T2Net

Task Transformer Network for Joint MRI Reconstruction and Super-Resolution (MICCAI 2021)

[Paper][Code]

Dependencies

• numpy==1.18.5
• scikit_image==0.16.2
• torchvision==0.8.1
• torch==1.7.0
• runstats==1.8.0
• pytorch_lightning==0.8.1
• h5py==2.10.0
• PyYAML==5.4

Data Prepare

1. Download and decompress data from the link https://pan.baidu.com/s/1OdIoBwJy3GZB979JPBJS6w Password: qrlt

2. Transform .h5 format to .mat format "python convertH5tomat.py --data_dir XXX/T2Net/h5"

3. You can get the dir of as following:

• h5
  • train
  • val
  • test
  • mat
  • train
  • val
  • test

4. Set data_dir = 'XXX/T2Net/h5' at the line 4 of ixi_config.yaml

[Training code --> T2Net]

git clone https://github.com/chunmeifeng/T2Net.git

Train

single gpu train

python ixi_train_t2net.py

multi gpu train you can change the 65th line in ixi_tain_t2net.py , set num_gpus = gpu number, then run

python ixi_train_t2net.py

:fire: NEWS :fire:

• We have upload the mask file.
• Before our project, you need to transform the .nii file to .mat file at first.
• We have provided the code of converting the .nii file to .mat file as well as the .mat data.

Citation

@inproceedings{feng2021T2Net,
  title={Task Transformer Network for Joint MRI Reconstruction and Super-Resolution},
  author={Feng, Chun-Mei and Yan, Yunlu and Fu, Huazhu and Chen, Li and Xu, Yong},
  booktitle={International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI)},
  year={2021}
}