Edit model card
YAML Metadata Warning: empty or missing yaml metadata in repo card (https://huggingface.co/docs/hub/model-cards#model-card-metadata)

CascadeNet-fastmri


tags:

  • TensorFlow
  • MRI reconstruction
  • MRI datasets:
  • fastMRI

This model can be used to reconstruct single coil fastMRI data with an acceleration factor of 4.

Model description

For more details, see https://www.mdpi.com/2076-3417/10/5/1816. This section is WIP.

Intended uses and limitations

This model can be used to reconstruct single coil knee data from Siemens scanner at acceleration factor 4. It cannot be used on multi-coil data.

How to use

This model can be loaded using the following repo: https://github.com/zaccharieramzi/fastmri-reproducible-benchmark. After cloning the repo, git clone https://github.com/zaccharieramzi/fastmri-reproducible-benchmark, you can install the package via pip install fastmri-reproducible-benchmark. The framework is TensorFlow.

You can initialize and load the model weights as follows:

from fastmri_recon.models.functional_models.cascading import cascade_net

model = cascade_net()
model.load_weights('model_weights.h5')

Using the model is then as simple as:

model([
    kspace,  # shape: [n_slices, n_rows, n_cols, 1]
    mask,  # shape: [n_slices, n_rows, n_cols]
])

Limitations and bias

The limitations and bias of this model have not been properly investigated.

Training data

This model was trained using the fastMRI dataset.

Training procedure

The training procedure is described in https://www.mdpi.com/2076-3417/10/5/1816 for brain data. This section is WIP.

Evaluation results

This model was evaluated using the fastMRI dataset.

Contrast PD PDFS
PSNR 33.98 29.88
SSIM 0.811 0.6251

Bibtex entry

@article{ramzi2020benchmarking,
  title={Benchmarking MRI reconstruction neural networks on large public datasets},
  author={Ramzi, Zaccharie and Ciuciu, Philippe and Starck, Jean-Luc},
  journal={Applied Sciences},
  volume={10},
  number={5},
  pages={1816},
  year={2020},
  publisher={Multidisciplinary Digital Publishing Institute}
}
Downloads last month
0
Unable to determine this model's library. Check the docs .