[Insight-users] Mean squares metric

[Luis Ibanez]

Hi Gerald,


The fact that you are getting QNAN values, leads to suspect that
the metric is dividing by zero, and the typical division in this metric
is by the number of pixels inside of the fixed image region.

It would seem that you are passing an empty FixedImageRegion
to the registration class.


Regarding your choices of components, the MeanSquares metric
is a reasonable choice for same-modality registration.

You could use almost any optimizer in this case.

Most of our examples use RegularGradientDescent, and GradientDescent,
but at the end, what is most important is that you pick an optimizer that
your are familiar with.  In this way it will be easier for you to fine tune
its parameters.


      Regards,


             Luis


-----------------------------------------------------------
On Tue, Nov 24, 2009 at 4:36 AM, Lodron, Gerald
<Gerald.Lodron at joanneum.at> wrote:
> Hi,
>
> I am making 3D image registration of CT images where it is possible that the images have different sizes/spacings. Because of the same modality and performance reasons i want to use the mean squares metric with nearest neighbour interpolator. My printed iteratin event says following:
>
>
> Iteration XX    Value 1.#QNAN   Position[*1.#QNAN, etc]
>
> Does anyone know what i made wrong. Any suggestions which optimizer would work best?
>
> Best regards
>
> _____________________________________
> Powered by www.kitware.com
>
> Visit other Kitware open-source projects at
> http://www.kitware.com/opensource/opensource.html
>
> Kitware offers ITK Training Courses, for more information visit:
> http://www.kitware.com/products/protraining.html
>
> Please keep messages on-topic and check the ITK FAQ at:
> http://www.itk.org/Wiki/ITK_FAQ
>
> Follow this link to subscribe/unsubscribe:
> http://www.itk.org/mailman/listinfo/insight-users
>