This behaviour is completely as expected. The segmentation is supposed to
only work with healthy adult human brains and does this by including prior
knowledge about brain structure. The model also includes other prior
knowledge about tissue of the same class having similar intensities, the
inhomogeneity field being mostly low frequency and multiplicative etc.
http://en.wikipedia.org/wiki/Ugly_duckling_theorem
http://en.wikipedia.org/wiki/Bayes_theorem
If there is prior knowledge, then it should be possible to achieve more
accurate results by using it. If someone does a manual segmentation, then
they would also use prior knowledge - but it would typically involve a more
complicated model of the data than those currently used for image
segmentation. It's difficult to make image processing algorithms as
effective as the human visual system.
How should the algorithm have behaved with the phantom?
Best regards,
-John
On Sunday 24 May 2009 03:53, Vince D. Calhoun wrote:
> Hi,
> I found this by accident, but we ran a phantom through the unified
> segmentation in SPM5. I've attached a screen shot of the input (which was
> a spherical phantom) "vbm_problem_remprage.jpg" and the output GM map
> "vbm_problem_wc1.jpg"...which looks very much like a brain (and spatially
> correlated with the prior at a value of about 0.94). This makes me worry a
> bit about the influence of the priors on the unified segmentation. Has
> anyone seen this before?
>
> VDC
>
> _____________________________________________
> Vince D. Calhoun, Ph.D.
> Director, Image Analysis and MR Research
> The Mind Research Network
>
> Associate Professor
> Depts. of Electrical and Computer Engineering,
> Neurosciences, and Computer Science
> University of New Mexico
>
> Associate Professor, Adjunct
> Dept. of Psychiatry
> Yale University
>
> MAILING ADDRESS:
> The Mind Research Network
> 1101 Yale Boulevard, NE
> Albuquerque, NM 87131
> MSC 116040
> Tel: (505) 272-1817, Fax: (505) 272-8002
> Email: [log in to unmask]
> _____________________________________________
>
> "An author [or researcher] should never conceive
> himself as bringing into existence beauty or
> wisdom which did not exist before, but simply
> and solely as trying to embody in terms of his
> own art some reflection of eternal Beauty and
> Wisdom...And always, of every idea and of every
> method the Christian will ask not, 'Is it mine?'
> but 'Is it good?'"
> ----CS Lewis----
|