The masking is only for estimating the parameters, such as the intensity
distributions of the tissue classes, the bias fields and the deformations.
When the tissue class images are written, all the voxels are assigned some
value.
If you do a literature search, you'll find a few segmentation algorithms that
deal with such lesions. Unfortunately, I don't think that many of these
algorithms have been packaged up and made publicly available (except for EMS,
but this is based around a very old version of SPM).
Best regards,
-John
On Monday 23 March 2009 17:22, Eytan Raz wrote:
> Dear SPM experts,
>
> I am using VBM for multiple sclerosis (MS) patients.
>
> After having done the lesion load, I prepared binary lesion masks of each
> patient.
> Using SPM5, I run <segment>, and in the <custom>, the last option <masking
> image>, I add the lesion mask previously created.
> However, with great astonishment, after segmentation I checked the wc1
> images and the lesions that were included in the masks, were not excluded;
> instead they were considered part of GM.
> Why?
> Where is the error?
> Does anyone have the same experience?
> thanks, Eytan
|