Hi, I have the same problem. I ran a temporal concatenation over 11 subjects. Each subject has 3 sessions, and I first concatenate the sessions for each subject first (since not all sessions have the same number of time points). After this, each subject has a data set of 609 scans. melodic gave me a total of 236 IC components.From component 172, it give a negative variance count. I will follow up on this thread. Thanks, Lingling ----- Original Message ---- From: Christian F. Beckmann <[log in to unmask]> To: [log in to unmask] Sent: Tuesday, February 26, 2008 6:02:57 AM Subject: Re: [FSL] group ica Hi, this should not happen, it might be an overflow somewhere due to really huge data, can you tar up a report foleder and upload it to http://www.fmrib.ox.ac.uk/cgi-bin/upload.cgi and send me the upload ID, please. Thanks Christian > Hi Christian, > > Sorry, I still have questions to need your kind helps. > > From the group ICA output results, sometimes I can see the last > component has negative variance proportion value, like -1.06% of > explained variance; -0.61% total variance. I am confused with it. > Would > you please explain it for me? > In addition, group ICA also has a simple plot and a boxplot that show > the relative effect size across the different sessions/subjects, which > come from the values in the first column of matrix s..txt I guess. My > questions are what the physics meanings of these values is? How to > compute the values and they have units? What does negative value mean? > Sorry to bother you so much. > > Any help would be great appreciated! > > With best regards, > Weiming > > ____________________________________________________________________________________ Looking for last minute shopping deals? Find them fast with Yahoo! Search. http://tools.search.yahoo.com/newsearch/category.php?category=shopping