One way to do it is, as you suggest, just put all the runs into the same bold directory, then just use mc as registration. This works to some extent. The main problem with it is that you can get fairly different B0 distortion between sessions because of the different head/body position in the scanner, and this can cause the mc algorithm to moderately misalign the sessions. You can try it out and see how well it works using tkregister2 (can't fixed it with tkregister2 though).
The other thing you can do is to analyze each session separately, then combine them together in talairach space with a fixed effects analysis (isxavg-fe-sess), assuming you've done the first couple of freesurfer analysis steps.
is it necessary/helpful to record low-res anatomicals before
functional runs
I don't find it helpful for registration purposes, though you do need good anatomicals at some point.
doug
Sebastian Moeller wrote:
Dear fs-fast wizards,
doing monkey fMRI in Bremen (block desing so far) I run into the problem that analysing the data from one session does not lead to well defined activation maps. Therefore I would like to pool over several sessions (always within one subject). Unfortunately I have did not manage to always position the slices at the same position, I recorded a different number of runs each session. Now my assumption is all I need to do is define according df and sf files and be done with. But at what stage (mc-sess, spatialsmooth-sess, mkanalysis-sess.new, selxavg-sess, mkcontrast-sess or stxgrinfer-sess)? Is it advisable to motion correct all session runs together, or combine the data after that step? So excuse me for my uneducated question, but I would be delighted to tap into this list's vast knowledge before debarking into computationally long experiments. One more question, is it necessary/helpful to record low-res anatomicals before functional runs to register separate sessions, or is this step avoidable.
Regards Sebastian