thanks to all for your replies. I scripted things up and hope that someone could comment on the procedure. Even though the Jeanette's tutorial is quite clear, the values I am getting appear quite low.
(just a reminder: each subject had 3 runs of the same event-related WM task BEFORE, and 3 runs AFTER treatment)
the scaling factor:
the 1st contrast of interest is [onset task stimulus - onset control stimulus], so [1 -1]. In this case I don't need a contrast fix.
both stimuli are 10s long, so using the table 2x5s for gamma HRF is 1.2942
so the scale factor is 129.42.
To see if it's working, I extracted a mask from the feat stats AFTER - BEFORE for that contrast using a spheric mask around the highest local maximum
for each of the 3 runs per session for each participant I do:
fslmaths cope -mul scalefactor -div mean_func -mul mask percentage_image
fslstats percentage_image -M >> results.txt
I then average the percentages for the 3 runs for each subject, and pool these per "group" (ie before and after). This gives me an average DEactivation of 0.93% for BEFORE and an average activation of 0.27%, both relative to the control stimulus.
This contrast (AFTER - BEFORE) is significant in feat (so are the percentage differences I calculated as expected). Would this procedure be correct? I guess the only problem could be with the scaling factor...
Thanks for all comments.
Cheers,
Torsten
|