Hi all,
We are processing longitudinal structural MRI data on Freesurfer 5. 0. As our images have some intensity problems, we were running an N3 correction outside of FreeSurfer using a mask in order to maximize this step (as per recommended by Boyes et al., 2008). We first ran autorecon1 and then used the brainsmask.mgz as a mask for the N3 correction, using the following steps.
mri_convert orig.mgz orig.mnc
mri_binarize --i brainmask.mgz --min 1 --o binmask.mgz*
mri_convert binmask.mgz binmask.mnc*
mincnorm orig.mnc orig_norm.mnc
nu_correct -mask binmask.mnc -distance 50 -iterations 1000 -stop 0.0001 -shrink 2 -fwhm .05 orig_norm.mnc nu.mnc
mri_convert nu.mnc nu.mgz*
This nu.mgz was then fed into freesurfer and autorecon1 was run over again without nuintensity correction. However, during this step, many of our scans are crashing, with the T1.mgz being very dark. When the same scans were run through Freesurfer with the nuintensitycor-3T flag (without using a mask), there were no crashes. Do you have any advice on why the N3 correction using the mask is crashing when re-running autorecon 1? Should we be using different nu_correct parameters to help the process?
Thanks for your help!
Nandi