[MINC-users] Mritotal - how many parameters?

[Andrew Janke]

Hi Daniel,

A 9 parameter transform is the mritotal default, and a nine parameter
fit will generally provide a good linear fit. A 6 or 7 parameter
version will often not cut the mustard.

As for calculating sizes, this will be no problem, either
back-transform the traced structures to the native images (invert the
xfms when using mincresample) or simply scale the volumes you obtain
in MNI-space by the inverse of the three scale parameters that
xfm2param outputs for scale.


a

On 6/1/06, Daniel Schwarz <schwarz at cba.muni.cz> wrote:
> Dear minc-users,
>
> I have a question which might be too elemental. However I will be happy for
> any comments:
>
> I am going to ask my colleagues-physicians to segment manually several
> features from a set of brain images. I would like them to do it in the MNI
> space with Display. Before that, I will do correction for INU artifact with
> N3 and transformation into the MNI space with the use of ICBM152 template
> and Mritotal. How many parameters of the transform should I use? If I
> perform 7-parameter transform (the same scaling factor for x,y,z), I will
> not have to worry about changing the proportions of the images and
> computations of the featrue volumes will be very easy. Are there any
> advantages when using 9-parameter transform? Is it better to use 9-parameter
> transform and then recompute the feature volumes according to the various
> scaling factors?
>
> Thank you in advance
>
> Daniel Schwarz
>
> _______________________________________________
> MINC-users at bic.mni.mcgill.ca
> http://www.bic.mni.mcgill.ca/mailman/listinfo/minc-users
>


-- 
Andrew Janke   (a.janke at gmail.com || http://a.janke.googlepages.com/)
Canada->Montreal                                   Cell: +1 (514) 924 2012