TLR-BM-MRF

A simple implementation of the TLR-BM-Net for MRF reconstruction: "Learned Tensor Low-CP-Rank and Bloch Response Manifold Priors for Non-Cartesian MRF Reconstruction"

Files

• 'configs': contains the configuration files
• 'data': contains the data required for the training and testing
• 'dataset': data loader
• 'model': model definition
• 'utils': utility functions
• 'train.py': training script
• 'test.py': testing script
• 'demp.ipynb': demo script

  Data preparation

• Download the data from here
• put the data in the 'data' folder
• the data folder should look like this:

  data
  ├── MRF_dataset
  │   ├── csm_maps
  │   ├── k_samples
  │   ├── para_maps
  │   ├── X
  │   ├── ktraj_nufft.mat
  ├──output
  │   ├── Model_save
  │   ├── TLR_BM_X_4_5_2023-04-17-13 (Note: a output data folder for example)

  Demo

• complete the "Data preparation" step
• run the 'demo.ipynb' file to see the results

  Training

  prepare your data as the above data structure

• run the following command to train the model

  python train.py

• the results will be saved in the 'data/output/TLR_BM_X_4_5_2023-04-17-13/' folder

  Testing

  run the following command to test the model

  python test.py

• the results will be saved in the 'data/output/TLR_BM_X_4_5_2023-04-17-13/test/' folder

Requirements

• torchkbnufft
• PyTorch >=1.13.0
• numpy
• scipy
• matplotlib

References

• https://ieeexplore.ieee.org/document/10210402
• @ARTICLE{10210402, author={Li, Peng and Hu, Yue}, journal={IEEE Transactions on Medical Imaging}, title={Learned Tensor Low-CP-Rank and Bloch Response Manifold Priors for Non-Cartesian MRF Reconstruction}, year={2023}, volume={}, number={}, pages={1-1}, doi={10.1109/TMI.2023.3302872}}