new-class metrics | DANN · Nuclei PanNuke to Lizard (5-shot) | mF1*: 4.94% worse than Sup-only on mF1* despite 4.68% gain on mF1 | Hurt |
| apply FT-CIDA on cancer tissue phenotyping versus supervised only baseline | FT-CIDA · Cancer tissue phenotyping (5-shot) | mAcc: 46.82 -> 43.72 | Hurt |
| propose method versus second-best on cancer tissue phenotyping | Optimization trajectory distillation · Cancer tissue phenotyping | mAcc and mF1: +0.62~5.32 mAcc; +1.12~4.81 mF1 | Improved |
| increase lambda and kappa to large values (ablation) | Optimization trajectory distillation · Cancer tissue phenotyping (10-shot) | classification accuracy | Hurt |
Closest papers by meaning
- Perturbating, Tuning, and Collaborating: Harnessing Vision Foundation Models for Single Domain Generalization on Medical Imaging (AAAI 2025 · similarity 0.86)
- Knowledge Transfer Scaling Laws for 3D Medical Imaging (NeurIPS 2026 · similarity 0.86)
- Samba: Severity-aware Recurrent Modeling for Cross-domain Medical Image Grading (NeurIPS 2024 · similarity 0.86)
- AIF-SFDA: Autonomous Information Filter Driven Source-Free Domain Adaptation for Medical Image Segmentation (AAAI 2025 · similarity 0.86)
- Open-Set Heterogeneous Domain Adaptation: Theoretical Analysis and Algorithm (AAAI 2025 · similarity 0.85)
- Constructing and Exploring Intermediate Domains in Mixed Domain Semi-supervised Medical Image Segmentation (CVPR 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.
