[Freesurfer] normalizing activation maps

Dear Freesurfers, I have a question regarding normalization of functional activation maps. Let us assume I have functional activation maps on a vertex space (e.g. mgh files reflecting degree centrality, which do not need previous normalization of functional data). After resampling activation maps into individual vertex space (bbregister and vol2surf), I want to investigate if any differences across two groups are detectable regarding this activation maps. Which pipeline/command/stream is better to use for first normalize the activation maps into a common template and then create an output that can be implemented into mri_glmfit ? Should fsfast be used? Or is mri_surf2surf better? Thanks a lot,

Lorenzo