Thank you Andreas. Makes sense.
One last question. When we are running the t-test, should we correct for multiple comparisons?
Hi,
you won't get a spatial map of the difference in your case. You get a spatial map of the average effect and, by using the time point entries (=subjects in your case) you can test for differences of the average effect across groups (i.e. ordered subjects)
outside FSL (using R, matlab - whatever you fancy). Does that make sense?
Cheers,
Andreas
"Franklin, Crystal G" <[log in to unmask]> hat am 9. April 2013 um 22:44 geschrieben:
Steve,
We have taken the melodic_mix data and performed an unpaired t-test on the components we are interested in. Now we are having trouble interpreting the results. What does it mean if we see a significant group difference in a component? We are also noticing
for some components we are getting a negative and positive group mean, how do we interpret this?
Thank you,
Crystal
You already have this - the spatial ICA maps.
Will we be able to get a map from this?
Crystal
Hi
In this case the ICA "time series" are the subject weights - so (e.g.) read melodic_mix into Matlab, identify the column number that matches the spatial map component number that you "like" - and then do a statistical test between different timepoints
(subjects) in that column of weights.
Steve.
Thank you for your quick response.
So I have concatenated all my single subject images into a 4D NIFTI image and ran them through Melodic using single-session. Now I have my components, but I am stuck on how to get my group differences. Any suggestions?
Thank you,
Crystal
Hi - yes this should be possible. If you have "time series" data for each subject then you should use group-ICA and dual-regression; if you only have a single summary image per subject then you should concatenate all images
into a 4D NIFTI image and run "single-session" (default) melodic.
Steve.
Dear FSL,
I have a FDG PET study that contains 79 subjects that are either FH+ or FH-. The data has already been preprocessed and registered into standard space. I am now wanting to do an ICA analysis to get the components and then do group comparisons with these
components. Is this possible and if so, how do I go about setting this analysis up.
Thank you,
Crystal
---------------------------------------------------------------------------
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
---------------------------------------------------------------------------
---------------------------------------------------------------------------
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
---------------------------------------------------------------------------
---------------------------------------------------------------------------
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
---------------------------------------------------------------------------