Hi Antonin
I think you are correct that the brainmask.mgz voxel intensities are irrelevant as long as they are "on" (which I believe we define as >5)
cheers Bruce On Wed, 29 Mar 2017, Antonin Skoch wrote:
Dear experts,
I am reprocessing a group of subjects originally processed and edited in v5.3, by version v6.0. Due to the fact that some subjects had brainmask.mgz edited, I kept the brainmask.mgz from v5.3 to save time in editing (i.e. did not use -clean-bm in recon-all -all). I noticed that the brainmask.auto.mgz (from v6.0) is more aggresively filtered than brainmask.mgz (from v5.3) in basal ganglia (they have intensity 110 in v6.0 in contrast to v5.3) and also generally in white matter (much more voxels are 110 in v6.0 than in v5.3). Is it an issue? As I looked to the recon-all stream, the brainmask.mgz is in fact probably used only as mask for mri_em_register, mri_ca_normalize, mri_ca_register and mri_normalize, so the voxel intensities does not matter. Could you confirm that? Does it pose a problem to mix the subjects with brainmask.mgz created in v5.3 (and edited) with new subjects processed from scratch by v6.0?
I considered to create v6.0 brainmask.mgz while keeping edits by remasking v6.0 T1.mgz by v5.3 brainmask.mgz using mri_mask T1.mgz brainmask_v5.3.mgz brainmask.mgz -keep_mask_deletion_edits but I do not know whether this is in fact necessary. Could you please comment on?
Regards,
Antonin Skoch