Research map / Federated learning and privacy / Personalized and heterogeneous FL
DSFedMed: Dual-Scale Federated Medical Image Segmentation via Mutual Distillation Between Foundation and Lightweight Models
Hanwen Zhang, Qiaojin Shen, Yuxi Liu, Yuesheng Zhu, Guibo Luo
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 mutual KD (ablation) | DSFedMed · Prostate | Dice: 0.780 -> 0.791 | Improved |
| add learnability-guided selection (ablation) | DSFedMed · Prostate | Dice: 0.780 -> 0.794 | Improved |
| add learnability-guided selection (ablation) | DSFedMed · training time | time (s): 399.010 -> 209.827 | Improved |
| add both mutual KD and LG selection (ablation) | DSFedMed · Prostate | Dice: 0.780 -> 0.813 | Improved |
| add both mutual KD and LG selection (ablation) | DSFedMed · CHAOS | Dice: 0.893 -> 0.927 | Improved |
| add mutual KD alone vs LG selection alone (ablation) | DSFedMed · CHAOS | Dice: 0.916 (KD) vs 0.911 (LG) vs 0.893 none | Mixed |
| use TinySAM client instead of U-Net (ablation) | DSFedMed · Fundus-OC | Dice: 0.739 -> 0.797 | Improved |
| replace SAM-B server with larger MSA-L (ablation) | DSFedMed · Fundus-OC | Dice: 0.797 -> 0.795 | No clear effect |
| use ControlNet instead of DDPM or Pix2Pix generator (ablation) | DSFedMed · Fundus generated data | FID: 34.86 -> 7.83 | Improved |
| DSFedMed vs FedSAM | DSFedMed · Average over 5 datasets | Dice and communication: 0.810 -> 0.829; 71,538 MB -> 8,920 MB | Improved |
| DSFedMed vs FedSAM on Prostate | DSFedMed · Prostate | Dice: 0.814 -> 0.813 | No clear effect |
| select samples with rate 0.2-0.8 and lambda 0.1-0.9 (ablation) | DSFedMed · five datasets | Dice | Improved |
Closest papers by meaning
- FedVCK: Non-IID Robust and Communication-Efficient Federated Learning via Valuable Condensed Knowledge for Medical Image Analysis (AAAI 2025 · similarity 0.87)
- Dual-calibrated Co-training Framework for Personalized Federated Semi-Supervised Medical Image Segmentation (AAAI 2025 · similarity 0.87)
- FedA3I: Annotation Quality-Aware Aggregation for Federated Medical Image Segmentation against Heterogeneous Annotation Noise (AAAI 2024 · similarity 0.87)
- Federated Black-Box Adaptation for Semantic Segmentation (NeurIPS 2024 · similarity 0.86)
- FedPIA – Permuting and Integrating Adapters Leveraging Wasserstein Barycenters for Finetuning Foundation Models in Multi-Modal Federated Learning (AAAI 2025 · similarity 0.86)
- Steady Progress Beats Stagnation: Mutual Aid of Foundation and Conventional Models in Mixed Domain Semi-Supervised Medical Image Segmentation (CVPR 2025 · similarity 0.86)
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.
