Dear list,
I am in the process of analyzing a large event-related fMRI study. As previously
posted and discussed, the amount of data is too large (11.700 volumes) to be
processed using spm99b in its current state. What I am trying to do now is to
use a two-level analysis as recommended by Karl and I would appreciate some
advise on how to do this.
I have three event types presented pseudorandomly, modeled with the synthetic
hemodynamic response function and its derivative. I have about 20 events of each
type per scan run of 4 minutes. I want to analyze five scan runs of this length
per subject at this time. My main interest is examining activation decreases
over time (priming experiment) and comparing this effect between different event
types.
1) Is it appropriate to produce contrast images reflecting the evoked activity
for every event and scan run (resulting in 3(number of events)*5(scan run)
contrast images per subject), then take these contrast images to the second
level and look for linear decreases using an appropriate contrast (like 3 1 0 -1
3 or something like that)? Or should I produce one contrast image reflecting the
linear decrease per subject? I would prefer the former since I would like to
allow myself some freedom at how to model the decrease (which might not be
linear) and I would like to try several contrasts without having to go back to
the first level each time.
2) Once I have the contrast images, I assume that neither global normalization
nor grand mean scaling makes sense. However, it confuses me that, after
responding 'no' to both GloNorm and GMsc, I still have to make a choice about
thresholding. How should I respond to this question?
3) I would like to restrict the analysis of the decrease (priming) effect to
those voxels that show a main effect of event presentation vs. baseline. Could
you advise me on how to do that?
Sorry for being wordy,
and thanks in advance,
Andreas
Andreas Meyer-Lindenberg, M.D.
Unit on Integrative Neuroimaging, Clinical Brain Disorders Branch, National
Institute of Mental Health
Bldg. 10 Room 4C101, 9000 Rockville Pike, Bethesda, MD 20892-1365, USA
email: [log in to unmask], phone: 301-496-9672, fax: 301-496-7437
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|