Dear Monica,
As you realised, you will need to transform your coordinates and you can
do this with std2imgcoord. To use this you will need to give it the transformations
from your feat/reg directory in the form:
std2imgcoord -img example_func -std standard -warp highres2standard_warp -premat example_func2highres.mat -vox coord_file.txt
where the coord_file.txt contains the coordinates you want to transform (each line
should have one coordinate - three numbers per line - and be in standard space
mm values).
Then you can use the coordinates it gives you with fslmeants and the -c option.
You can either use the filtered_func_data as the input file for this or the res4d file
in the stats folder. If you use res4d then it will have the covariates removed from
it (your 9 EVs) whereas filtered_func_data has all the preprocessing but still
contains these covariates.
All the best,
Mark
On 22 Jul 2011, at 17:58, Monica Wey wrote:
> Dear experts,
>
> I am trying to use fslmeants to extract time series data for a seed-based resting-state fMRI analysis.
> I would like to define the "seed" using some MNI coordinates published in the literatures.
> However, the fslmeants function seems to read the input coordinates (with -c option) as "voxel coordinates but not the MNI.
> Any suggestions on how to convert the coordinates?
>
> I am doing
> (1) Pre-processing: bet, motion correction, band-pass filtering, registration: subject's 4D EPI -> subject's T1 -> MNI 1mm brain.
> (2) generate a 9 columns text file for confound EVs (6 parameters motion, WM, CSF, and global)
> (3) extract time series from target regions (based on MNI coordinates) for FEAT analysis
> Those these steps sound correct?
>
> Thanks for any advices in advance.
>
> Monica
>
|