[Insight-users] Problem with BSplines-Registration and Mutual Information Metric

[Luis Ibanez]

Hi Carola,

It certainly looks like you need to fine tune the parameters of the
MutualInformation metric.

Could you tell us more about the images that your are registering ?

1) Image modality
2) Image size (in pixels along X, Y, Z )

Have you connected an observer to the optimizer ?

If so,

Could you please post to the list the output values of  Image metric values
at every iteration of the registration.

Also, one of the critical aspects of using MattesMI with a BSpline transform
is that you have to make sure that there are enough samples for population
*every* cell of the BSpline grid.

Are you using the same BSpline grid parameters of
DeformableRegistration8.cxx ?

---

You may want to try using all the samples from the image.

If you have turned ON the option ITK_USE_OPTIMIZED_REGISTRATION_METHODS,
you could use the API:

     metric->UseAllPixelsOn();


Please let us know about the questions above,


    Thanks


           Luis


-----------------------------------------------------------------------------------------------
On Fri, Dec 19, 2008 at 11:00 AM, Carola Schmidt <
carola.schmidt at igd.fraunhofer.de> wrote:

> Hi, itk users.
>
> I am implementing the BSpline Registration to register CT-Data of the
> abdomen. I use the LBFGSBOptimizer
> and use the parameters of the deformableRegistration8.cxx file
>
>   OptimizerType::BoundSelectionType
> boundSelect(transform->GetNumberOfParameters());
>   OptimizerType::BoundValueType
> upperBound(transform->GetNumberOfParameters());
>   OptimizerType::BoundValueType
> lowerBound(transform->GetNumberOfParameters());
>
>   boundSelect.Fill(0);
>   upperBound.Fill(0.0);
>   lowerBound.Fill(0.0);
>
>   optimizer->SetBoundSelection(boundSelect);
>   optimizer->SetUpperBound(upperBound);
>   optimizer->SetLowerBound(lowerBound);
>
>   optimizer->SetCostFunctionConvergenceFactor( 1e+7 );
>   optimizer->SetProjectedGradientTolerance( 1e-4 );
>   optimizer->SetMaximumNumberOfIterations( 500 );
>   optimizer->SetMaximumNumberOfEvaluations( 500 );
>   optimizer->SetMaximumNumberOfCorrections( 12 );
>
> As metrics I tried itkMeanSquaresImageToImageMetric and
> itkMattesMutualInformationImageToImageMetric. When I use the
> itkMeanSquaresImageToImageMetric the registration takes really long but the
> results are quite ok.
> When I use itkMattesMutualInformationImageToImageMetric the registration
> only takes a few seconds but most of the time the warped image looks like a
> hardly deformed moving image. Since the registration time is very short, it
> seems that the registration stops after a few steps and warps the image
> then. I use those parameters for the mututal information:
>
>   metric->SetNumberOfHistogramBins( 40 );
>   const unsigned int numberOfSamples = 10000;
> //fixedRegion.GetNumberOfPixels()/ 10;
>   metric->SetNumberOfSpatialSamples( numberOfSamples );
>   metric->ReinitializeSeed( 76926294 );
>   metric->SetMovingImage(m_itkimage);
>   metric->SetFixedImage(m_itkatlas);
>   metric->SetFixedImageRegion(fixedRegion);
>
> Why does the Mutual Information Metric not work as expected? Should I
> change the parameters of the optimizer?
> Thanks for an answer.
>
> Ciao,
> Carola
> _______________________________________________
> Insight-users mailing list
> Insight-users at itk.org
> http://www.itk.org/mailman/listinfo/insight-users
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.itk.org/pipermail/insight-users/attachments/20081221/6244f0b3/attachment.htm>