Hi - the problem is that you've done the temporal concatenation yourself without important pre-steps such as removing the mean from every timecourse before concatenation - hence the results you're seeing. The ICA is dominated by the changes in mean intensity. It's easier and better to use the -a concat option in melodic and it will take care of this for you. Cheers On 9 Nov 2012, at 15:58, Matteo Diano wrote: > Dear FSLroots, > i used T-ICA to perform analysis across subjects and obtain informations about correlation between time course and model. Each subjects have 4 runs. My steps: > 1) preprocessed every runs for each subject( bet, motion correction, spatial smoothing, high pass filter and registration). > 2) for each subject, i concatenated the 4 runs and then i built an unique model to use for the correlation with time courses extracted from ICA components. > 3) then i started the analysis with this command: > > melodic -i list_subjs -o tica -v --nobet --bgthreshold=10 --tr=2.0000000000 --report --bgimage=bg_image -d 40 --mmthresh=0.5 --Oall -a tica --Tdes=tica.mat --Tcon=tica.con > > At the end, i obtained only 2 components that correlate with model and these components are very different compared with the Feat results. I attach to you the time course of one component. Is it right that the time course is "not linear" like this ? Most of the components have time courses similar to this one. > > > > Thank you very much. Regards > > Matteo <t11.png> --------------------------------------------------------------------------- Stephen M. Smith, Professor of Biomedical Engineering Associate Director, Oxford University FMRIB Centre FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK +44 (0) 1865 222726 (fax 222717) [log in to unmask] http://www.fmrib.ox.ac.uk/~steve ---------------------------------------------------------------------------