Hi,
if it is resting-state FMRI you should run Melodic in the temporal concatenation mode.
You should probably put the subjects into one group ICA first, regardless of this issue.
Hth,
Andreas
________________________________________
Von: FSL - FMRIB's Software Library [[log in to unmask]] im Auftrag von Jie Yang [[log in to unmask]]
Gesendet: Montag, 1. August 2011 20:52
An: [log in to unmask]
Betreff: [FSL] Questions about group ICA
Hi,
Sorry to bother you. We are using MELODIC to conduct group ICA. We have two groups of subjects: one group of English speakers and one group of Japanese speakers. We hope to run ICA for both groups and examine group differences.
Our questions are about the way of doing the group ICA: Shall we conduct tensor-ICA for two groups separately? Or shall we combine the two groups together and run a large T-ICA?
If we conduct the tensor-ICA for two groups separately, what output can be used to conduct contrast analysis (e.g. T-test)?
If we run a large T-ICA, shall we run the contrast analysis on the subject/session mode for some "best-fit" ICs? Then how to generate the spatial maps for each group and the maps for the contrast results?
Thanks very much for the help,
Jie Yang
|