[Insight-developers] [Fwd: Re: [Insight-users] itkCurvatureRegistrationFilter]

[Torsten Rohlfing]

Hi --

could someone please put the following fixes that came up on the 
itk-users list into the CVS repository.

The first adds a trivial access method needed to get the mean squares 
registration function to work with the curvature registration filter. 
The second fixes a template instantiation/compilation error with VC++.

Thank you!
  Torsten

Index: Code/Algorithms/itkMeanSquareRegistrationFunction.h

===================================================================

RCS file: /cvsroot/Insight/Insight/Code/Algorithms/itkMeanSquareRegistrationFunction.h,v

retrieving revision 1.5

diff -r1.5 itkMeanSquareRegistrationFunction.h

145a146,150

>    /** Get the metric value. The metric value is the mean square difference

>    * in intensity between the fixed image and transforming moving image

>    * computed over the the overlapping region between the two images. */

>   virtual double GetMetric() const

>     { return this->GetEnergy(); }

Index: Code/Algorithms/itkCurvatureRegistrationFilter.h
===================================================================
RCS file: /cvsroot/Insight/Insight/Code/Algorithms/itkCurvatureRegistrationFilter.h,v
retrieving revision 1.1
diff -r1.1 itkCurvatureRegistrationFilter.h
132c132
<   typedef Image<RealTypeDFT,TDeformationField::ImageDimension> DeformationFieldComponentImageType;
---
>   typedef Image<RealTypeDFT,::itk::GetImageDimension<TDeformationField>::ImageDimension> DeformationFieldComponentImageType;




-------- Original Message --------
Subject: 	Re: [Insight-users] itkCurvatureRegistrationFilter
Date: 	Thu, 18 Aug 2005 08:51:55 -0700
From: 	Torsten Rohlfing <torsten at synapse.sri.com>
To: 	Martin Urschler <martin at urschler.info>
CC: 	insight-users at itk.org
References: 	<4303B67E.4090109 at synapse.sri.com> 
<43045E05.5040609 at urschler.info>



Martin:

You are right again. Apply the following (trivial) patch:

Index: Code/Algorithms/itkMeanSquareRegistrationFunction.h

===================================================================

RCS file: /cvsroot/Insight/Insight/Code/Algorithms/itkMeanSquareRegistrationFunction.h,v

retrieving revision 1.5

diff -r1.5 itkMeanSquareRegistrationFunction.h

145a146,150

>    /** Get the metric value. The metric value is the mean square difference

>    * in intensity between the fixed image and transforming moving image

>    * computed over the the overlapping region between the two images. */

>   virtual double GetMetric() const

>     { return this->GetEnergy(); }


I will also forward this to the developers list so someone with CVS 
write access can put it into the repository.

As a final note, I myself typically use the 
itk::FastSymmetricForcesDemonsRegistrationFunction class as the driving 
force for the curvature-based registration.

Best,
 Torsten

Martin Urschler wrote:

> hi torsten,
>
> your advice resolved my compilation problem, but there occurred 
> another one when actually using the registration filter
>
> the meansquare registration function that you are using by default for 
> computing the updates lacks the GetMetric method which leads to a 
> compiler error
> are you using a different registration function? does it compile on 
> your machine with the default one?
>
> thanks,
> Martin
>
> Torsten Rohlfing wrote:
>
>> Just a hunch - try replacing
>>
>> typedef Image<RealTypeDFT,TDeformationField::ImageDimension> 
>> DeformationFieldComponentImageType;
>>
>> with
>>
>> typedef 
>> Image<RealTypeDFT,::itk::GetImageDimension<TDeformationField>::ImageDimension> 
>> DeformationFieldComponentImageType;
>>
>>
>> It seems that everyone except me used the latter version (or 
>> itkGetStaticConstMacro()) to achieve the desired effect. If you have 
>> problems compiling itkCurvatureRegistrationFilter but everything else 
>> compiles fine, this may fix it.
>>


-- 
Torsten Rohlfing, PhD          SRI International, Neuroscience Program
Research Scientist             333 Ravenswood Ave, Menlo Park, CA 94025
 Phone: ++1 (650) 859-3379      Fax: ++1 (650) 859-2743
  torsten at synapse.sri.com        http://www.stanford.edu/~rohlfing/

    "Though this be madness, yet there is a method in't"




-- 
Torsten Rohlfing, PhD          SRI International, Neuroscience Program
 Research Scientist             333 Ravenswood Ave, Menlo Park, CA 94025
  Phone: ++1 (650) 859-3379      Fax: ++1 (650) 859-2743
   torsten at synapse.sri.com        http://www.stanford.edu/~rohlfing/

     "Though this be madness, yet there is a method in't"