st: what type of analysis for this fMRI issue

[Richard Goldstein <richgold@ix.netcom.com>]

Dear all:

I am working on my first functional magnetic resonance imaging project.  An 
fMRI is being performed on each of about 50 traumatically brain injured (TBI) 
subjects.  A number of analyses are planned and I am not clear regarding one.

For this analysis, the outcome is a score for the person on a memory test 
after they have had some memory training.  The two prime predictors are the 
pretest on the same test and percent-fMRI-signal-change.  The complication is 
that the percent-fMRI-signal-change is measured for each of, say, 30,000 
voxels in the brain -- clearly, the test scores do not change for each voxel. 
 Thus, we have

	posttest=f(pretest, %change)

for each of 30,000 voxels.  Clearly these 30,000 models are related to each 
other (since the posttest and pretest do not change).

My question: is there a "correct" model for this situation and, if yes, what 
is it and what command covers it?

Thanks,

Rich

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/