Hi. I presume that you mean for multi-channel segmentation with mfast.
Indeed, this is a good question. For the main (EM-based) iterations with
Gaussian mixture-model fitting, it doesn't make any difference if you
arbitrarily scale one image. For the k-means-based initialisation, it
_does_, as we don't "normalise" the images first. I just tried a 2-channel
segmentation, running twice; the second time I multiplied one of the
channels by a factor of 10. The initial segmentation was then quite
different, but by the end of the EM iterations, the EM had converged to
virtually identical results.
You could "specify" the initial scaling of the inputs using avwmaths ;-)
Yes, there is a separate bias field estimated for each input image.
Cheers, Steve.
On Mon, 6 Oct 2003, Adolf Pfefferbaum wrote:
> How does the relative scaling of the inputs affect the computation? For
> instance if one supplies spgr, an early echo and late echo spin-echo set
> of inputs, are they equally weighted or weighted by their native
> magnitude?
>
> Can one specify the weighting of the inputs?
>
> Is a separate bias field estimated for each of the inputs?
>
> Thanks,
>
> Dolf
>
>
> ____________________
> Adolf Pfefferbaum,MD
> Neuroscience Program
> SRI International
> 650-859-2927 phone
> 650-859-2743 FAX
>
>
> CONFIDENTIALITY NOTICE: Information contained in this message and any
> attachments is intended only for the adressee(s). If you believe that you
> have received this message in error, please notify the sender immediately
> by return electronic mail, and please delete it without further review,
> disclosure, or copying.
>
Stephen M. Smith MA DPhil CEng MIEE
Associate Director, FMRIB and Analysis Research Coordinator
Oxford University Centre for Functional MRI of the Brain
John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK
+44 (0) 1865 222726 (fax 222717)
[log in to unmask] http://www.fmrib.ox.ac.uk/~steve
|