|
Many thanks Paul,
Can I ask a more specific question: are the ventricles completely
surrounded by a thin layer of GM? That appears to be the case looking
at the GM TPM. I'm not sure whether your comment about the caudate
nucleus "lining" the ventricles has already answered this question, or
if you just meant that it's alongside parts of the ventricles.
Presumably if the TPMs are expert-segmented then partial-volume
effects shouldn't be too important, although they of course will be in
SPM/FAST/etc.
Thanks again,
Ged.
Paul Laurienti wrote:
> Andreas and Ged,
>
> Your question is pretty interesting and is likely a combination of real anatomy and some volume averaging. For example, if you look at the template image that was attached in Ged's email the crosshairs are located directly on the body of the caudate nucleus. This structure is definitely gray matter and lines the ventricles. There is also the stria terminalis adjacent to the caudate that contains gray matter. Both of these structures would be located on the floor of the body of the lateral ventricle in a coronal sections but are in the roof of the temporal horn of the lateral ventricles. The fornix is located adjacent to the roof in the body of the lateral ventricles and includes considerable amounts of non-white matter tissue including choroid plexus that may be segmented into gray matter. Of course as Ged noted, volume averaging between white matter and csf may play a role as well.
>
> Paul
>
>
> ------------
> Paul J. Laurienti, MD, PhD
> Department of Radiology
> Wake Forest University School of Medicine
> Medical Center Blvd
> Winston-Salem, NC 27157
> 336-716-3261
> [log in to unmask]
> www.fmri.wfubmc.edu/laurienti.htm
>
> -----Original Message-----
> From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] On Behalf Of DRC SPM
> Sent: Friday, March 10, 2006 6:55 AM
> To: [log in to unmask]
> Subject: Re: [SPM] Segmentation question
>
> Hi Andreas, everyone,
>
> I can't offer much help, but I can confirm that this occurs with SPM5 too. The attached screengrab shows the native and the modulated-normalised segmentations of the avg152T1 brain, using the default parameters (except for changing template to average-sized, since the image and the template should match pretty well here...).
>
> Andreas, I don't know whether you used a smoothed version of the
> avg152 (as in the spm5/templates dir) but I've used the unsmoothed version here, as you can see there is definitely GM around the ventricles.
>
> I too had heard that this shouldn't be the case, and my explanation/guess was that the partial volume effect of blending CSF and WM was resulting in an average intensity that looked like GM, hence SPM5's segmentation (incidentally, FSL FAST gives similar results, see third attachment).
>
> What's confusing me now though is that I looked at the GM TPM (screenshot attached), and that also shows GM around the ventricles -- I was under the impression that this was the result of expert manual segmentation, so are you certain that there shouldnt' be any grey around the ventricles (I'm not! But hopefully there's at least one person reading this who's brave enough to step forward as a neuroanatomy expert...)
>
> As for whether this is a problem, well, pragmatically, if your spm results have significant GM change where you expect no GM then you're free to ignore those areas, but of course, it would be nice if one didn't have to do too much "interpretation"
> (censoring/fishing/smooth-talking) for the results...
>
> Interested to hear more on this,
> Ged.
>
>
> On 09/03/06, Andreas Jansen <[log in to unmask]> wrote:
>> Dear list -
>>
>> a question about the segmentation algorithm in SPM2.
>>
>> I segmented some T1-weighted images. Everything looked fine, expect
>> that - according to the segmented images - the ventricles are
>> surrounded by a closed ring of grey matter. I was told that this does
>> not reflect the typical neuroanatomy.
>>
>> I was not sure whether this is something that arises only in my images
>> or generally in all images, so I took the canonical SPM single-subject
>> T1-template and segmented it as well. Again, the ventricles were
>> surrounded by gray matter.
>>
>> I am just asking me whether this can become a problem in e.g. VBM studies?
>>
>> Best wishes, Andreas
>>
>> --
>> Echte DSL-Flatrate dauerhaft für 0,- Euro*!
>> "Feel free" mit GMX DSL! http://www.gmx.net/de/go/dsl
>>
>
|
