Freesurfer surface inaccuracy

[Lestropie]

This has been at the back of my mind for quite some time, but I just spotted it in HCP 100307 I'm using for another purpose, and so it warrants documenting and looking into at some point. Pretty sure there's a systematic bias in there, not just imprecision.
Most likely culprit is the meshconvert -transform fs2real operation that is utilised within 5ttgen hsvs.
It might just be an off-by-one problem, but I probably need to contrast 1.0mm with non-1.0mm data to confirm that any fix is accurate.

[Lestropie]

• Check for possible influence of DWI metadata fixes for master #3011

[Lestropie]

From https://surfer.nmr.mgh.harvard.edu/fswiki/CoordinateSystems:

3. I have a point on the surface ("Vertex RAS" in tksurfer) and want to compute the Scanner RAS in orig.mgz that corresponds to this point:

ScannerRAS = Noriginv(Torig)[tkrR tkrA tkrS 1]'

Norig: mri_info --vox2ras orig.mgz Torig: mri_info --vox2ras-tkr orig.mgz
Test: click on a point in tksurfer. Use "Vertex RAS" to compute ScannerRAS. Save Point. In tkmedit, Goto Saved Point, compare ScannerRAS to "Volume Scanner Coordinates".

mri_info --vox2ras orig.mgz is just the voxel2scanner transform. Need to figure out exactly how mri_info --vox2ras-tkr produces what it does.