Hi,
On 24 Nov 2008, at 08:17, Andreas Pedroni wrote:
>> Dear FSL list
>>
>> We are currently trying to analyse an event related fmri set with
>> melodic. Our plan looks as follows:
>>
>> We would like to identify components which timecourses are task
>> related. We managed this by time-locking the signal timecourses to
>> the stimulus onsets and then tested (with a Hotellings T2) for
>> significant differences of the differences between timepoints from
>> zero.
I presume you convolved the stimulus timings with an HRF to enable
them to better match the melodic component timeseries?
>> This works fine for single subject ICA. We are able to sort out
>> meaningful component maps. But when running the same procedure with
>> a group ICA (concatenated in time) results are not interpretable.
>> It looks like only few (most of the time one) subject is
>> contributing to single components. Does it make sense to timelock
>> such components to the onsets of individual stimuli, since only one
>> person is responsible for the component?
If tensor ICA is not giving activation components that include more
than one subject at a time (look at the subject-mode plots) then that
probably means that you have too much subject variability for the
tensor model to be a good one, and you're maybe better off sticking
with running ICA on each subject separately.
>> Is there a way to identify spatially similar components across
>> subjects using single subject ICA? E.g. is there a tool like fslcc
>> for more than two 4d images (e.g. pca for 4d images)?
Not trivially, though presumably if you have already found the
activation components in each subject's ICA you have already
identified the components to compare across subjects?
Cheers.
>>
>>
>> Cheers
>>
>> Andreas
>>
>>
>> Andreas Pedroni
>> Universität Zürich
>> Psychologisches Institut
>> Lehrstuhl für Neuropsychologie
>> Binzmühlestr. 14/25
>> CH-8050 Zürich
>> Switzerland
>> [log in to unmask]
>>
>>
>>
>
---------------------------------------------------------------------------
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
---------------------------------------------------------------------------
|