Dear FSL experts,
I have 12 subjects to perform 3 types of cognitive tasks (A,B,C) in an fMRI run, and subjects were asked to rate difficulty levels of each trial. My aim is to compare activations between the 3 tasks. However, I noticed that the difficulty levels were statistically different among the 3 tasks. I want to remove the effects of difficulty so as to clarify activation differences. My 1st-level model has set 6 contrasts (A,B,C, A-B, B-C, A-C). My 2nd-level model is actually group-level analysis by performing 1-sample t-test of these contrasts.
How can I remove the confounding effects of difficulty (in my case, each subject have 3 difficulty scores for tasks A, B and C)? My idea is to remove it at group-level analysis. For example, if I want to see group map for A, I can add a new EV representing each subject's difficulty level for task A in a new group-level analysis. If I want to see group map for A-B, in another group-level analysis, I can add a new EV representing each subject's difference in difficulty level between tasks A and B (i.e., A-B). So if I want to get all 6 map, I have to to additional 6 group-level analyses. Am I correct? Also I wonder if it's possible to remove effects of difficulty in 1st-level model?
Thanks in advance. Mike
|