[Insight-users] Wich one is the correct Deformable Registration ?

[Luis Ibanez]

Hi Javier,


For the endoscopic images of uterine cancer that you are processing,
you probably want to use the BSpline deformable transform.


Examples on how to use this code are available at:


        Insight/Examples/Registration/
                   DeformableRegistration4.cxx
                   DeformableRegistration6.cxx
                   DeformableRegistration7.cxx
                   DeformableRegistration8.cxx


The optimizer used in the examples:

                   DeformableRegistration7.cxx
                   DeformableRegistration8.cxx

is known to give better results than the one used in 4,6.


You should simply try these two examples, by reducing
the image Dimension to 2.


Note also that this examples use MattesMutualInformation,
which is not essential in your case since your two images
are of the same modality.  MattesMutualInformation should
work here, but you still have the choice of many other
metrics available in ITK.



For a full account of the metrics and other registration
classes available in the toolkit you should read the
registration chapter from the ITK Software Guide:


    http://www.itk.org/ItkSoftwareGuide.pdf


In this document you will find details about the standard
deviation to be used in the MutualInformation metric.



Regards,


    Luis



----------------------------------
javier silva bravo wrote:

> Hi My name is Javier, I from Mexico and work at www.inaoep.mx 
> <http://www.inaoep.mx> , I try to analize medical Images so I want to 
> know which is the best Deformable Registration for my Images.
> I send a both of examples.
> I use itk-2.0.0 and run examples within it.
>  
> thanks.
>  
> Also somebody know what and which are the Standard deviation for 
> MutualInformationImageToImageMetric ??
>  
>