[Insight-users] Problem with TransformPoint

[brian avants]

hi clemens

the transform, when applied to resample a moving image to a fixed
image,  calls transformpoint on each fixed point to get the
corresponding moving point.

you might want to try using the inverse of your transform when mapping
from the moving space to the fixed space.  i.e. your code should be
almost the same except that you use the inverse transform.


brian





On Thu, Mar 1, 2012 at 8:39 AM, Clemens M. Hentschke
<cmh at isg.cs.uni-magdeburg.de> wrote:
> Hi,
>
> I want to transform points with the TransformPoint() method from one
> coordinate system (moving image) to another coordinate system (fixed image).
> So far, I checked the transformation works fine by applying it to the image,
> the result image was as I expected correct.
> As I am only interested in single pixels (and their respective transformed
> coordinates), I only used the TransformPoint() method of
> Similarity3DTransform insted of transforming the whole image.
> I converted from pixel to physical coordinates on the moving image,
> transformed this coordinates and converted the physical coordinates to pixel
> coordinates on the fixed image.
> However, the pixels from the moving image do not correspond to the fixed
> image.
> I use ITK 3.20 with Windows 7 and the MinGW compiler.
> Has anyone an idea what the problem may could be?
> Thanks in advance.
>
> I attached a code snippet:
>
> typedef TransformType::InputPointType CenterType;
>
>    typedef itk::Similarity3DTransform<double> TransformType;
>    TransformType::Pointer transform =TransformType::New();
>
>      CenterType center;
>      center[0] = 3.3874403492;
>      center[1] = -17.7840881162;
>      center[2] = 50.2830146457;
>      transform->SetCenter(center); // setting the center coordinates
> manually
>
>      readerMoving->Update();
>      readerFixed->Update();
>      ImageType::Pointer mImage = readerMoving->GetOutput(); //the moving
> image
>      ImageType::Pointer fImage = readerFixed->GetOutput(); //the fixed image
>
>      TransformType::ParametersType
> paramsInput(transform->GetNumberOfParameters());
>      paramsInput[0] = 0.244805;
>      paramsInput[1] = 0.009105;
>      paramsInput[2] = -0.056415;
>      paramsInput[3] = -7.963562;
>      paramsInput[4] = -29.462040;
>      paramsInput[5] = 61.358929;
>      paramsInput[6] = 0.981010;
>
>      transform->SetParameters(paramsInput); //setting the parameters
> manually
>
>      ImageType::IndexType indexIn;
>      indexIn[0] = 162;  //some sample pixel from the moving image
>      indexIn[1] = 218;
>      indexIn[2] = 81;
>
>      TransformType::InputPointType pt;
>
>      mImage->TransformIndexToPhysicalPoint(indexIn, pt);
>
>      TransformType::OutputPointType output =  transform->TransformPoint(pt);
>
>      ImageType::IndexType indexOut;
>      fImage->TransformPhysicalPointToIndex(output, indexOut);
>
>      for (int i=0; i<3; i++)
>          cout << indexOut[i] << endl; //the output coordinates are obviously
> wrong, they do NOT match to the input (moving image) coordinates
> _____________________________________
> 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