Surface-based ICA for resting-state fMRI

Hello NeuroStars,

I’m attempting to create individualized RSNs within my participants. I’ve run all participants’ rs-fMRI data through fMRIPrep and then XCP_D. I’ve done some seed-seed analyses with atlas-based parcellations (e.g., Yeo/Schaefer 400p/7n), but I would now like to create individualized networks.

My hope is to use the “cleaned” rs-fMRI data from XCP_D for ICA. The rs-fMRI subcortical data is now in an average space (MNI152) and cortical signal is projected onto the surface (subcortical and cortical stored in fsLR).

I appreciate that ICA requires all rs-fMRI data be in one space, but does this mean I should project “cleaned” rs-fMRI data back into the cortex and combine both hemisphere with the subcortical rs-fMRI data? Is there another way to run ICA when data is “cleaned” and in fsLR format?

Any help would be much appreciated!