You could try various skull-stripping approaches, but most of these are likely
to chop out much of the CSF - so the estimates are not so useful. I am
currently working on segmentation that does a better job of dealing with CSF
(among other things), but this won't be ready for a while. Others have used
multiple-channels of MRI data (eg T2- and PD-weighted) to obtain a cleaner
segmentation of CSF. Single channel segmentation sometimes works well, but
there needs to be a good separation between the intensity distribution of
bone and that of CSF.
Best regards,
-John
On Monday 07 April 2008 17:07, Benson, Randall wrote:
> List members,
>
> I am trying to look at age effects on global grey, white and CSF volumes.
> I can get reliable measurements of GM and WM but CSF is not reliable using
> segmentation in SPM5. Does anybody know a good way to do it either with
> SPM or with another program, e.g., FSL? I would greatly appreciate ANY
> help on this.
>
> Randy
>
> ------------------------------------------------------------------------
> Randall R. Benson, M.D.
> Assistant Professor of Neurology
> Wayne State University School of Medicine
> UHC-8D 4201 St. Antoine
> Detroit, MI 48201
> Tel: 313-993-2726
> Fax: 313-745-4216
>
>
>
> -----Original Message-----
> From: Benson, Randall
> Sent: Fri 4/4/2008 9:47 AM
> To: John Ashburner; [log in to unmask]
> Subject: RE: [SPM] Segmentation-Results display
>
> So, John and fellow SPMers,
>
> Since we are trying to use SPM5 to do volumetric analysis of WM, GM and
> CSF, specifically looking at age effects and possibly longitudinal analysis
> in dementia and TBI patients, we really need either a reliable measure of
> CSF or at least a good measure of intracranial cavity volume (without dura,
> skull, etc.). Is there a good way of doing this in SPM proper or
> extensions of SPM? How have people done these analyses?
>
> Randy
>
> ------------------------------------------------------------------------
> Randall R. Benson, M.D.
> Assistant Professor of Neurology
> Wayne State University School of Medicine
> UHC-8D 4201 St. Antoine
> Detroit, MI 48201
> Tel: 313-993-2726
> Fax: 313-745-4216
>
>
>
> -----Original Message-----
> From: SPM (Statistical Parametric Mapping) on behalf of John Ashburner
> Sent: Fri 4/4/2008 9:17 AM
> To: [log in to unmask]
> Subject: Re: [SPM] Segmentation-Results display
>
> > We are trying to do volumetric analysis from segmentation using SPM5. We
> > would like to get accurate measurements of GM,WM and CSF volumes. We are
> > running into some problems.
>
> CSF volumes are not likely to be very accurate, as there are no tissue
> probability maps for non-brain tissues. Maybe the skull-stripped version
> would help here - but be careful not to strip out too much CSF.
>
> > 1) We are finding that when we skull strip the MPRAGE, the segmentation
> > algorithm never gives a solution. Has anybody else observed this?
>
> Depending how the skull-stripping is done, it is possible that the
> orientation information in the headers is being lost, or filled with junk.
>
> > 2) When we apply a mask of the brain to the MPRAGE image in
> > segmentation, which of the 3 options (data, probability map, mask) should
> > be used?
>
> Masking is just for cases where there are lesions in the brain, which you
> don't want to influence the way that GM, WM etc in other parts of the brain
> is segmented.
>
> > 3) There is no result displayed on the graphics window after the
> > segmentation program is done- Does anybody know why this is the case?
>
> Because I was too lazy to write the code for displaying the results. It is
> easy to check the results with the Check Reg routine, and this allows you
> to display the images that you consider most useful.
>
> > 4) There is no PS file written in the working directory after the
> > segmentation program is done.- Does anybody know why this is the case?
>
> Again, because I was too lazy.
>
> Best regards,
> -John
|