SPM gurus,
I am struggling with the following design problem, and couldn't find a
clear answer in the database: I have two groups in the same 4-condition PET
experiment (one scan/condition). Given this, I cannot compute *con.img's to
take to a second level analysis (comparing the two groups). That is, either
a "single subject" or "multisubject condition x subject interaction" design
makes an unestimable design with 1 scan/condition.
How, then, does one easily obtain contrast images to apply to a 2-group
t-test at the second level? The only way I can think of is extract the
globals, and then using the image calculator, manually scale/normalize to 50
and do each condition subtraction image on a subject-by-subject basis.
These manually created subtraction images would then be tested at the second
level. So, for each subject (i1/Global)*50 - (i2/Global)*50.
1. Is this a valid analysis?
2. Is there a more automated way of doing this (its a lot of typing, with
lots of room for error; I don't know much about batch processing in SPM)?
3. By tracking the same voxel I think that I've determined that the mean
image calculator will properly proportionately scale just one image- getting
rid of the need to extract the globals (correct?). But, that still leaves a
lot of manual subtractions.
4. Other suggestions?
Thanks much.
________________________________
David A. Kareken, Ph.D.
Assistant Professor & Director
Neuropsychology Section, Department of Neurology
RI 1773
Indiana University School of Medicine
Indianapolis, IN 46202
(317) 274-7327 (voice)
(317) 274-1337 (fax)
|