Neuroimaging Workflow

This workflow currently covers structural and diffusion imaging analyses and is missing information about ASL analyses. As those workflows are established, they will be added to the workflow.

Beginning of Study / Initial Recruitment

These steps can take several months or more to pilot.

  1. Initialize manual that contains study specific scripts to be used for processing and analyzing data. The manual typically is 100+ pages of code and documented descriptions of all methods and procedures. Data and information will be uploaded to a private Github repository.
  2. Download DICOM directories and sort files
    • Verify the PatientName or PatientID matches across all DICOM headers and to the directory
      • Additional checks may involve acquiring scan dates
    • Count the number of slices across participants and make sure that each participant has the same number depending on scan protocol. It is often the case that there will be missing slices for T1 and T2 images and missing volumes for DWI images
    • Verify scanner manufacturer for DWI images (Siemens, Phillips, and GE have different procedures for analyses)
    • Validate the number of volumes / directions for DWI matches the bvec and bval files
    • Check that there is, in fact, a B0 for DWI
    • Validate the Phase Encoding direction for DWI (scan protocol names and Phase Encode directions don’t always match)
    • Validate the number of volumes for rsfMRI (if available)
  3. Convert DICOMs into NIFTI format and organize in the “Brain Imaging Structure” (BIDs)
    • For some studies, it will be necessary to preprocess images (e.g. topup for DWI images)
    • Visually check for too much head motion in T1 images
    • Visually check for too small FOV in DWI images
  4. Validate BIDS directory
  5. Complete brain analyses
    1. Preprocess T1 images
      • ACPC
      • N4 Bias field correction
      • Crop images
    2. Align T2 / FLAIR to T1 if available
    3. FreeSurfer v6.0.0
      • Amygdala / Hippocampal subregions
      • Brainstem subregions
      • Thalamus subregions
      • Run longitudinal pipeline if applicable
    4. FreeSurfer v7.1.0
      • Amygdala / Hippocampal subregions
      • Brainstem subregions
      • Thalamus subregions
      • Run longitudinal pipeline if applicable
    5. Qoala-T
    6. TOPUP / EDDY
    7. EDDY quality control
    8. Fit Tensors
    9. Run TBSS analysis
      • TBSS step 1
      • TBSS step 2
      • TBSS step 3
      • TBSS step 4
      • ENIGMA ROI analysis
      • Repeat steps 1-5 for non-FA images
    10. Preprocess DWI for tractography (VISTASOFT)
    11. Run AFQ analysis
      • Clean and render tractography fibers
    12. Extract DTI measures from FreeSurfer aparc
    13. MCFLIRT intra-modal motion correction for fMRI
    14. Jeff Anderson’s resting state analysis
  6. Complete additional analyses as requested
  7. Review results and steps with larger group and discuss any changes need to be made to the study

During Recruitment

These steps can take anywhere from 1-2 weeks depending on download and upload speeds and number of participants. In the case of larger studies, it can take several months (i.e., CENC).

  1. Update and maintain Github repository.
  2. Complete Steps 2 - 4 from above (Beginning of Study / Initial Recruitment)
  3. Establish regular communication with study coordinator
    • Reconcile BIDS organized data with the coordinator to confirm all participants have been processed
    • Confirm missing images / data with the coordinator
    • Brain analyses cannot be completed until the dataset can be reconciled with the recruitment list

Analyses REQUESTED During Recruitment

Brain analyses are run in parallel as a single batch (according to best practices) on a computing cluster. Depending on which analyses are needed, it will take ~4-6 weeks to complete. PLEASE ALLOW ENOUGH TIME TO COMPLETE ANALYSES. Note that these analyses are PRELIMINARY and values will change when participants are rerun at the end of recruitment. If images / figures are needed, please allow additional time.

End of Recruitment / Final Analyses

All participants are rerun as a single batch at the end of recruitment to ensure that all images receive the same treatment. It is important to use the most up-to-date version of each program and to run participants on computers with the exact same architecture and OS. Note that most neuroimaging programs are updated daily / weekly (i.e., adding features and fixing bugs) and so it is important to use the latest / most stable version. Analyses include those initially piloted at the beginning of study and any additional analyses requested or that are now avaliable because of new software / pipelines.

Additional steps for the final analyses include:

  1. Finalize manual that contains study specific scripts used to process and analyze data
  2. Upload all data and images to remote server for long term storage

Analyses REQUESTED After The Fact

If analyses are requested after a study has been completed and I was not invovled in data organization during recruitment, PLEASE ALLOW EXTRA TIME for data organization before analyses can be run. Note that this can be the longest step in all the processes because it is important to get everything organized properly into BIDs organization and to verify the accuracy of images (see steps above under DURING RECRUITMENT).