Defacing

The first step before releasing the data is to deface the T1w and T2w images for all sessions. To perform defacing, we are using a software called PyDeface (Gulban et al. 2019). To proceed, run the following command in the command line:

    bash ./code/defacing/run_pydeface.sh

Release checklist

1. Verify defacing results
2. Open a subset of T1w and T2w images with your preferred viewer (e.g., FSLeyes).
3. Confirm that all facial features were removed while brain tissue remains intact.
4. Run quality checks
5. Execute MRIQC on all sessions as described in the MRIQC guidelines.
6. Inspect individual and group reports, addressing any issues before continuing.
7. Package and version the dataset
8. Save new files with DataLad:

   datalad save -r -m "add: session <session_id>" sub-001/ses-<session_id>
   

9. Tag the dataset with a new version number:

   git tag -a "vX.Y.Z" -m "HCPh release X.Y.Z"
   

10. Upload the release
11. Push data and tags to the remotes:

    datalad push --to ria-storage
    datalad push --to origin
    git push origin --tags
    

12. Publish the new version on GitHub and upload the dataset to the designated repository (e.g., OSF or Zenodo).
13. Update documentation
14. Record the release in changes.md.

References

[1]: Gulban, Omer Faruk, Dylan Nielson, Russ Poldrack, John Lee, Chris Gorgolewski, Vanessa Sochat, and Satrajit Ghosh. 2019. “Poldracklab/Pydeface: V2.0.0.” Zenodo. doi: 10.5281/zenodo.3524401.