Dear CONN experts,

I am working on a ROI-to-ROI connectivity analysis to be done in subject space, importing ROIs (both tissue segmentations and atlas-based parcellation) obtained with CAT12.

I have some concerns about the functional/structural coregistration step.

I don't have fieldmap acquisitions to perform proper susceptibility-distortion correction, but EPI distortions are not massive in my dataset, with affine coregistration giving quite acceptable results.
Anyway, I was trying to implement a nonlinear functional/structural coregistration step to partially correct for EPI distortions.
I tried the functional_coregister_nonlinear step in CONN but I am not fully convinced by the results, probably because of its indirect "via-MNI" nature.

Is it possible to implement a direct nonlinear functional/structural coregistration in CONN?
Otherwise, do you have any experience/suggestions on the integration with other tools (spm normalization? the ants syn registration used in fmriprep? others?).

Also, given that my objective is to obtain ROI-to-ROI connectivity matrices in the subject space, I was wondering if reslicing (and interpolating) fMRI data could affect the results.
Do you suggest keeping the original fMRI resolution or reslicing fMRI volumes to 1x1x1mm (the original resolution of structural scans) or 2x2x2mm?

Thank you as always for your precious help,
Giuseppe