Dear sir/madam,
I am quite new to fMRI analysis as well as to SPM, and I am concerned my preprocessing might have gone wrong.
I am using SPM12. After realigning the volumes from different time points per subject and generating a mean image, I intended to coregister the mean image as well as all separate time points to the T1 structural image. Now, in my first run, there was a minor
notation error in my 'other images' specification (using scripts, not the GUI), however, the programme did not kill the run but simply skipped 'other images' and only coregistered the mean image to T1. I found out halfway, killed the process, fixed the typo
and restarted it, but now I am worried this made it even worse. After the first (faulty) run, the mean images were reoriented and overwritten, correct? SPM seems to assume that the 'other images' simply require the same reorientation as the 'source image',
but this does not work anymore if the 'source image' has been reoriented in a previous run already.. So my question is, does the coregister estimate function also look at the relative orientation of 'source' and 'other' images, or does it simply determine
one reorientation matrix for 'source' and imposes this onto all 'other'? If the latter is the case, is there a fix or do I redo the preprocessing all the way from before the realignment/mean step? (I found that in my test-run, the reorientation parameters
in the file headers were identical for mean and other images, whereas they were not for the runs that had been stopped and restarted).
In general, I wonder to what extent SPM overwrites files 'irreversibly' in realignment and coregistration operations?
Thanks in advance!
Iris
