fsaverage_sym as possible output space?

[spitschan]

How difficult would it be to add fsaverage_sym as a possible output space? fsaverage_sym is the hemisphere-symmetric atlas, and it would require an additional step in recon-all, namely:

# Invert the right hemisphere
xhemireg --s bert
# Register the left hemisphere to fsaverage_sym
surfreg --s bert --t fsaverage_sym --lh
# Register the inverted right hemisphere to fsaverage_sym
surfreg --s bert --t fsaverage_sym --lh --xhemi

(as described in https://github.com/scitran-apps/retinotopy-templates)

[effigies]

Probably not too difficult. It would require:

1. Add fsaveragesym (note lack of underscore; this is to fit BIDS naming convention) as an option to --output-space
2. Wrapping those three steps into a small workflow, creating nipype interfaces for xhemireg and surfreg, if they don't already exist. This would conditionally follow recon-all.
3. Adjust the mri_vol2surf step as needed to correctly resample to the template.

Anything I'm missing? And are you interested in submitting a PR?

[effigies]

I ran the above lines on the bert subject distributed with FreeSurfer 6.0. At least without any multithreading turned on, the second and third steps each took >1 hour, so we would definitely not be turning this on by default, and should make a fairly prominent warning about the increased runtime to potential users.

The xhemireg and surfreg tools do not currently exist in nipype, so they would need to be added, and multithreading should be looked into.

[spitschan]

I'm a little bit out of my depth trying to add this in nipype but I can give it a shot. This would be implemented in nipype/interfaces/freesurfer/utils.py?

[effigies]

I might go in registration instead of utils, but to be honest, the organizational scheme is somewhat arbitrary.