Hello SPMers,
I'm using Matlab6.5/SPM2 to work with a couple of fMRI data sets that we are
applying the following steps to: slice-time correction, realignment
(coregister & reslice), coregistration (anatomical with middle functional
volume), normalization, and smoothing.
For one of the data sets, I fear that some of the runs are being distorted
more than others because the length of the runs are determined by
participants' responses. For example, if run1 is 120 volumes and run2 is
150, isn't more likely that run2 has more overall slice-time correction and
realignment? Is this going to be a problem for the validity of the data?
For the other data set, my concern is about the stimulus types (modalities)
being different. Each run is 128 volumes but the first has the subject
listening to a siren wail, the second has a moving optical illusion, and for
the third, the subject has the task of tapping his or her index finger on
the scanner bed. We use behavioral data software for the purpose of
recording stimulus start and finish time, however all stimuli are triggered
automatically (therefore we record no responses from the Ss). My
predecessor would pre-process the three runs for a given individual
together, realigning to the middle volume of the middle run. Again, is our
output valid?
***********
A separate question: I understand that the output from slice-time
correction is basically temporally homogenized, and that between realignment
and coregistration, 3D space is mapped and matched from one volume to the
next, but is there a procedure (implicit to SPM2 or otherwise) that will
correction for motion within a volume? (a procedure that does what
slice-timing correction does but spatially, not temporally?) I think we may
already be applying one and I don't realize it...
Please advise & Many thanks in advance,
Felix Garcia
[log in to unmask]
|