Spatially localized fMRI metrics as predictive and highly distinct state-independent fingerprints

This repository contains code to generate fingerprints and perform identification. The details are described in the paper "Spatially localized fMRI metrics as predictive and highly distinct state-independent fingerprints". The preprint is available on biorxiv.

Code

Please run the setup.sh file which will get the DataLad dataset and spm12. Note that you need datalad available in your path.

The main analysis code is available in the code folder. We also provide two convenience files calculate_fingerprints.m and calculate_identification.m. The calculate_fingerprints.m will calculate fingerprints using the fMRIPrep derivates of ds000115 dataset. The script will download the DataLad dataset and use corresponding files to calculate several fingerprints as described in the paper. For demonstration only 10 subjects are used but this can be changed in the code. The output will be stored in the output folder. The calculate_identification.m script can be then used to calculate identification accuracies from those fingerprints. For convenience we also provide mask and atlas files in the correct space.

This code has only been tested on a Linux machine.