Dear FSL-experts,
My question is regarding PPI-analysis.
I want to show that my ROI interacts with some areas more in Task A than in Task B. How to do create such a PSY (A-B) x PHSY (timecourse of ROI)interaction is very well explained in Jill O'Reilly's pages: http://www.fmrib.ox.ac.uk/Members/joreilly/frequently-asked-questions-ppi
What I do not really understand is how to design the contrast & f-tests in my GLM for PPI.
I have a(n original) design with 8 EVs and many many contrasts for everything such as A>B, A>C, A>C+D, C>B...
For the PPI, I need to model A-B (PSY) and A+B, timecourse of ROI (PHSY) and the PPI (PSY*PHSY). I should not model separately for task A or B. So in my PPI design I would have 10 EVs and I should also include all other task regressors to my PPI-GLM. Am I right?
EV1 EV2 EV3 EV4 EV5 EV6 EV7 EV8 EV9 EV10
A-B 1 0 0 0 0 0 0 0 0 0
A+B 0 1 0 0 0 0 0 0 0 0
PHSY 0 0 1 0 0 0 0 0 0 0
PPI 0 0 0 1 0 0 0 0 0 0
Are these four contrasts enough for further analysis, if I am only interested in the areas, which interact with my ROi? Is it meaningful (or common) to have contrasts such as PPI +1, EV6 -1 ?
I hope it is not a very silly question, but I am new to PPI and your help is much appreciated!
Best,
Elif
|