z-fabian / lowfieldsim_py

Python implementation of usc-mrel/lowfieldsim low field MRI simulator with sample code on the fastMRI dataset.

Python Code for Low SNR Simulation at Low Field Strengths

This is a direct Python implementation of the Low Field Simulator from the Magnetic Resonance Engineering Laboratory at USC.

For example usage take a look at fastMRI_to_lowfield.py that converts the fastMRI knee dataset (1.5T/3.0T) to simulated low field data.

The fastMRI dataset can be downloaded here.

Requirements

Low Field Sim Py:

• python >= 3.7.6
• numpy >= 1.19.2

fastMRI to low field:

• h5py >= 2.8.0
• xmltodict > 0.12.0

Running the script

To simulate low field noise on fastMRI data, set INPUT_PATH to the folder containing .h5 files of the dataset, OUTPUT_PATH to an empty folder where the output will be saved and B_LOW to the desired low field strength. Then run

python3 fastMRI_to_lowfield.py --input-path INPUT_PATH --output-path OUTPUT_PATH --B-low B_LOW