I was wondering if I can get an opinion on the approach that I have taken
on an spm between group analysis.
I have performed a BOLD-fMRI experiment of 7 social phobic subjects and 7
healthy controls who are "less anxious than most" making public
speeches in the scanner with the following cycle:
rest(1min)-->anticipation(1min)-->making a speech in front of an audience
--> thinking about how the speech went (1min) --> rest(1min)
repeated 4 times. Movement was not a problem on the non-speech scans.
One comparison I performed is anticipation scans minus rest scans for each
group separately at a height threshold of p=0.01 and spatial extent
threshold p=0.05.
The regions of activation for both group are extremely similar.
Thus, I wondered whether the intensity of activation differed in any
region. To answer this question, I found whether:
1: local maxima on the anticipation minus rest cluster map for the
social phobia group are also points of activation (not necessarily
local maxima) on the between-group difference map:
(social phobia group antic minus rest) minus (control group antic minus
rest)
[Note: The between group difference map was created by first making
minimally thresholded z-maps for the control and patient groups on the
anticipation minus rest comparison (ie. p =.49; cluster extent (k) =1).
I then performed the following calculation to make the unthresholded group
difference map:
((patient group antic minus rest) - (control group anticipation minus
rest)) / (square root of 2 to account for variance)
This between-group difference z-map was then thresholded at p=0.01;
spatialm extent(p)=0.05 to find the regions of activity on this
diference map.]
2: local maxima on the anticipation minus rest cluster map for the
control group are also points of activation (not necessarily
local maxima) on the between-group controls minus patients difference map.
Calculations are done similarly to #1.
Is there any problem with this approach??
Thanks
Jeff Lorberbaum
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|