-d 50 -------------------- Stephen M. Smith, Professor of Biomedical Engineering Associate Director, Oxford University FMRIB Centre FMRIB, JR Hospital, Headington, Oxford. OX3 9 DU, UK +44 (0) 1865 222726 (fax 222717) [log in to unmask] http://www.fmrib.ox.ac.uk/~steve ---------------------- On 27 Sep 2012, at 13:18, Benjamin Kay <[log in to unmask]> wrote: > Depends on exactly what you're trying to do, but that would be a good start. Hmmm... I'm suprised this option hasn't percolated into the Melodic GUI. > > On Thursday, September 27, 2012 08:08:34 you wrote: >> Hi Ben, >> >> Is it >> >> melodic -i filtered_func_data -n 50 -o filtered_func_data.ca >> >> ? >> >> Thanks. >> >> Mark >> >> 2012/9/27 Benjamin Kay <[log in to unmask]> >> >>> Why are you generating 300 components? That's way too many for >>> interpretation! I'm guessing you let MELODIC try to pick the number of >>> components automatically. I have found that this often fails for group ICA >>> with less-than-pristine datasets. Try using the -n option to limit the >>> number of components to around 50 (or a # of your choice). As I recall, >>> this will discard all but the most 50 relevant PCA components prior to >>> running ICA. >>> >>> On Wednesday, September 26, 2012 21:59:14 you wrote: >>>> Hi, >>>> >>>> I believe this might be a simple question but I've checked previous >>> lists but still can not find answer...... >>>> >>>> I am a new user of MELODIC and try to use its output data to remove >>> artefact in my subject's FMRI data. I did FSL's FEAT analysis and selected >>> MELODIC analysis in Pre-stats in one subject's long-run FMRI data (a >>> complex cognition task containing 800 volumes). Report of FEAT Pre-stats >>> showed that the translation of this subject is less than 1.5 mm. In MELODIC >>> report, it produced 300 independent components, which retains about 85% of >>> total variability. The first ~30 components captured about 40% of the >>> variance across the entire functional data. >>>> >>>> Is the number reasonable? The problem is, it will be grueling to go >>> through these 300 ICs to remove noise components! Could I just check the >>> first ~30 components to remove noise? >>>> >>>> >>>> Thanks for any help. >>>> >>>> Mark >>>> >>> >> >