[Freesurfer] normalized SPGR as an input for inflation

Dear experts,

I run my statistical analysis in SPM, while after coregistration I normalize my data (SPGR and EPI). Since my models and activations images are based on normalized data it's natural for me to use the normalized SPGR as an input for freesurfer inflation. Is it a good enough? Since I run inflation for each individual brain, I basically can do it for non-normalized anatomy, which is preferable, I think. However, then the coordinates are not MNI / Talariach. Attached the comparison of activations for normalized (left side) vs. non-normalized. It looks pretty similar, though the brains look of different shape. So, am I OK to do it on normalized brain?

Thanks for help, Vadim