Hi there,
I often have a similar problem using an earlier version of bet. I found that
it could be simply fixed by applying a 3D bounding box mask which removes
everything outside a rectangular box which encloses the brain.
This of course requires some user-intervention, but for my particular
application (visualisation in mri3dX) that isn't a big hassle.
Of course, if you need to automatically skull-strip lots of volumes, without
user-intervention, the bounding-box approach isn't helpful.
All the best,
Krish
On 11/8/03 10:48 am, "Christopher Joseph Bailey" <[log in to unmask]> wrote:
> Dear All,
>
> I and others using a 1.5T Siemens Vision scanner have found that with
> our (common) MPR parameters the automatic brain extraction most often
> fails. The symptom is without exception left over non-brain tissue just
> anterior to the pons and inferior to the temporal cortices (will send
> .gif if needed).
>
> I have found a work-around using a modified version of betpremask, after
> which bet (-f 0.4) works fine.
>
> However, after reading the BET section in a MEDx manual, it strikes me
> that the "cure" for my misbehaving bet could be the adjustement of the
> initial brain/non-brain thresholding ("10% of the way between the 2% and
> 98% histogram values").
>
> Is it possible that the centre of gravity of the tesselated sphere is
> consistently misplaced because of gray level properties of our MPRs? If
> so, would it be possible to add a parameter to the BET command line
> specifying a different threshold/histogram parcellation?
>
> Best regards,
>
> Christopher Bailey
--
Dr K.D. Singh
Senior Lecturer and Convenor of the Neuroimaging Research Group
Neurosciences Research Institute
Aston University, Birmingham B4 7ET, U.K.
Tel/Fax: +44 (0)121 [359 3611 ext 5176/5190]/[333 4220]
[log in to unmask], http://www.aston.ac.uk/lhs/staff/singhkd/
|