Dear FreeSurfer,

Hello,

I have been using the preprocessed images (which have already undergone N3 correction) from the Alzheimer's Disease Neuroimaging Initiative database as input for recon-all -all.( version 5.3.1)

I have 3 questions:

(1) Does recon-all use the nu.mgz during subcortical segmentation and cortical parcellation?

(2) If so, then would my image be effectively undergoing N3 "twice", and does this affect the accuracy of the segmentation in any way?

(3) A bit of a side question, but does rescaling to isovoxels and performing ACPC aligment before recon-all make the segmentation results more accurate?

Thank you in advance.

Best regards,

Kristine