Alignment of different sessions


We’re planning to use RSA and we have acquired our data over multiple sessions (3) so we were checking the output of fmriprep (1.4.1) to make sure that all the runs and sessions are well aligned to each other. This is crucial for our analysis because we want to compare activity profiles over (the same) voxels between sessions.

When overlaying runs from different sessions and navigating from one to the next, we seem to see a small shift in the voxels, especially in temporal and frontal areas that are part of our ROIs (see attachment: ). Could this be due to the distortion correction yielding different solutions for the different sessions, or can this have a different reason?

May there be ways to make the output space more similar across all sessions for a participant?
One option may be to first re-align runs to each other and only then coregister them to the T1w image. Could that help (and could that be implemented in fmriprep)?

For some background: We are using an inverse scan for susceptibility distortion correction and asking for the output in native space.

Thanks a lot for any help!