Thanks David and Wolf,
All my subjects have the same number of confounders. I am not sure how
to script the fsf files. Is there anywhere I can find examples?
David can you please clarify a bit further the meaning of the example
script you sent?
thanks for the help
Paul
On 9/22/2011 5:19 PM, wolf zinke wrote:
> Hi,
>
> You might want to look into the unconfound tool:
> unconfound <in4d> <out4d> <confound.mat>
>
> However, you should be aware, that it is not the same, if you fit your
> full modell and remove confounds based on this fit, or if you remove
> the noise irrespective of the regressors of interest. See the
> explanation by Christian Beckmann:
> https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=FSL;722a5c62.0807
>
> I would rather recommend to script the creation of your fsf-files, so
> that a complete model is created for each subject, and use
> fsl_regfilt. If it is just a problem, that the same 10 variable have
> different names it shoul dbe no problem, to create a loop over all
> subjects and replace the specific strings of the file names. It will
> be more tricky though possible, if each subject has got a different
> number of confound regressors that you want to filter.
>
> I hope this helps,
> wolf
>
> On 22/09/11 22:48, Paul Geha wrote:
>> Dear FSL users,
>> is there a quick way to regress out of the filtered_func_data the
>> confounders (confoundersev.txt) and obtain the residual 4D data
>> without having to run the whole analysis in feat. I need to use the
>> 4D after correction for motion and other noise outside of feat.
>> Fslreg_filt seems like an option but I will need to enter 10
>> variables for each subject manually in Glm to obtain a design.
>> thanks
>> Paul
>
|