Group ICA Script in Python for Preprocessed Data

This is part of the fMRI Preprocessing for Causal Modeling Project. The goal of this is to perform the same task as aa-gift but without using GIFT and using ICA-Torch instead.

Container Files

The container_data directory contains work toward using GIFT software to run ICA in a singularity container.

These files might be adapted to later accomdate the standalone group ica python script.

Standalone Group ICA Script

The standalone_group_ica_script directory contains a script that is intended to run group ICA on a study's set of preprocessed fMRI files.

The directory also contains a subdirectory called input_data which contains a script for generating an input string of an entire study's processed fMRI filepaths.

Each filepath is the full path to each subject's processed fMRI data with each subject's fMRI file separated by a string.

The group_ica_script.py Python Script

This script operates on a study's fMRI data and produces N spatial maps, where N is the number of subjects.

Each spatial map is saved as ${subjectID}_SM.nii.gz where ${subjectID} should be replaced by the subject's unique identifier. Each spatial map file contains K volumes where K is the number of components estimated by ICA. Each volume in the spatial map file represents an activation network that was detected via ICA.