Dear all,

Apologies for this simple question.  I have tried to search the forum but I can't quite find a question that addresses my very basic lack of understanding of something.

I want to use Conn to run CompCor and then move to SPM for my first level analyses.  When I run the denoising step on a few sample functional runs, I have folders called "data" and "results".  "data" has files that are prepended "COND", "COV", "DATA", and "ROI" and one file called REX.mat.  It looks like the information from running ART to get outliers is in the COV files, but I have no idea where information about the CompCor filtration is stored, or what form it takes -- is it a resliced image?  A modification to the headers of my preprocessed image?  Some mat file somewhere? So I don't know how to use the filtered image in my preprocessing.

Also, a related question: if the CompCor filtration *is* an edit to a header file, and I was getting the warning message "no unsmoothed volume found; using original functional volume instead" does that mean that the volumes that are having their header changed are my original functional volumes?  I originally preprocessed my data using DARTEL and it smoothes and normalizes in one step.  

Thanks very much, and thanks in advance for your help.

 Best, 
Katie

I've made some progress in my understanding of this issue, I think, so I thought I would post my updated understanding and subsequent question.

The files in the directory /preprocessing/presults with the prefix "DATA" seem to have the right dimensions to be my filtered data.  I don't know offhand how I make .mat files into .nii files, but I can investigate that.  If the conn users have somewhere helpful to direct me, I'd take the advice.  

Thanks again,
Katie

Ok, now I actually get it.  I have to check "create confound-corrected time series" in the options tab of Setup.  Then it makes Nifti files for me.

A friend of mine who is a computer programmer said that at his job they kept a plush duck to be an audience to help people work through problems out loud.  Thank you, conn help forum, for being my plush duck.

Hi Katie,

Glad to hear everything worked out. And regarding your question about the "no unsmoothed volumes" message, you may find the relevant info in this post (http://www.nitrc.org/forum/message.php?m...)

Hope this helps
Alfonso

Originally posted by Katie Surrence:

Thx for the information.
I'm interested by using this way to compute Dynamic Causal Modeling in SPM.

Do i need to define some regressor in SPM to make and estimate the fmri model with the denoised images ?

After estimate my fmri model, i want to use atlases from conn to defined some roi that i'll use in the DCM. With the SPM tool, volume of interest, i choose the atlases or network nifti file from Conn as a 1st mask image, the mask.nii (created by fmri model estimation) as the second mask and i1&i2 as the expression. Is it the right way ? how can i be sure that the ROI delimitation is good ?

Thank you for your support
Best
JB