Code for the article "Comprehensive decoding mental processes from Web repositories of functional brain images", published in Nature Scientific Reports 2022.
- Setup a virtual environment (conda, virtualenv, pyenv, ..) and install requirements
virtualenv venv -p python3.6 . venv/bin/activate pip install -r requirements.txt
- If you have Neurovault already downloaded on your machine, create a symlink in the
Data/
folderNote: For Parietal users, Neurovault is already downloaded atsudo ln -s /path/to/already/downloaded/neurovault Data/neurovault
/storage/store2/data/neurovault
on Drago. - Add folder path to the PYTHONPATH
export PYTHONPATH="$(pwd):$PYTHONPATH"
All scripts to perform the experiments are stored in the Experiments/
folder. In particular:
- the
a_preparation_pipeline
script performs all preprocessing for images and labels - the
b_decoding_experiment
script performs training (from the output of the previous scripts) - the
c_plot_perf_and_models
generate some figures of the paper
For explanations on each argument, you can call python {a_preparation_pipeline.py|b_decoding_experiment.py|c_plot_perf_and_models.py} --help