Hello, 

I would like to perform resting state connectivity (seed-to-voxel) analysis at each subject's native space. Since my population is characterized by brain pathologies, I was thinking to use FreeSurfer's reconstruction algorithms for optimal segmentation (it also allows for manual corrections of defects).

If I am to do that, could I still use Conn for data denoising? I mean to say, which FreeSurfer images are the equivalent for the gray matter, white matter and CSF? Maybe other tips that could be helpful as well?

Thank you so much in advance!

Best,

Athena

Hi Athena,

If you have processed your subjects anatomical data in FreeSurfer you can very easily import that into CONN to import that info and run all of your analyses on those cortical surfaces. To do this:

1) in the Setup.Structurals tab select for each subject the anatomical files in the freesurfer-generated subject*/mri/ folder (e.g. the T1.mgz or .mgh files there) and CONN will automatically import each subject's surface information and it will offer you the choice to also import the corresponding gray/white/CSF masks from the aseg.mgz files (or if you are using batch scripts you may also use the conn_importaseg script to automaticlaly import those masks)

2) set the 'Analysis space' option in the Setup.Options tab to 'surface (FreeSurfer fsaverage)'

3) If you also have a *.annot file with your choice of ROIs defined on the fsaverage space you may also add those to the list of ROIs in Setup.ROIs. You may also enter there other subject-specific *.img/.nii/.mgh ROIs that you may have (note that if you have all of your data in subject-space you should not be using any MNI-space ROIs and your ROIs should be in subject-space as well) 

4) last, make sure you run one of the "subject-space" preprocessing pipelines (instead of the "MNI-space" preprocessing pipelines) so your functional data is kept in subject-space and appropriately coregistered to the FreeSurfer-generated surfaces

Hope this helps
Alfonso
 
Originally posted by Athena Demertzi:

Thank you, Alfonso.

Can I ask you some more questions:

1. When using the proprocessing surface-based (single subject) pipeline, I need to remove the segmentation part, correct?

2. When doing so, why my functional data are of different dimension compared to the T1,mgz and the segmented images?

3. Conn does not automatically load the art_outliers_and_movement file for the scrubbing 1st level covariate, is that normal?

Thanks so much!

athena

Hi Athena,

Some thoughts on your questions below
Best
Alfonso
Originally posted by Athena Demertzi:
Yes, if you prefer to use the FreeSurfer generated masks and have already entered those in Setup.ROIs then simply remove the segmentation step from the preprocessing list so those FreeSurfer-generated masks are not replaced by the SPM-generated masks.

The functional data is coregistered (but not resampled) to the structural volumes, so they will be in the same space (both in subject-space) but not necessarily having the same voxel dimensions. You may check that whether the structural and functional data is appropriately coregistered by going to the Setup.Functional tab and selecting 'functional tools. Check functional/anatomical coregistration'

CONN should be automatically loading, separately, the art_outliers file into a first-level covariate called 'scrubbing', and the estimated subject-movement parameters into a first-level covariated named 'realignment'. If it is not doing so check that a) the ARToutlier identification step was included as part of the preprocessing steps; and b) that the procedure terminated without any error messages

Hope this helps
Alfonso

Hi both,

Just to add on to this, I have fMRIprep data that has also been run through freesurfer. What if I'm interested in conducting a seed to ROI analysis using a specific seed, segmented by freesurfer (e.g., the nucleus accumbens)? What would I select as my ROIs? (it seems that the freesurfer segmentation just has the whole segmented brain, plus a table, but no individual seeds as their own files?)

Hello,

Did you ever figure out how to do this?

Derya