preprocessing_parallelized.py

Purpose

This script executes the fMRI preprocessing workflow. It is designed to facilitate preprocessing and quality control for fMRI data analysis using interleaved acquisitions from a ten minute run with 750 volumes.

Functions

• transform_fmri_to_standard: Aligns fMRI data to a standard space - the first volume - for each subject.

  • Also removes the first 10 scans and corrects for field inhomogeneity (e.g. correcting for the anterior -> posterior and p -> a transformations).
  • Uses get_fmri2standard_wf (from bold2T1_wf.py), which constructs the workflow for transforming fMRI to standard space, including steps for field inhomogeneity correction.

• execute_coregistration: Coregisters BOLD images to T1-weighted images using SPM.

• extract_wm_csf_masks: Extracts white matter and cerebrospinal fluid masks, needed for nuisance correction.

• run_nuisance_regression: Sets up the workflow to remove nuisance signals using the predefined masks.

  • Uses:
    • get_nuisance_regressors_wf (from bold_nuisance_correction.py): Creates the workflow for nuisance correction.
    • motion_regressors (from bold_nuisance_correction.py): Computes motion regressors based on realignment parameters, aiding nuisance regression.
    • cosine_filter_txt (from bold_nuisance_correction.py): Creates a discrete cosine transform (DCT) basis for modeling low-frequency drifts in the data.
    • merge_nuisance_regressors (from bold_nuisance_correction.py): Combines multiple nuisance regressors into a single matrix for filtering.

• mni_normalization: Performs MNI normalization on T1 and fMRI images, aligning them to the standardized brain atlas.

• apply_nuisance_correction: Applies processing steps to regress out and remove signal originating from movement, white matter and cerebrospinal fluid.

  • Also runs a band-pass filter that removes all signal that are below the threshold for being considered a BOLD signal response.

• fmri_quality_control: Conducts framewise displacement calculations and DVARS computation for fMRI quality assessment.

• prepare_and_copy_preprocessed_data: Prepares necessary directories and copies preprocessed data for each subject to the designated output location.

• initialize_preprocessing_dirs: Sets up initial directories and retrieves a list of subjects to process.

• change_logger_file: Configures the logging settings for a each processing step.

• main: Initializes paths and directs the execution flow, conducting the preprocessing for all subjects.

Additional Functions

• get_fmri2standard_wf: Constructs the workflow for transforming fMRI to standard space, including steps for field inhomogeneity correction.

• motion_regressors: Computes motion regressors based on realignment parameters, aiding nuisance regression.

• cosine_filter_txt: Creates a discrete cosine transform (DCT) basis for modeling low-frequency drifts in the data.

• merge_nuisance_regressors: Combines multiple nuisance regressors into a single matrix for filtering.

Notes

• This pipeline does not use global signal regression or smoothing.
• The script supports parallel processing for efficiency.
• Dependencies include Nipype, SPM, and FSL.
• After running this script, to delete intermediate files that are no longer needed, run erase_intermetiate_files.py.