The problem with the SPM5 segmentation is that it does not use any
knowledge of what lies outside the brain. In the new segment of SPM8,
there are tissue probability maps of skull etc, which means that the
algorithm can determine the approximate intensity distribution for bone,
and have some chance of identifying differences between bone and fluid.
In SPM5, it is quite common (depending on the contrast) for skull to
sometimes be classed as CSF. The new segmentation algorithm is
essentially the same as the one in SPM5, except with different tissue
probability maps and a bit more flexibility in the deformation model.
Best regards,
-John
On Wed, 2009-12-02 at 18:00 -0600, Dana Perantie wrote:
> I believe the "CSF" segment is really not considered CSF, but classification for "not white or gray matter". So if sinus sagitalis is classified as "CSF" then you should not be worried. I find that the "thorough clean" option helps eliminate the inclusion of this area as gray matter. I believe some people use FSL's "brain extraction tool" (BET) prior to segment to help prevent it.
> -Dana
>
> -----Original Message-----
> From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] On Behalf Of Winkler
> Sent: Friday, November 06, 2009 4:50 AM
> To: [log in to unmask]
> Subject: [SPM] vbm5 segmentation problem
>
> Hi,
>
> a problem came up during my MRI-data analysis with VBM5 of SPM5.
> In order to receive die volume of the BPF, so far, I segmented the a
> T1-MPR with the toolbox:VBM5 of SPM5, which seperates the GM,WM and CSF
> volumes. The problem is, that some of the segmented images detect e.g. the
> Sinus sagitalis superior as part of the grey matter or as CSF, which
> logically influences our volume measures.
> What is the best way to solve this problem?
> Do you have any proposals?
>
> Thank you very much in advance.
>
> Liza Winkler
>
--
John Ashburner <[log in to unmask]>
|