Thanks for your quick reply. Ingo Am 09.05.08 22:52 schrieb "Christian F. Beckmann" unter <[log in to unmask]>: > Hi, > > as you said, Tensor ICA does assume consistent temporal responses > between subjects - if there is a consistent difference then it won't > be the best model. The easiest is to run TICA first and then have a > look at the subject specific time courses (saved in the tXX.txt files > inside the report directory - the last N columns are the individual > temporal dynamics) You can then check if there is a consistent(!) > temporal difference between group A and B - if so, re-run with concat. > hth > Christian > > On 9 May 2008, at 21:45, Ingo Wolf wrote: > >> Dear FSL team, I have a question regarding between group analysis. I >> apologize if this has already been answered but I couldnīt find what >> I was >> looking for in the archives. >> I conducted a fMRI study with movies in two groups, a clinical and a >> control >> group. I would like apply a model free Tensor ICA approach for group >> comparison. Would you recommend this method or is it more plausible >> to apply >> the concatenation approach due to prevenient temporal dynamics in the >> different groups? >> Thanks in advance for your help, >> >> Ingo