All data in the OpenFMRI database are processed using a standard processing stream that is implemented on the Lonestar compute cluster at the Texas Advanced Computing Center. The processing steps include the following , with links to the code that implements each operation. To obtain the code mentioned below, see the OpenFMRI GitHub repository.

  • conversion to compressed NiFTi (.nii.gz) format and organization into the appropriate directory structure
  • removal of facial information from high-resolution anatomical image (code: deface.py)
  • motion correction using FSL mcflirt tool (code: run_mcflirt.py)
  • brain extraction for fMRI data using FSL bet tool (code: run_betfunc.py)
  • brain extraction for inplane anatomical images using FSL bet tool (code: bet_inplane.py)
  • convert high-resolution anatomy image into freesurfer format (code: fs_setup.py)
  • brain extraction and cortical reconstruction for high-resolution anatomical images using freesurfer (code: run_autorecon.py)
  • copy brain-extracted high-resolution anatomy back into main subject directory (code: run_copy_stripped.py)
  • creation of first-level design files and analysis using FSL feat (code: mk_level1_fsf.py, mk_all_level1_fsf.py, design_level1.stub)
    • Design details are identified from onset files in the model directories as well as key files in the main models directory
    • Motion parameters are automatically included in the model as nuisance regressors
  • creation of second-level design files and analysis using FSL feat (code: mk_level2_fsf.py, mk_all_level2_fsf.py, design_level2.stub)
  • creation of third-level design files and analysis using FSL feat (code: mk_level2_fsf.py, mk_all_level2_fsf.py, design_level3.stub)
The parameters used in the analyses can be determined from the stub files linked above for each level.
Many of these programs automatically submit the relevant job to the Lonestar grid, using launch_qsub.py.
Currently, several manual quality control checks are implemented:
  • checking the quality of the brain extraction for the high-resolution anatomy image (code to create report: check_brainmask.py)
  • checking first-level feat directories for proper completion of the analysis (code: check_featdir.py, check_all_featdirs.py)
  • checking quality of standard-space registration for high-resolution anatomy and functional image
  • comparison of the third-level results with the original aper (code to create report: get_group_report.py)