This is an official release of the paper Personalizing Federated Medical Image Segmentation via Local Calibration, including the network implementation and the training scripts.
Personalizing Federated Medical Image Segmentation via Local Calibration,
Jiacheng Wang, Yueming Jin, Liansheng Wang
In: European Conference on Computer Vision (ECCV), 2022
[arXiv][Bibetex][Supp]
- [3/1 2023] Codes for Head Calibration are tuned.
- [7/12 2022] We have released the training codes.
- [7/25 2022] We have uploaded the test scripts.
- [7/12 2022] We have released the pre-print manuscript.
- [7/11 2022] We have released the pre-trained weights on the polyp segmentation.
- [7/4 2022] We have released the pre-processing scripts.
- [7/4 2022] We have created this repo.
- Network
- Pre-processing
- Training Codes
- Pretrained Weights
For more details or any questions, please feel easy to contact us by email (jiachengw@stu.xmu.edu.cn).
In this paper, we perform the experiments using three imaging modalities, including the polyp images, fundus images, and prostate MR images. They could be downloaded from the public websites, or copied from FedDG and PraNet.
After downloading the data resources, please run the file utils/prepare_dataset.py. Note that the file directory should be replaced with yours.
Run the train script $ python scripts/train_lcfed.py.
Please download the pre-trained weights from Baidu Disk (https://pan.baidu.com/s/10HkQ90xeFcHMaNgfIyT0iw, a1sm) and put them in the project directory.
Rename the directory as logs/{dataset}/{exp_name}/model/.
Run the test script $ python scripts/test.py.
The test IoU scores and ASSD scores on the Polyp dataset are:
If you find LC-Fed useful in your research, please consider citing:
@inproceedings{wang2022personalizing,
title={Personalizing Federated Medical Image Segmentation via Local Calibration},
author={Wang, Jiacheng and Jin, Yueming and Wang, Liansheng},
booktitle={European Conference on Computer Vision},
pages={456--472},
year={2022},
organization={Springer}
}