help > Running standard fMRI analysis in SPM using CONN images
Mar 11, 2018  02:03 PM | Elena Pozzi
Running standard fMRI analysis in SPM using CONN images
Hi All,

I have a simple fMRI task with three conditions (faces and shapes, alternated by rest) and I have preprocessed my images with CONN to perform some gPPI analysis. Now I would like to perform some additional standard fMRI analysis in SPM, basically I would like to compare the BOLD signal in the faces vs shape conditions including some additional regressors of interests (behavioural covariates).

I thought I had two option (but please me correct me if I am mistaken):
- use the swau.nii images and include motion regressor parameters and run first level again in SPM - but that would not produce the denoised images that I used in CONN;
- use the images resulting from the denoising step.
As I wanted to use the most comparable images, I decided for the latter. Following previous suggestions, I converted the DATA_Subject*_Condition*.matc in a nifti file using conn_matc2nii. However, if I understood right, this file contains all the conditions collapsed together. I run spm_file_split('niftiDATA_Subject001_Condition000.nii') but this command splits the file according to the number of session - not conditions, and I am interested in the different conditions.

Therefore my questions are:
- how do I generate separate nifti files for each condition?
- the denoising step computes the effect of each task condition controlling for all other confounding effects - including the other task conditions - so would it be conceptually "wrong" to try setting up a second level GLM in SPM including my condition 1 images vs condition 2 (i.e. setting up a contrast 1 -1) using images where I previously controlled for these effects? If so, should I re-run the denoising step not including the effect of the different conditions (perhaps just rest - which I am not interested into) or is there an alternative? 

Hope that my message is clear enough - thank you so much in advance for your help!