I don't have any good ideas about how to fix this one. The image has
tissue with similar intensity to GM, in a place that is fairly close
to where you'd expect to find GM. Fully automated segmentation
algorithms are simply not yet 100% reliable (largely because of a lack
of good training data).
Sometimes, it is worth trying various approaches to see what is
successful for any particular dataset. Where one segmentation model
fails, it is sometimes possible that another one will work. It
depends on the modelling assumptions used, and whether your data are a
good fit that particular model.
On 27 May 2011 18:42, Xin Di <[log in to unmask]> wrote:
> Dear experts,
> I am using the new segment tool to run segmentations on several MRI images.
> For one of the subject, the tissues outside the cortices are mis-classified
> as gray matter and warped into the cortices after normalization (please see
> attached figures). I use the default settings of number of Gaussians for
> each tissue type. Can someone kindly give me some suggestions on how to
> avoid this mis-classification? Thank you!
> Xin Di, PhD
> Postdoctoral Researcher
> Department of Radiology
> University of Medicine and Dentistry of New Jersey
> 30 Bergen Street, ADMC 575
> Newark, NJ 07101