processing-scripts > nuisance signal regression
Showing 1-6 of 6 posts
Display:
Results per page:
May 23, 2011  08:05 PM | cirong liu
nuisance signal regression
The scipt of nuisance signal regression defautly regresses out 6 nuisances, including Global, CSF, WM, and 6 motion parameters.

What am I going to do if I just wanna regress out Global and  6 motion parameters?

I have tried to modify the nuisance template, ''nuisance_cocaine.fsf". I set the number of EV to 7 and deleted those settings for CSF and WM, but it didn't work.


May 27, 2011  08:05 AM | Maarten Mennes
RE: nuisance signal regression
Hi cirong liu,

I took a quick look and for me adjusting the model works. What type of error do you get? Or is there no error at all, just no difference with your previous results (i.e. those with 9 nuisance signals)?

If you are using the batch_process.sh or 0_preprocess.sh you will have to edit 0_preprocess.sh. On line 92 you have to change the last input to the call to 5_nuisance.sh to your new template. Alternatively you can just run 5_nuisance.sh using your newly created .fsf file.

Does this make sense?
Maarten
Jun 7, 2011  04:06 AM | cirong liu
RE: nuisance signal regression
Thanks。

yes, I had "edited 0_preprocess.sh. On line 92 you have to change the last input to the call to 5_nuisance.sh to your new template".

And in order to creat the new template I just modified the old one, where I set the number of EV to 7,  deleted those settings for CSF and WM and rearrange the number of remaining nuisances . However, the new template didn't work.

Do you mean that I need to create a totally new template rather than modify the old one?
If yes, how should I create a new one.


I attached my modified template with this message.
Attachment: nuisance_modify.fsf
Jun 8, 2011  07:06 AM | Maarten Mennes
RE: nuisance signal regression
Hi Cirong,

I took a look at your model and it matched the one I succesfully used. The only difference is that your contrast (which is not used anyway) is empty. That might prevent the model from running successfully. Just set EV1 to 1 in the global contrast and see it that solves it. With the empty contrast, feat fails to make the design matrix and other needed files for the analyses.

Tested it here with my model. It fails if I make the contrast empty like yours. So if you set that contrast correctly it should hopefully  work. Otherwise we're missing something else.

Maarten
Jan 20, 2014  08:01 AM | Viola Wittstock
RE: nuisance signal regression
Dear all,

first of all, many thanks for the great scripts!!

I have a similar problem: I want to run an analysis without global signal regression, so I delete the global signal EV from the design. However, that's not working, the error message is: 

Running feat model
terminate called after throwing an instance of 'NEWMAT::ProgramException'
/media/fmriDATA_2/fcon_1000/scripts/5_nuisance.sh: Zeile 82:  2157 Abgebrochen             feat_model ${nuisance_dir}/nuisance

Can anybody help me with this?

Many thanks!
Viola

ps I'll attach my design file
Attachment: nuisance-noglob.fsf
Feb 6, 2014  03:02 AM | Maarten Mennes
RE: nuisance signal regression
Hello,

are you running the nuisance_regression as part of the scripts or are you running it in some other form? The .fsf file you attached completes a full first-level analysis, while the one that we use only runs stats + poststats. 

The error message you provide is too isolated to see what the problem is. Maybe it cannot find the .fsf file or it's permissions are wrong.

I've attached the nuisance_noglobal.fsf file that worked here previously. Make sure you indicate you want to use that as the template (your file was also missing some sting values the template script is looking for).

Finally, a word of caution. The scripts are no longer supported and are not in accordance with current standards in the field (e.g., the nuisance variables are not filtered similar to the data). Therefore I suggest users to upgrade to C-PAC: http://fcon_1000.projects.nitrc.org/indi/cpac/index.html

Best,
Maarten