Longitudinal Freesurfer with T2

Does anyone know what the correct recon-all usage is for longitudinal data that has both a hi-res T1w and a hi-res T2w image? I know I can use the T2 for pial refinement and the cross-sectional is easy to specify.

Being that the cross-sectional version creates both T2.mgz and T2.norm.mgz (and some T2 related images in transforms), is it sufficient after base creation to simply pass --T2pial or do I need to respecify the path to the original T2 image?

I found a very old conversation (2016) about this on the Freesurfer ListServ but nothing since then that seems to clearly answer the question.


With your updates to fMRIPrep’s longitudinal processing, any thoughts or insights?

I did such preprocessing recently, and relying on the freesurfer BIDS-app allowed me to abstract these. I don’t remember what is does from the top of my head ( I can check the logs later if needed), but maybe checking the logic for longitudinal options here freesurfer/run.py at master · bids-apps/freesurfer · GitHub will give you clues about what options are passed to recon-all for these stages of the longitudinal preprocessing.


I looked into the BIDS freesurfer app and found that it actually doesn’t do anything for the longitudinal data. T2s/FLAIRs are used in the first step (cross-sectional) but are not propagated downstream at all to the longitudinal step.

You are correct, I cannot see any T2 in the logs of the recon-all at base or long stages, so these are not picked-up automatically from cross-sectional outputs.
There are no examples on freesurfer wiki for longitudinal with T2/FLAIRs. I will try to find time and see if that can be passed at these stages.
Maybe open an issue on the BIDS-app repo.