I am writing to inquire about performing seed-voxel FC and ROI-ROI FC analyses using the output of XCP-D. As the xcp-Engine is no longer supported, I am seeking alternative post-analytic approaches. I would like to know if Nilearn is a recommended solution for this purpose.
If you are talking about ROIs in an atlas, then XCP_D already supports that as part of its workflow! Otherwise it wouldn’t be too hard to extract timecourses from the postprocessed BOLD from XCP in specific ROIs to do your own connectivity analysis.
Thank you for your assistance! I still have a question regarding working with the self-defined ROIs and try to preform ROI-ROI FCs. I would greatly appreciate any suggestions or guidance you can provide.
My question is, if I use Nilearn to extract timeseries from Bold img, then I should load the xcp-d output bold img (i.e., sub-01space-MNI152NLin2009cAsym_desc-denoised_bold.nii.gz) and the confounds file. I am not sure which confounds files should be used (_design.tsv, _motion.tsv, or _outliers.tsv?)