Hi,
Hm - if the tasks are spread across runs, then creating an EV
incorporating the contrast will probably not help. Another approach of
course would be to run Dual Regression (e.g., across the runs with the
1-back and the 2-back), and then compare the runs. This would yield
complementary information.
But may I ask a question in return? How do you do the correlation of
ICA timecourses with a first level regressor, exactly. I know how to
get the timecourses - but then? Could you point me to a step-by-step
tutorial? Or do you have a command line example handy for me? This
would be a great help, as I am trying to do exactly this right now.
Thanks a lot in advance,
Cornelius
On Thu, Oct 27, 2011 at 12:50 PM, Ciara Greene <[log in to unmask]> wrote:
> Hi, I'm attempting to use the fsl_glm command to correlate the timecourse of independent components from melodic with contrasts defined in Feat. This has worked very well in previous experiments, but I'm having a problem using this method with a study where the 4 conditions (0-back, 1-back, 2-back & 3-back) were performed in 4 different runs.
>
> Under normal circumstances, a timecourse can be extracted from each component for each run within each subject, and this can be correlated with the first level design.mat and design.con files. In this new experiment, the contrasts (e.g. 2-back > 1-back) are necessarily specified at the second level rather than the first level, meaning that there is no useful information in the first level files to correlate with the individual run timecourses. The obvious solution seems to be to run melodic at the second level (averaging across runs) and use the second level contrast files, but I can't get this to work, I think because while filtered_func_data at the first level is a 4D file, the average across the 4 runs (mean_func) is 3D.
>
> I'd really appreciate any advice on this. Is it possible to run melodic on second-level data, or alternatively, is there some way I can use fsl_glm to correlate the second level design.con and design.mat files with the ICs from the individual runs?
>
> Thanks,
> Ciara
>
--
Dr. med. Cornelius J. Werner
Department of Neurology
RWTH Aachen University
Pauwelsstr. 30
52074 Aachen
Germany
|