Hi
You'll need to correlate the spatial maps rather than the time courses, so you should be using melodic_IC. Make sure that all data sets are in the same space, either by registering the data first or by running a melodic group analysis and have this done for you.
The first column in the Tmodes file is the Eigen-timeseries from the other 10 time series (rank-1 approximation)
hth
Christian
On 2 Jan 2010, at 15:50, SUBSCRIBE FSL B. Jeong wrote:
> Dear FSL expert,
>
> I'd like to perform correlation analyses among ICs of Melodic output at each
> subject (resting state data).
> I can see two candidate file, melodic_mix and melodic_Tmodes in
> groupmelodic.ica directory.
> Which one I have to use for the correlation analyses among ICs?
>
> Here is another question about melodic_Tmodes.
> I could get a 250 * ((1+10) * 49) matrix file if melodic show 49 ICs with 250
> timepoint data of 10 subjects.
> My question is what the value of the first column of (1 + 10) column set is
> derived from? It doesn't seem like mean or sum of the rest 10 columns of (1 +
> 10) column set.
>
> Thanks,
>
> B Jeong
|