To run:

Requirements:

FreeSurfer/6.0.0 python/3.6+

Python packages:

nibabel numpy numba scipy https://www.github.com/satra/mapalign

Processing steps:

1. Project volume to surface using FreeSurfer:

Currently set to project from MNI space, but could also be modified for native spaces.

#!/bin/bash

while read subject;
do
  ./x.mri_vol2surf.sh ${subject}
done <subject_list.txt

2. Correlate RS data across sessions and embed

This command will call load_fs.py to read in the data. This is the most time consuming step, but the script has some fancy numpy optimizations to speed it up dramatically.

#!/bin/bash

while read subject;
do
  python pipeline.py ${subject}
done <subject_list.txt

3. Rotate embeddings across individuals to match

Need to set the path to the subject_list.txt file.

python combine_subjects.py

A single file with all individual embeddings is created that can be read into another software for group-level analyses.