Hi all,
I've just begun looking at doing PPI in SPM2 and ran into a few questions.
1.) Our current data has 8 sessions, and there seem to be two different
suggestions on the list archives. The first that's been repeated several
times is to model the 8 sessions as one large session and do a PPI and
analysis on that. The second that I've seen mentioned once is to just do a
PPI on each session individually and model the second analysis as an 8
session model. Is one of these to be preferred over the other? Right now
I'm just modelling it as an 8 session analysis, because that made more
sense to me. Couldn't the first method lead to problems if the runs scaled
differently?
2.) When I tried getting the PPI results from our multi-session SPM.mat
originally the PPI.Y variable was filled with NaN. I found this post
http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind04&L=spm&P=R351544&D=0&I=-1
that suggested a workaround by replacing the inv function in line 204 of
spm_peb_ppi.m with pinv. This works in that it gives me values for PPI.Y,
but another post I was reading here
http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind04&L=spm&D=0&I=-1&P=83665
suggests using the Y or xY.u from the VOI or PPI.xY.u variable in place of
PPI.Y, saying they should all be identical. This is not the case with the
pinv solution. Should I just use the Y values or are they not identical to
the PPI.Y?
3.) With my 8 session model, a T contrast of [1 0 0 1 0 0 1 0 0 1 0 0 1 0
0 1 0 0 1 0 0 1 0 0 0 0 0 0 0 0 0 0] would be correct to give me the
positively covariant regions correct? (assuming my model was 3 regressors,
PPI.ppi, PPI.P, PPI.Y)
Thanks in advance for any answers.
-Mike
|