Since it has be reiterated many times that aseg.stats is calulated using estimates of partial volumes that increase accuracy over obtaining volumes from binarized masks, I wonder whether I might still incorporate this advanced procedure while still bisecting a previous segmentation by some arbitrary plane, say z=27.

I propose the following procedure as an example for splitting the left amygdala (for example) and still using the partial volume estimation:

  1. Make copies of aseg.mgz, one each for the partial segmentation you want to create (e.g. aseg_anterior.mgz and aseg_posterior.mgz, to split the amygdala by the plane z =27 on some particular brain, by some particular protocol).
  2. Edit  aseg_posterior.mgz (e.g. by tkmedit) to remove all amygdala anterior to z = 27. Similarly edit aseg_anterior.mgz to remove posterior segments.
  3. Run the stats on these edited segmentations, and record the data in aseg_posterior.stats and aseg_anterior.stats
Does this sound reasonable and wise procedure? If so, is there a quick way to remove the unwanted segmentations without manually using a brush, slice by slice in tkmedit? Loading it into Matlab, like Doug suggests, would allow the edits I think, but then I'd have to rewrite the file from matlab.. I think.

Josh





On Wed, Nov 30, 2011 at 2:49 PM, Bruce Fischl <fischl@nmr.mgh.harvard.edu> wrote:
yes

On Wed, 30 Nov 2011, Joshua Lee wrote:

Actually, I have one last-last question.
If I use mri_binarize on my aseg.mgz file, does it remain in the same
'space'? Meaning, if I identify a z-coordinate of interest in aseg.mgz, will
that z -coordinate still be valid in nifti file i'd have mri-Binarize
create?

Josh


On Wed, Nov 30, 2011 at 2:32 PM, Joshua Lee <jkilee@ucdavis.edu> wrote:
     I'm glad you caught that because I suddenly got really confused!
     Thank Bruce for all your hard work!
     Josh
     -


On Wed, Nov 30, 2011 at 2:21 PM, Bruce Fischl
<fischl@nmr.mgh.harvard.edu> wrote:
     oops, I meant the aseg.stats, not the aseg.mgz

On Wed, 30 Nov 2011, Joshua Lee wrote:

     Hi Bruce,

     I'll see if I can phrase this right

     In aseg.mgz each region has essentially been given a
     unique number (e.g. 17
     for left hippocampus, for example). ROI taken from
     aseg.mgz volumes are
     non-probabilistic, yes-no affairs already. 
     MRI_binarize just pulls this out
     when we want to look at particular regions, say left
     hipp???

     Josh




     On Wed, Nov 30, 2011 at 1:39 PM, Bruce Fischl
     <fischl@nmr.mgh.harvard.edu>
     wrote:
          probably not, as the aseg.stats calculation
     uses partial volume
          estimates.

          cheers
          Bruce

     On Wed, 30 Nov 2011, Joshua Lee wrote:

          Hey Doug, THanks. That looks simpler. I'll
     probably use
          that.

          Question: Will the total volume of a ROI taken
     from
          aseg.mgz correspond to
          the volume given in aseg.stats?

          Josh



          On Wed, Nov 30, 2011 at 9:55 AM, Douglas N
     Greve
          <greve@nmr.mgh.harvard.edu>
          wrote:
               If you have matlab, you can load the aseg
     into
          matlab, like
               aseg = MRIread('aseg.mgz');
               vol = length(aseg.vol(:,:,n:256) == ID);
               where ID is the code number for the ROI
     you want  as
          found in
               $FREESURFER_HOME/FreeSurferColorLUT.txt
               doug

               Joshua Lee wrote:
          Hi All,

          I would like to determine the volumes of a
     segmented
          region
          below or above an arbitrary line in one of the
     planes. For
          example, what would be the remaining volume of
     a label
          when z >=
          nth slice, where I would have previously
     determined the
          appropriate value of n.



          Thanks,
          Josh
        
      ------------------------------------------------------------------------

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


          --
          Douglas N. Greve, Ph.D.
          MGH-NMR Center
          greve@nmr.mgh.harvard.edu
          Phone Number: 617-724-2358 Fax: 617-726-7422

          Bugs:
     surfer.nmr.mgh.harvard.edu/fswiki/BugReporting
          FileDrop:
        
      www.nmr.mgh.harvard.edu/facility/filedrop/index.html



          The information in this e-mail is intended only
     for the
          person to whom
          it is
          addressed. If you believe this e-mail was sent
     to you in
          error and the
          e-mail
          contains patient information, please contact
     the Partners
          Compliance
          HelpLine at
          http://www.partners.org/complianceline . If the
     e-mail was
          sent to you
          in error
          but does not contain patient information,
     please contact
          the sender
          and properly
          dispose of the e-mail.