Have you considered using a finite impulse response (FIR) model to analyze your event-related data? Such a model would allow you to estimate the average stimulus-locked BOLD response to each of your trial types across time. Comparisons of these response shapes for your 2 subject groups might then be more easily (i.e., visually) interpretable.
Hope this helps,
Daniel Weissman, Ph.D.
Department of Psychology
University of Michigan
1012 East Hall
530 Church Street
Ann Arbor, MI 48109
From: SPM (Statistical Parametric Mapping) [[log in to unmask]] On Behalf Of Liliana Demenescu [[log in to unmask]]
Sent: Friday, March 27, 2009 11:22 AM
To: [log in to unmask]
Subject: [SPM] Hemodynamic response shape testing.
I have an event-related design. The 1st level included the canonical HRF and its time and dispersion derivatives, because we want to look if there are difference in response shape between 2 groups. At the 1st level analysis, I defined a t-contrast for each basis function (HRF, TD and DD). The 2nd level analysis employed an ANOVA 3 (basis functions) x 2 (groups). I did an F test defined as:
1 -1 0 0 0 0
0 0 1 -1 0 0
0 0 0 0 1 -1
(group1 - group2: HRF; group1 - group2: time der. and group1 - group2: dispersion der.). This tells me which voxels have different activity between the 2 groups with respect to either amplitude, either time, or dispersion. I plotted the results, but I found it difficult to understand/explain, as it's an F test. Moreover, as I understood from Friston's paper (Friston et al. 1998, Neuroimage, 7:30-40) this tells me how well the model fits or have I misunderstood the things?
My question is how can I look for differences in response shapes between 2 groups combining the 3 basis functions?
Any suggestion is more than welcome.