README.md

ICW-GANs

This is a Tensorflow implementation of my paper:

FMRI data augmentation via synthesis, The IEEE International Symposium on Biomedical Imaging (ISBI'19)

Peiye Zhuang, Alexander Schwing, and Sanmi Koyejo

Prerequisites

• Tensorflow 1.x
• Python3
• NVIDIA GPU + CUDA CuDNN
• Scipy 1.1.0, Nilearn etc.

Data

We used a public dataset BrainPedia 1952. You may either download the dataset by yourself or use our preprocessed data on GoogleDrive.

Pretrained models

You may download our pretrained model checkpoint on GoogleDrive.

Training & testing

 python icw_gans.py

You may change default parameter settings in the argparse. We did not write an independent python file for testing. Instead, we used the function save_test_img in the code to save test images after amount of training epochs.

Citation

If you use this code for your research, please cite our paper:

@article{icwgans,
  title={FMRI data augmentation via synthesis},
  author={Peiye Zhuang, Alexander Schwing, Sanmi Koyejo},
  journal={The IEEE International Symposium on Biomedical Imaging (ISBI)},
  year={2019}
}