 |
 |
Dear FSL experts, I am analyzing fMRI data acquire while subjects play a game (event-related task). Due to the complexity of the game, it is impossible to record the time seed of game events, therefore, I decide to use ICA instead of GLM. I have 4 groups of subject: control group and three training strategy groups (called trained 1, 2 and 3). Each subject has two MRI sessions (pre and post training). There are 40 subject each groups, which lead to 40 (subjs) x 4 (groups) x 2 (times) = 320 scans total. I want to see if there is any difference in functional connectivity (and later on construct a Bayesian graphical model) before and after training. I also want to see if there the functional connection varies between groups. Therefore, is using concat ICA followed by dual_regression and FSLNets a good way to conduct the analysis ? If so, should I run 4 analyses separably for each group (each analysis is a concat ICA follow by dual regression with design contrast seek for the difference between pre and post session, and FLSNets) then compare the output of FSLNets to find if there is any difference between groups. Or I should run concat ICA and dual regression on the whole data (320 scans total). I believe it will takes a great deal of time to run. In this case, it would be appreciated if anyone can suggest me how should I set up the design matrix and contrast. Best regards, Khoi, Huynh.