I have a FEAT analysis that uses timeseries as the inputs: one for "timecourse of interest", and three more for "noise timecourses." Each timecourse is its own EV, and in the contrasts the noise-timecourses all have zero weight (i.e. [1 0 0 0]), leaving me with a map of areas that activate with the interest-timecourse while controlling for noise-timecourses. So far, so good! The whole-brain results make sense.
However, I'm having trouble figuring out how to do a ROI analysis of this data. I'd like to ask "In this ROI, what is the noise-controlled timecourse?" In other words, I'd like to see the timecourses of the output of the FEAT analysis.
I thought I could get this from running Featquery on the COPEs, but that turns out to be incorrect: the COPE shows the exact same timecourse as the PE. (i.e., max_cope1_ts.txt and max_pe1_ts.txt are identical.) Whereas I hoped the PE would give me the EV's timecourse, and COPE the full contrast's timecourse.
I don't know whether I'm missing something basic, or I'm making a more fundamental/complicated error. Can anyone help me figure out how to get the timecourse of the output of a FEAT analysis? Or if I want to see "one timecourse controlling for another" do I have to take those PE timecourses and go do multiple regression in Matlab?
Thanks,
-Benjamin Philip
|