Required packages:

• Nibabel,
• Nilearn,
• Scikit Learn,
• validators.

usage: app.py [-h] [-p PATH] [-a ATLAS] [-r REGIONS] [-d DOWNLOADPATH] [-k KIND] [-f FILTER] -s SUBJECTS

Currently supported connectivity measures:

• correlation,
• partial correlation,
• tangent,
• covariance,
• precision,
• power predictive score.

Computes connectivity matrices of fmris.

arguments:
  -h, --help            show this help message and exit
  -p PATH, --path PATH  Path to a folder containing fMRI
  -a ATLAS, --atlas ATLAS
                        URL or Local Path to an atlas
  -r REGIONS, --regions REGIONS
                        Local Path to a list of brain regions
  -d DOWNLOADPATH, --downloadpath DOWNLOADPATH
                        Path with filename for the downloaded atlas. Default:
                        ./downloaded_atlas.nii.gz
  -k KIND, --kind KIND  Comma separated list of nilearn's kinds (e.g.:
                        "partial correlation,correlation,tangent,covariance,pr
                        ecision"). Default: "correlation,partial correlation"
  -f FILTER, --filter FILTER
                        Regex filter to select fMRI. Default:
                        "**/*bandpassed*.nii.gz"
  -s SUBJECTS, --subjects SUBJECTS
                        <Required> Path to a csv file or a list of csv files,
                        with IDs in the first column. Every fmri should have
                        its ID in its relative path

Example:

python app.py -p './data/BIDS/' -a 'https://my_url.com/msdl_rois.nii' -d './msdl_atlas.nii' -k 'partial correlation' -f '**/*bandpassed*.nii.gz' -s './subjects1.csv,subjects2.csv'

The above command will compute a partial correlation matrix for each .nii.gz file containing the word bandpassed, in the folder ./data/BIDS/, including subfolders, using an atlas downloaded from https://my_url.com/msdl_rois.nii to ./msdl_atlas.nii.

Note that -a supports both local path and valid url.

If you provide an URL, it is good practice to define a download path using -d, to ensure the online atlas and the downloaded one have the same extension (.e.g.: .nii, .nii.gz, etc.) Default download location: ./downloaded_atlas.nii.gz.

When no atlas is specified, it defaults to MSDL.

Each path should contain an unique identifier (in a folder or file name) for a given subject. Identifiers should be reported in the first column of ./subjects1.csv and subjects2.csv or any other specified csv file by -s. The script also supports list of csv files.

Matrices will be saved as values in a Dict, with IDs as keys, in a pickle file saved in <path>/{time}_{n}subjects_[kinds].pkl.