Hi Martin,
the sample is small (30), and the design is longitudinal, but these structures 
are not the main focus, which is CT (already manually edited). 
Thanks for help,
Garret




Date: Thu, 24 Sep 2015 14:14:52 -0400
From: [email protected]
To: [email protected]
Subject: Re: [Freesurfer] segmentation accuracy / aseg edits


  
    
  
  
    Hi Garret,

    

    again, I am not an editing expert, but some of these look more
    severe. The question is what you need this for. If this is a
    population study with 1000 of subjects then obviously editing will
    be tough. For a small study I would recommend manual edits to make
    sure you are removing as much noise as possible. Especially if you
    focus on exactly these structures in your paper. Maybe someone with
    more editing experience will chime in.

    

    Cheers, Martin

    

    

    

    On 09/24/2015 01:30 PM, Garret O'
      Connell wrote:

    
    
      
      Hi Bruce,
        

        
        as you can see in attached image, the hippocampal label
          (brownish) seems to bleed into 3-4 rows of voxels into a
          ventrical. In the "amg" image, the amygdala volume (light
          blue) seems to "wrap" on the outside around the hippocampus a
          bit, when obviously it should be more spherical. These are the
          most common errors I find for these respective regions.
        

        
        Within ranges of acceptability?
        

        
        Thanks kindly for help,
        

          Garret

          

          

          

          
            From: [email protected]

            Date: Thu, 24 Sep 2015 11:58:18 -0400

            To: [email protected]

            Subject: Re: [Freesurfer] segmentation accuracy / aseg edits

            

            Feel free to upload a representative subject and we
              will take a look 
            Cheers
            Bruce
            

              On Sep 24, 2015, at 11:54 AM, Martin Reuter 
<[email protected]>
              wrote:

              

            
            
               Hi Garret, 

                this is not really a longitudinal question, rather what
                accuracy can be expected and when do you need to do aseg
                edits. I therefore changed the title.

                

                From what I know, a few voxels here and there will not
                change the volume measures much. Furthermore, there
                should be some partial volume correction anyway, so a
                strong intensity difference should also change the
                proportional volume. But it would be good if someone
                with more knowledge about the segmentation details
                answers.

                

                You could edit a single case and see how much difference
                that makes, I bet it will be small.

                

                Best, Martin

                

                

                On 09/24/2015 11:19 AM,
                  Garret O' Connell wrote:

                
                
                  
                  Hello FSers,
                    

                    
                    quick question on the validity of FS automatic
                      segmentation.
                    

                    
                    I have found longitudinal changes in
                      hippocampus/amygdala volumes. Upon visual
                      inspection, there are some concerning issues, such
                      as the labelling of CSF in the amygdala. Mostly,
                      these amount to a few clusters of between 5-20
                      voxels around the boundaries of these structures
                      in around 15-20 % of slices. 
                    

                    
                    My question is do these errors sounds
                      abnormally high (and don't think so, having seen
                      examples in some lit about FS accuracy), and are
                      these errors acceptable (publishable with a mild
                      degree of confidence) in the context of a
                      longitudinal design, given the high test-retest
                      reliability of FS segmentation? Or are edits
                      necessary?
                    

                    
                    I know there are probably no right/wrong
                      answers on this, but any advice would be greatly
                      appreciated.
                    

                    
                    Best,
                    

                    
                    Garret

                      

                      

                    
                  
                  

                  
                  

                  _______________________________________________
Freesurfer mailing list
[email protected]
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
                
                

                -- 
Martin Reuter, PhD
Assistant Professor of Radiology, Harvard Medical School
Assistant Professor of Neurology, Harvard Medical School
A.A.Martinos Center for Biomedical Imaging
Massachusetts General Hospital
Research Affiliate, CSAIL, MIT
Phone: +1-617-724-5652
Web  : http://reuter.mit.edu 
              
            
            
              _______________________________________________

                Freesurfer mailing list

                [email protected]

                https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
            
            

            _______________________________________________
            Freesurfer mailing list
            [email protected]
            https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
            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.
        
      
      

      
      

      _______________________________________________
Freesurfer mailing list
[email protected]
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
    
    

    -- 
Martin Reuter, PhD
Assistant Professor of Radiology, Harvard Medical School
Assistant Professor of Neurology, Harvard Medical School
A.A.Martinos Center for Biomedical Imaging
Massachusetts General Hospital
Research Affiliate, CSAIL, MIT
Phone: +1-617-724-5652
Web  : http://reuter.mit.edu 
  


_______________________________________________
Freesurfer mailing list
[email protected]
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer


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.                                    
_______________________________________________
Freesurfer mailing list
[email protected]
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer


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.

Reply via email to