Dear all,
I have preprocessed my resting state data using FEAT without highpass
filtering, then denoised them using ICA-AROMA as described in the
manual.
I have then created CSF and WM segmentations using fast, eroded the
segmentations, and extracted the CSF and WM timeseries to regress out
the CSF and WM signals using fsl_glm (with demeaning) to output the
residuals. Finally, I have readded the means of the residuals,
highpass filtered the data and registered the images to standard space
(using applywarp and the warp field generated by FEAT).
Now I need to do a seed-based analysis (I know this is not generally
recommended, but please bear with me).
Would it be sufficient to do the following:
1) Create a common mask
for i in (subjects)
do
fslmaths func_$i -Tstd -bin mask_$i
done
fslmerge -t maskALL mask_*
fslmaths maskALL -Tmin mask
2) Extract time series and calculate correlation maps
for i in (subjects)
do
fslmeants -i func_$i -o roi_$i -m roi
fsl_glm -i func_$i -d roi_$i -o map_$i -demean -m mask
done
3) Merge maps and run randomise
fslmerge -t mapALL map_*
randomise -i mapALL -o output -d design.mat -t design.con -e
design.grp -m mask -T 5000
I would appreciate any comments. Thank you very much.
All the best,
Anders
|