[MINC-users] more minctracc questions!

[Andrew Janke]

Hi,

the following is from 'man minctracc':

-w_translations <w_tx> <w_ty> <w_tz>: Optimization weight of
translation in x, y, z (default =  1.0  1.0 1.0).

-w_rotations  <w_rx> <w_ry> <w_rz>: Optimization weight of rotations
around x, y, z (default = 0.0174533
0.0174533 0.0174533).  Internally, the rotations are stored as
radians, although all user  input  is  in
degrees.  The value 0.0174533 makes one degree equivalent to 1 mm in
the optimization.

So to expand on this, these values are the inital amounts that will be
used in the optimiser for each iteration.  ie: translate 1.0mm in x, y
or z per iteration..  For rotation the amount given roughly translates
to the amount of rotation required for a movement of 1mm at the
surface of a "head shaped" object.

So yes, you will have to modify a bunch of parameters but it will
work!  As an initial quick hack, I suggest you scale your mouse heads
to roughly human size (use param2xfm and mincresample, or just
mincedit and change the step and start in the header) and get things
working.  Once this is good then scale all the values for the -w_
params and -step.

The simplex size is expressed in terms of -step so you should not have
to change this.


a


On 9/13/07, mojdeh.zamyadi at utoronto.ca <mojdeh.zamyadi at utoronto.ca> wrote:
> Hi all,
>
> Can someone please explain what exactly '-w_rotation' and
> 'w_translation' mean in minctracc? I know these are weight parameters
> that are used in the simplex optimization method that is used in
> minctracc, but I was wondering if anyone can give a more detailed
> description of these params.
>
> I am trying to register mouse embryos using minctracc. Since the
> parameters in minctracc has been optimized for human brain (right?!) I
> have to change them since mouse embryos are about 20 times smaller
> than human brain. The other thing is that, unlike brains, these
> embryos can be in very different orientations (eg. one can face left
> and the other one can be upside down and facing right!). Considering
> the scale and orientation, can someone suggest how I should optimize
> these params for registration of mouse embryos?
>
> Also, how are -step and -simplex values determined.
>
> Sorry for so many questions ;) and thanks in advance :)
>
> Mojdeh
>
>
> ###################################
> Mojdeh Zamyadi
> MSc Candidate
> Mouse Imaging Centre (MICe)
> Hospital for Sick Children
> Toronto Centre for Phenogenomics
>
> _______________________________________________
> 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/)
Canberra->Australia    +61 (402) 700 883