Within-subject alignment across 3 sessions

I’m currently working on a longitudinal analysis of fMRI data across three timepoints and am trying to puzzle out how to do within-subject alignment of the three structural images (and their corresponding functional images) in an unbiased fashion- i.e., not just align timepoints 2 and 3 with timepoint 1 but align all three timepoints to a more ‘neutral’ position so that all three datasets are being manipulated to a similar extent. In the past, I’ve followed Daniel Glen’s recommendations for how to do this across two timepoints (https://sscc.nimh.nih.gov/sscc/dglen/alignmentacross2sessions) but it’s not obvious to me how to scale this up for three (or more) timepoints without biasing toward one of the timepoints.

In a pinch, I know that I could just transform each timepoint into standard space separately, but we are planning on doing our analyses in native space and as these participants are older stroke patients, transformation to a template is not always straightforward (or desirable), so I’d like to avoid that if possible.

Any thoughts or suggestions from the AFNI experts?


  • Kate

You could just use the same in-between S1 and S2 time point, or you could add on to that to calculate a new in-between S2-S3 anatomical. Then compute the in-between of the in-betweens. There is no existing script for this, but it could be fairly easy to extend the two-session script to that. If you have many sessions, then you could generate a new template for that using the @_to_MNI_Qwarpar script. A lot of this depends on how far you want to go in terms of complexity. Most just use an initial session and settle on that.

Daniel’s suggestion is probably easier and likely about the same end result. I might suggest you use Freesurfer[/url] to build an unbiased anatomical template using their [url=https://surfer.nmr.mgh.harvard.edu/fswiki/LongitudinalProcessing]Longitudinal pipeline. Then using their version of an unbiased brain, I would run @SUMA_Make_Spec_FS to generate a NIFTI version of brain.mgz (now brain.nii) and use that as your skull-stripped anatomical for future processing in AFNI.

Happy to try and break this down further!