UltraFedFM — Federated Ultrasound Foundation Model (mirror)

This is an unmodified mirror. All credit goes to the original authors.

• Original code: https://github.com/yuncheng97/UltraFedFM
• Original weights: OneDrive / Google Drive / Baidu links in the upstream README (this is log_2024-07-16_13_53_08/checkpoint.pth from the OneDrive folder)
• Paper: npj Digital Medicine 8, 714 (2025)

A privacy-preserving federated ultrasound foundation model, collaboratively pre-trained with self-supervised learning (MAE) across 16 institutions without sharing raw data — ~1M images, 19 organs, 10 modalities — for downstream diagnosis and segmentation.

Files

File	Purpose
checkpoint.pth (1.34 GB)	Pre-trained UltraFedFM checkpoint — place in output_dir/ per the upstream README
LICENSE	Upstream Apache-2.0 license text

License

Apache-2.0, as released by the original authors. See LICENSE.

Citation

@article{jiang2025pretraining,
  title     = {From pretraining to privacy: federated ultrasound foundation model with self-supervised learning},
  author    = {Jiang, Yuncheng and Feng, Chun-Mei and Ren, Jinke and Wei, Jun and Zhang, Zixun and Hu, Yiwen and Liu, Yunbi and Sun, Rui and Tang, Xuemei and Du, Juan and others},
  journal   = {npj Digital Medicine},
  volume    = {8},
  number    = {1},
  pages     = {714},
  year      = {2025},
  publisher = {Nature Publishing Group UK London}
}