processing-scripts > DEMEANING after signal noise regression
Showing 1-2 of 2 posts
Results per page:
Nov 30, 2011  11:11 AM | Michael Huetel
DEMEANING after signal noise regression
Hello Everyone,
I'm quite new in the field of fMRI especially resting-state fMRI.
I tried to understand the certain kind of preprocessing steps, where I stopped at line: 
## 8. Demeaning residuals and ADDING 100
3dTstat -mean -prefix ${nuisance_dir}/stats/res4d_mean.nii.gz ${nuisance_dir}/stats/res4d.nii.gz
3dcalc -a ${nuisance_dir}/stats/res4d.nii.gz -b ${nuisance_dir}/stats/res4d_mean.nii.gz -expr '(a-b)+100' -prefix ${func_dir}/${rest}_res.nii.gz
Could somebody please give some explanations why these step is useful. 
In case of just considering resting-state fMRI, are there some preprocessing steps to avoid?


Dec 15, 2011  10:12 AM | Maarten Mennes
RE: DEMEANING after signal noise regression
Hi Michael,

appologies for the slow response.

The processing step you're asking about is in essence a workaround step to avoid problems with FSL later on if you were to use FEAT. These problems might not exist anymore, but we kept the step anyway just to be sure.

The demeaning is done to make sure the mean of the residuals is 0 as it should be. From the residuals calculation the means are very close to 0 but not entirely, so we do the demeaning just to be sure.

The 100 is then added to avoid problems with FEAT, where we ran into an error exactly because the mean of the residuals is 0. FEAT might not expect you to use residuals for your analysis.

In essence this step does not alter the data.

All the scripts are meant to process resting state fMRI. They are not intended for task-based fMRI. Therefore we would recommend to use all steps described in the scripts. But, of course you are free to use/replace whatever step according to your own likings!

Hope this helps,