Universal Coil Configuration-Adaptive model for Multi-coil MRI Reconstruction

Universal Coil Configuration-Adaptive model for Multi-coil MRI Reconstruction using Hypernetworks

HyperCoil-Recon: A Hypernetwork-based Adaptive Coil Configuration Task Switching Network for MRI Reconstruction

Dependencies

Packages

• PyTorch
• TensorboardX
• numpy
• tqdm

An exhaustive list of packages used could be found in the requirements.txt file. Install the same using the following command:

 pip install -r requirements.txt

Folder hierarchies

The hierarchy is dataset or experiments/<model_name>/results folder

<base_path>/experiments/<model_name>/results/DATASET_TYPE/MASK_TYPE/ACC_FACTOR

<model_name> - this is the folder with the model name in which the model files are stored. results - this is the folder in which all the predicted test files are stored in .h5 format for each acquisition context.

Train code

sh train.sh

Test code

sh valid.sh