I'm trying to use my own skull stripped T1 when running recon-all. For this, I'm doing following steps:

1) run recon-all with -autorecon-1 -noskullstrip

2) mri_convert --in_type nii --out_type mgz -rl ./FreeSurfer/ID/mri/orig.mgz -rt nearest ./own_skull_strip.nii.gz ./FreeSurfer/ID/mri/brainmask.mgz

3) run recon-all with -autorecon-2 and autorecon-3

My original own_skull_strip.nii.gz has matrix size 192x240x256. Therefore, when transforming to the size of brainmask.mgz (256x256x256), it deforms. I tried nearest, cubic, and interpolate, but none of them gives actually a good result in terms of retaining the contrast. I was wondering if it is important to keep a good contrast for brainmask.mgz , or whether this brainmask.mgz is just used to determine the brain borders for segmentation. (ie.: in which steps of recon-all is the brainmask.mgz actually used?)

Any recommendations?

Thank you!

Julie