search

facebookresearch / fastMRI

A large-scale dataset of both raw MRI measurements and clinical MRI images.
https://fastmri.org
MIT License
1.29k stars 370 forks source link

Add code for emulated single coil #202

Open mmuckley opened 2 years ago

mmuckley commented 2 years ago

This issue would add code for simulating the single-coil data from multi-coil data described in the following paper:

Tygert M, Zbontar J. Simulating single-coil MRI from the responses of multiple coils. Communications in Applied Mathematics and Computational Science. 2020 Nov 19;15(2):115-27. https://msp.org/camcos/2020/15-2/p01.xhtml

sriram-1 commented 2 years ago

Hi, am interested. Please let me know the next steps.

mmuckley commented 2 years ago

Hello @sriram-1, that would be fantastic! I think the first thing would be to read the paper and understand the mathematics of how the single-coil data was simulated. Then you could create a function with a signature like

def simulate_singlecoil_from_multicoil(kspace: Tensor) -> Tensor:
    # write function here

that would return the single-coil k-space based on the paper.

Create a new folder under fastmri_examples called singlecoil_simulation and include a Jupyter notebook using your function and plot the reconstruction. Compare the reconstruction from single-coil to the reconstruction from multi-coil. The single-coil should be noisier.

Once you have all this you can open a pull request and I'll review it. Once everything looks good we can merge it into the repo.