I'm currently interested in subcortical structure volumes, particularly hippocampus, so I'm looking in my subjects dirs at stats/aseg.stats to get those values. It's not clear to me if those measurements are in native or FS space and when I skim over the code for mri_segstats I don't see anything in there that applies a scaling factor to take into account the scale parameter of the native space to FS space transform, so I've been assuming that the volumes reported in aseg.stats are actually in FS space although that seems odd to me. So first off, are aseg.stats measurements in native or FS space?

As a naive first attempt to calculate the scale factor myself I simply ran xfm2param on mri/transforms/talairach.xfm and then multiplied together all three scale factors to end up with something I could multiply the aseg.stats results by to get back to native space measurements. In one particular case I got a result of about 1.26. However, when I follow the directions at http://freesurfer.net/fswiki/FsAnat-to-NativeAnat to move the aseg.mgz into native space and then use mincstats to generate a histogram of the label values in the segmentation and compare that histogram to the same histogram generated from the FS space aseg.mgz, I find the values I get back from both files (a) don't match aseg.stats measurements and (b) are not scaled by anywhere close to 126% relative to each other. The former is undoubtedly a result of mri_segstats taking partial volume effects into account when calculating measurements, which is great, but the latter is genuinely surprising and makes me wonder if I'm making a bad assumption about what talairach.xfm is really supposed to represent.