I have two runs of resting state data that I want to concatenate so that I can create correlation matrices. As I understand it, the best practice is to register the functional runs to each other and then register them to the T1 (and demean, etc.). However, I do not see an option in fmriprep to register functional runs to one another. Is this possible? Or if not, is it necessary?
My understanding is that your outputs should already be in the same space (whether that’s MNI as default or some other space you specify). If you want them to be aligned to the T1 image you can specify
anat in your output spaces. If you are not convinced the images are aligned then you can also align one run to the other by first calculating a rigid transform between the two bold reference images and then applying the transform to which ever run corresponded to the moving image in the calculation step. You can use FLIRT in FSL or antsRegistration/antsApplyTransforms in ANTs. But as fMRIPrep outputs each run in the spaces specified, there shouldn’t be a need for that (unless you only used the
func output space which keeps everything in the space it was collected in). Does that help?
That does help. Thank you. We were thinking about registering the resting state scans to the same space because we have another non-fmriprep-processed dataset and even though the functionals are both in MNI space, there was enough difference between the two runs that the correlation matrices looked funny. But maybe the registration from fmriprep is better so it’s less of an issue.