Hi Jason, Interesting to compare the two approaches -- I've thought some about the issues involved in the past, so I'll add my $0.02 in the hope it might benefits others on the list as well.

One conceptual potential note of caution: Because the aparc stats and aseg stats are two very different processing streams, there isn't really a perfect way to combine volumes from the two streams. That is, if you add/subtract volumes across the aseg and aparc to get a estimate of "total brain volume", there will inevitably be some % of the "true" volume that is "double-counted" and some other % that may not be counted at all.

One approach you might be interested in: mris_anatomical_stats can provide the total volume encapsulated within a given surface. If you use that for the pial surface (and add the resulting L and R values) the result should be a pretty good estimate of total cortical + white matter + subcortical + ventricular volume, since all of that is (generally) enclosed within the pial surface. Absent from volume defined in this matter would be the brain stem and cerebellum (which don't fall within the pial surface), possibly some voxels of the inferior-lateral ventricles (which sometimes fall outside the pial surface) and those voxels in the medial temporal area that fall outside the pial surface.

I suspect that volumes defined in this matter would still show approx. a 25% difference with eTIV, but I've never quite gotten around to actually comparing the two.

Best, Mike H.

On Wed, 2007-12-12 at 13:09 -0800, Jason Reed wrote:

Hi all,

We want to normalize our data using eTIV and compared the eTIV values generated in aseg.stats to the total brain volume we calculate by summing:

1. Non-cerebral cortex volumes from aseg.stats (including

ventricles) 2) Cerebral cortex volumes from lh and rh.aparc.stats We compared eTIV to the total brain volume as calculated above from aseg and aparc.stats for 40 subjects and found that, on average, eTIV is 27% larger. This difference is larger than we expect. Are we missing something?

Thank you!

Jason _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer