Documentation for active-mri-acquisition¶
active-mri-acquisition is a package that facilitates the application of reinforcement learning
to the problem active MRI acquisition. In particular, active-mri-acquisition provides a gym-like
environment for simulating the execution of policies for k-space sampling, allowing users to
experiment with their own reconstruction models and RL algorithms, without worrying about
implementing the core k-space acquisition logic.
Getting started¶
Installation¶
active-mri-acquisition is a Python 3.7+ library. To install it, clone the repository,
git clone https://github.com/facebookresearch/active-mri-acquisition.git
then run
cd active-mri-acquisition
pip install -e .
If you also want the developer tools for contributing, run
pip install -e ".[dev]"
Finally, make sure your Python environment has PyTorch (>= 1.6) installed with the appropriate CUDA configuration for your system.
To test your installation, run
python -m pytest tests/core
Global configuration¶
The first time you try to run any of our RL environments (for example, see our intro notebook), you will see a message asking you to add some entries to the defaults.json file. This file will be created automatically the first time you run an environemnt, and it will be located at $HOME/.activemri/defaults.json. It will look like this:
{
"data_location": "",
"saved_models_dir": ""
}
To run the RL environments, you need to fill these two entries. Entry data_location must point
to the root folder in which you will store the fastMRI dataset (for instructions on how to download
the dataset, please visit https://fastmri.med.nyu.edu/). Entry saved_models_dir indicates the
folder where the environment will look for the checkpoints of reconstruction models. Note that
saved_models_dir does not need to be set to use your own reconstruction model, but it is
required to use our example environments. For more details see Environment’s JSON configuration.