Tool used to filter data before generating synthetic motion in our Imaging Neuroscience article.
This repository can be used to train new models and replicate our study.
-
Clone the repository:
git clone <URL>
-
Set up a new Python environment using conda, venv, or any other tool (we used Python 3.11).
-
Install dependencies:
pip install -r requirements.txt
Before starting the server, you will need to: create a database, create a user, and add volumes to score. Fortunately, we provide CLI commands for all three tasks:
To create a new database, simply run:
flask --app motscore init-dbThen, create a user with:
flask --app motscore create-user --email <your_email>We use a simple user-code-based authentication system, as it is solely for rater identification. This tool is not intended to be deployed outside of local networks.
Finally, you can add your volumes from a BIDS dataset using:
flask --app motscore populate-volumes --dataset_path <Path_to_BIDS_root>Alternatively, to add multiple BIDS datasets at once, use:
flask --app motscore populate-volumes --multiple --dataset_path <Path_to_folder_containing_BIDS_roots>As this tool relies on Flask, you can run it using:
flask --app motscore runUpon arriving on the web interface, you will be asked for a user code. Once authenticated, you can start scoring.
After all volumes have been scored, you can export the labels as a CSV file using:
flask --app motscore export-csv --output <path_to_output.csv>For each command, use the --help argument to view available options, their purposes, and expected data types.
Once the repository is cloned, install the development dependencies with:
pip install -r dev_requirements.txtWe use:
pytestfor unit testspytest-covfor coverage reports (99% test coverage)
Run tests with:
pytest --covOther tools for code quality:
rufffor linting and formattingssort,pydocstyle,mypy, andpylintfor additional checks
All test data are extracted from MR-ART:
Nárai, Á., Hermann, P., Auer, T. et al. Movement-related artefacts (MR-ART) dataset of matched motion-corrupted and clean structural MRI brain scans. Sci Data 9, 630 (2022). https://doi.org/10.1038/s41597-022-01694-8
Charles Bricout.