[Insight-developers] Exception in itkImageToImageFilter.txx for itkPhilipsRECImageIOOrientationTest

[Don C Bigler]

A bug might exist.  However prior to adding VerifyImageInformation the
code I created in the test produced the desired result.  Additionally,
I've been using the same code routinely to do image fusion of images
acquired during a single scan session.  It works not just for the
itkPhilipsRECImageIO class but for DICOM images as well.

I should point out that I discovered that I needed to use the identity
matrix by accident because when I first created the test I tried to
create the transform manually.  It failed until I used the identity
matrix.  Note that I call UseReferenceImageOn() in the resample
filter.  I might be missing something here, but when I looked at the
resample code it appeared to me that ITK does transforms in real world
coordinates, not image coordinates.  So my test appears to be a
special case that only works when the target and reference images are
acquired during a single scan session.

On Thu, Jul 21, 2011 at 4:24 PM, Bill Lorensen <bill.lorensen at gmail.com> wrote:
> So itr is probably a bug somewhere. The physical spaces should verlap.
>
> The VerifyImageInformation is a recent addition.
>
> On Thu, Jul 21, 2011 at 5:12 PM, Don C Bigler <d.bigler1 at gmail.com> wrote:
>>
>> All,
>>  I am getting the following exception in
>> itkPhilipsRECImageIOOrientationTest after updating to the latest
>> Master branch:
>>
>> 627/1769 Testing: itkPhilipsRECImageIOOrientationTest2
>> 627/1769 Test: itkPhilipsRECImageIOOrientationTest2
>> Command:
>> "/Users/dbigler/Development/ITKV4Build/bin/ITK-IO-PhilipsRECTestDriver"
>> "--compare"
>> "/Users/dbigler/Development/ITKV4Build/ExternalData/Modules/IO/PhilipsREC/test/Baseline/itkPhilipsRECImageIOOrientationTest2.mha"
>>
>> "/Users/dbigler/Development/ITKV4Build/Testing/Temporary/itkPhilipsRECImageIOOrientationTest2.mha"
>> "itkPhilipsRECImageIOOrientationTest"
>>
>> "/Users/dbigler/Development/ITKV4Build/ExternalData/Modules/IO/PhilipsREC/test/Input/Test_Orientation_2_5.PAR"
>>
>> "/Users/dbigler/Development/ITKV4Build/ExternalData/Modules/IO/PhilipsREC/test/Input/Test_Orientation_2_1.REC.gz"
>>
>> "/Users/dbigler/Development/ITKV4Build/Testing/Temporary/itkPhilipsRECImageIOOrientationTest2.mha"
>> Directory:
>> /Users/dbigler/Development/ITKV4Build/Modules/IO/PhilipsREC/test
>> "itkPhilipsRECImageIOOrientationTest2" start time: Jul 20 23:18 CDT
>> Output:
>> ----------------------------------------------------------
>>
>> itk::ExceptionObject (0x1012cafc0)
>> Location: "void itk::ImageToImageFilter<TInputImage,
>> TOutputImage>::VerifyInputInformation() [with TInputImage =
>> itk::Image<short int, 3u>, TOutputImage = itk::Image<short int, 3u>]"
>> File:
>> /Users/dbigler/Development/ITK/Modules/Core/Common/include/itkImageToImageFilter.txx
>> Line: 253
>> Description: itk::ERROR: ResampleImageFilter(0x1012b72b0): Inputs do
>> not occupy the same physical space!
>> InputImage Origin: [-107.993, -73.195, 114.799], InputImage1 Origin:
>> [-132.601, -135.137, -62.593]
>> InputImage Spacing: [0.807, 0.807, 1], InputImage1 Spacing: [1, 1, 1]
>> InputImage Direction: 1 0 0
>> 0 0 1
>> 0 -1 0
>> , InputImage1 Direction: 1 0 0
>> 0 1 0
>> 0 0 1
>>
>>
>>
>>
>> <DartMeasurement name="ImageError"
>> type="numeric/double">0</DartMeasurement>
>> <DartMeasurement name="BaselineImageName"
>>
>> type="text/string">itkPhilipsRECImageIOOrientationTest2.mha</DartMeasurement>
>> <end of output>
>> Test time =   1.31 sec
>>
>> It looks like a check is made to verify that the origin and direction
>> cosines are the same.  Probably not a bad check for most situations,
>> but it appears to break this test that I created to test the
>> orientation of the Philips PAR/REC images.  Here's what I'm doing in
>> the test:
>>
>> 1.  I have two images acquired during the same scan session, one in an
>> axial orientation and the other is the same image in the sagittal
>> orientation created using multi-planar reconstruction.
>> 2.  Read both images.
>> 3.  Transform the sagittal image to the axial image using the identity
>> affine matrix.  The identity transform is used because the images
>> differ in orientation only.
>> 4.  Subtract the transformed sagittal image from the axial image.
>> Visual inspection using an image viewer shows that they are correctly
>> coregistered.
>>
>> The reason they are correctly coregistered is because ITK handles the
>> conversion from image coordinates to real world coordinates
>> "automatically" using the origin and direction cosines.
>>
>> Having explained all this (hopefully with adequate clarity and
>> correctness), it seems to me that this exception that is generated is
>> a bug.  I would appreciate it if someone else will verify the
>> exception that I'm getting and provide an explanation as to why this
>> check was added in ITK4.
>>
>> Don
>> _______________________________________________
>> 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://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-developers
>
>