Hi
melodic_mix contains the effective mixing matrix (i.e. the one
obtained if the data was decomposed as a simple (time*subjects) x
voxels matrix. That is, if you have 5 subjects, 200TRs each and
melodic estimated 20 components then melodic_mix will be of size 1000
x 20, each column containing the concatenated subject-specific time
courses.
The tXX.txt contains the same time courses scaled by the relevant
values from the subject mode. That is, the final N columns in the
tXX.txt files contain time series of the same 'shape' but different
scale from the ones in melodic_mix. This is done in order to
facilitate comparisons of 'shape' of regressors. These files also
contain the rank-1 approximation as the first column and possibly
also the total model fit (as the second time course) in case a
temporal design matrix has been supplied. That is, the final N time
courses in tXX.txt should match the XXs column of melodic_mix if you
apply the scalings in sXX.txt to the melodic_mix tcs.
I don't quite understand what you mean by 'raw time-course of each
person prior to the run of ICA' - that would be the raw data?
The spatial maps are the same for all subjects, there only is one set
of maps for the entire decomposition. You could can calculate subject-
specific maps if you need to by extracting the relevant subset from
melodic_mix and regressing it against the original data (use fsl_glm
for this)
hope this helps
christian
On 24 Aug 2007, at 01:28, Christopher Bell wrote:
> Hello,
>
> I ran a temporal concatenation of subjects and wish to retrieve the
> time
> courses for each subject's contribution to the spatial map
> components. After
> reading the fsl 4.0 page it was unclear to me whether these files
> were in
> the report/tXX.txt files or melodic_mix. I assume one of these two
> sets of
> time courses is the raw time-course of each person prior to the run
> of ICA,
> is this true? Is the best way to generate subject-specific spatial
> maps from
> these time course to do a run through feat post-stats or does
> Melodic now
> have a feature to display a subject-specific spatial map from group
> melodic
> run? Thanks in advance.
>
> Chris Bell
____
Christian F. Beckmann
University Research Lecturer
Oxford University Centre for Functional MRI of the Brain (FMRIB)
John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK.
[log in to unmask] http://www.fmrib.ox.ac.uk/~beckmann
tel: +44 1865 222551 fax: +44 1865 222717
|