Dear Freesurfer users,
I am wondering how are the volumes calculated from stats such aseg.stat. When I export the aseg and do a voxel count and multiple the volume per voxel, my total volume is different from that is given by the stats. For example, the hippocampal volume given by aseg.stats for one of my controls is 4000 cm3, but when I export the aseg.mgz into matlab and do a voxel count (my original voxel volume is 1mm3 isotropic) for that labeled mask, the total volume is aorund 3000 cm3. Does freesurfer does some sort of intensity normalization or partial volume correction? Even if so, the magnitude of difference is large that I am puzzled.
Thank you very much.
yes, we do a partial volume correction of the volumes, accounting for the fraction of the structure in each border voxel.
cheers Bruce On Mon, 15 Feb 2010, Hui J Yu wrote:
Dear Freesurfer users,
I am wondering how are the volumes calculated from stats such aseg.stat. When I export the aseg and do a voxel count and multiple the volume per voxel, my total volume is different from that is given by the stats. For example, the hippocampal volume given by aseg.stats for one of my controls is 4000 cm3, but when I export the aseg.mgz into matlab and do a voxel count (my original voxel volume is 1mm3 isotropic) for that labeled mask, the total volume is aorund 3000 cm3. Does freesurfer does some sort of intensity normalization or partial volume correction? Even if so, the magnitude of difference is large that I am puzzled.
Thank you very much.
Thank you for the quick response. If you do a partial volume correction, I would think the volumes from freesurfer would be less than simply counting all voxels, but it seems to be the opposite. Any thoughts?
----- Original Message ----- From: Bruce Fischl fischl@nmr.mgh.harvard.edu Date: Monday, February 15, 2010 10:42 am Subject: Re: [Freesurfer] Subcortical Volume Calculation To: Hui J Yu hjyu@ic.sunysb.edu Cc: freesurfer@nmr.mgh.harvard.edu
yes, we do a partial volume correction of the volumes, accounting for the fraction of the structure in each border voxel.
cheers Bruce On Mon, 15 Feb 2010, Hui J Yu wrote:
Dear Freesurfer users,
I am wondering how are the volumes calculated from stats such
aseg.stat. When I export the aseg and do a voxel count and multiple the volume per voxel, my total volume is different from that is given by the stats. For example, the hippocampal volume given by aseg.stats for one of my controls is 4000 cm3, but when I export the aseg.mgz into matlab and do a voxel count (my original voxel volume is 1mm3 isotropic) for that labeled mask, the total volume is aorund 3000 cm3. Does freesurfer does some sort of intensity normalization or partial volume correction? Even if so, the magnitude of difference is large that I am puzzled.
Thank you very much.
it's not necssarily less, as voxels that are not labeled part of the structure may contain some volume as well (e.g. ventricular voxels containing some hippocampus) On Mon, 15 Feb 2010, Hui J Yu wrote:
Thank you for the quick response. If you do a partial volume correction, I would think the volumes from freesurfer would be less than simply counting all voxels, but it seems to be the opposite. Any thoughts?
----- Original Message ----- From: Bruce Fischl fischl@nmr.mgh.harvard.edu Date: Monday, February 15, 2010 10:42 am Subject: Re: [Freesurfer] Subcortical Volume Calculation To: Hui J Yu hjyu@ic.sunysb.edu Cc: freesurfer@nmr.mgh.harvard.edu
yes, we do a partial volume correction of the volumes, accounting for the fraction of the structure in each border voxel.
cheers Bruce On Mon, 15 Feb 2010, Hui J Yu wrote:
Dear Freesurfer users,
I am wondering how are the volumes calculated from stats such
aseg.stat. When I export the aseg and do a voxel count and multiple the volume per voxel, my total volume is different from that is given by the stats. For example, the hippocampal volume given by aseg.stats for one of my controls is 4000 cm3, but when I export the aseg.mgz into matlab and do a voxel count (my original voxel volume is 1mm3 isotropic) for that labeled mask, the total volume is aorund 3000 cm3. Does freesurfer does some sort of intensity normalization or partial volume correction? Even if so, the magnitude of difference is large that I am puzzled.
Thank you very much.
freesurfer@nmr.mgh.harvard.edu
-
Bruce Fischl -
Hui J Yu