The idea is that multi-spectral segmentation should be more accurate, although
it will vary from dataset to dataset. If and when I get multi-spectral into
SPM5 or some other later SPM version, then this (I hope) should do a better
job than the old SPM2 segmentation.
Best regards,
-John
On Wednesday 03 May 2006 15:53, Mithun Mukherjee wrote:
> Hi John,
>
> The SPM manual currently says - "Note that multi-spectral segmentation
> (e.g. from a registered T1 and T2 image) is not yet
> implemented, but is planned for a future SPM version."
>
> So could this hack help in segmenting a pair of registered T1 and T2
> images for, say, a VBM study ?
>
> Thank you
> Mithun
>
> John Ashburner wrote:
> > I didn't have time to make the default segmentation multi-spectral. For
> > this reason, I left the code for the SPM2 segmentation in place. You
> > could get access to this by modifying spm_config.m at around line 212 so
> > that it includes spm_config_segment_old in the list.
> >
> > Best regards,
> > -John
> >
> > On Tuesday 02 May 2006 21:35, Marenco, Stefano (NIH/NIMH) [E] wrote:
> >> Does multi channel segmentation work in spm5?
> >>
> >> I tried but I get an error message that it cannot load the file called
> >> spm_krutil.mexglx
> >>
> >>
> >>
> >> My original problem was the following: I did a 3 channel segmentation in
> >> spm2, using images that had already been skull stripped (a T2 weighted,
> >> Fa and Mean diffusivity image) and obtained a result that assigns CSF to
> >> edges of the brain where it should not. How can I get around this?
> >>
> >>
> >>
> >> Stefano Marenco, MD
> >>
> >> GCAP, Clinical Brain Disorders Branch,
> >>
> >> NIMH
> >>
> >> 10 Center Drive, room 4S235
> >>
> >> Bethesda, MD 20892
> >>
> >> tel. (301) 435-8964
> >>
> >> fax. (301) 480-7795
> >>
> >> email: [log in to unmask] <mailto:[log in to unmask]>
|