Dear Ella and Cyril!
Thanks a lot for your response. The motivation for motion correction is clear. I like to realign two fMRI sessions to a volume, which lies in the middle of the two series. You may call it half through coregistration. If I follow the path Cyril pointed out (using the standard spm realign batch I end up with a coregistration (or realignment) to the first volume of the first session or the (mean) of the first session. I have attached an pdf to clarify it: Upper row: The images on the left and right symbolize fMRI-session 1 and 2 and are tilted by -0.2 and 0.2 relative to the middle one. The lower row shows the result after the running the realign batch (1 subject, two sessions). Unfortunately, the result was not the intended one.
What I like to do is to set up a batch that realign (and coregister) to the middle one (in space) ... not onto the first series or the second series.
All the best,
Luke(See attached file: IntersessionAlignment.pdf)