Dear SPMers
I have a question regarding the analysis of event-related fMRI data.
In our experiment we aquire single fMRI-images that are related to
events of intrest by using the external trigger function
of the scanner. We have two kinds of stimuli that have to
be contrasted. Each trial consist of the relevant stimulus, the
single fMRI-volume starting 4 sec after the stimulus presentation
(16slices, TR=1.6sec) and an additional "silent" period that results
in a mean interscan interval of about 12 seconds. Due to methodological
considerations, the interval between two scans varies and the
presentation of stimulus conditions is randomized over trials.
We preprocess our data with realignment, normalization
and smoothing.
My question is whether it is appropriate to apply the SPM99
basic model ANOVA for the statistical inference in single subjects
and the .con images for a second level group analyis. Our suggestion
is, that the use of basis functions in the analysis of our data results in
signal distortion which may be due to the fact that we see only a small
timespot within the bold response that might not be modelled adequately
within "fMRI models" option in SPM.
Any comments are appreciated
Bernhard Mueller
|