Dear FSL experts,
I have been analyzing data from an event-related design experiment.
Recently I came across a section in the FSL book that said a gamma
function for the HRF is not optimal for event-related designs. I get
activation for individual runs, but when I analyze across runs for 1
subject the activation is little to none. And therefore across subjects,
there is no activation. I think the problem is that I'm only using 5
subjects in this analysis and they have activations in different places
for each contrast, generally. I tried lowering the threshold to z=2.0
and p=0.05 and it helped but not a whole lot. Would it help to use a
different HRF (such as Gaussian) or should I just assume that activation
is very weak?
fyi - Each trial consists of 2 seconds of looking at a question mark, 4
seconds of stimulus, 3 seconds to respond, and 2 seconds of fixation.
There are 37 trials in each run.
Thanks very much!
Bo
|