PyTorch implementation of Multi-site fMRI Analysis Using Privacy-preserving Federated Learning and Domain Adaptation: ABIDE Results
Our manuscript is available on arxiv https://arxiv.org/pdf/2001.05647.pdf and published at Medical Image Analysis https://www.sciencedirect.com/science/article/pii/S1361841520301298.
- Python 3.6
- Pytorch 1.1.0
- tensorboardX
- nilearn
- deepdish
- numpy
sh run.shHere we show a few examples using different strategies listed in the paper. Please check the meaning of configurations in each script.
python single.py --split ${SPLIT} --site ${SITE}
python ensemble.py --split ${SPLIT} --site ${SITE}
python cross.py --trainsite ${TRAINSITE}
python mix.py --split ${SPLIT}
python federated.py --split ${SPLIT} --noise ${NOISE} --type ${TYPE}
python federated.py --split ${SPLIT} --pace ${PACE}
python federated_MoE.py --split ${SPLIT}
python federated_align.py --split ${SPLIT}
[1]Li, X., Gu, Y., Dvornek, N., Staib, L., Ventola, P. and Duncan, J.S., 2020. Multi-site fmri analysis using privacy-preserving federated learning and domain adaptation: Abide results. arXiv preprint arXiv:2001.05647.
Please cite our paper if you find this repository useful:
@article{li2020multi,
title={Multi-site fmri analysis using privacy-preserving federated learning and domain adaptation: Abide results},
author={Li, Xiaoxiao and Gu, Yufeng and Dvornek, Nicha and Staib, Lawrence and Ventola, Pamela and Duncan, James S},
journal={Medical Image Analysis},
year={2020}
}