[Insight-users] combining output from segmentation with label map attributes using Gaetan Lehmann's paper: "Label object representation and manipulation with ITK"

Luis Ibanez luis.ibanez at kitware.com
Fri Dec 11 11:20:55 EST 2009


Hi John,

Have you saved the output of the connectedThreshold filter to
a file, and verified that indeed it contains the segmentation of
the Ventricles ?

Alos, you are instantiating the connectedThreshold filter by
using as output an image of pixel type Float.

  typedef   float           InternalPixelType;
  const     unsigned int    Dimension = 3;
  typedef itk::Image< InternalPixelType, Dimension >  InternalImageType;

  typedef itk::ConnectedThresholdImageFilter<
     InternalImageType,
     InternalImageType > ConnectedFilterType;


Do you have any motivation for doing so ?

The output of a Connected Threshold filter is typically
a binary image, so you probably should do:

  typedef   unsigned char     BinaryPixelType;

  typedef itk::Image< BinaryPixelType, Dimension >  BinaryImageType;

  typedef itk::ConnectedThresholdImageFilter<
     InternalImageType,
     BinaryImageType > ConnectedFilterType;


and subsequently, you should instantiate the
BinaryImageToLabelMapFilter also using as
input this Binary image type.

Currently, you have:

  typedef itk::BinaryImageToLabelMapFilter<
     InternalImageType, LabelMapType > ConverterType;

.. and you should probably replace it with:

  typedef itk::BinaryImageToLabelMapFilter<
     BinaryImageType, LabelMapType > ConverterType;



Regards


             Luis


----------------------------
On Tue, Dec 8, 2009 at 2:38 PM, John Drozd <john.drozd at gmail.com> wrote:
> Hello,
>
> I have segmented the ventricles from an image using a connected threshold
> image filter.
> I then tried to convert my outputted segmentation to a label map filter and
> then to get the label's attributes.
>
> I am using parts of the example code attribute_values.cxx that is described
> in section 9.4 of
> Gaetan Lehmann's paper: "Label object representation and manipulation with
> ITK"
>
> My problem is that my code is telling me that the number of labels is 0.
> How do I apply a label to my segmented ventricles?
>
> Below is my code:
>
> /*
> to run type:
> ./ConnectedThresholdImageFilter correctedsubject5.dcm outsubject5.dcm 103
> 142 95 17100 17300
> */
>
> #if defined(_MSC_VER)
> #pragma warning ( disable : 4786 )
> #endif
>
> #ifdef __BORLANDC__
> #define ITK_LEAN_AND_MEAN
> #endif
>
>
> #include "itkConnectedThresholdImageFilter.h"
>
> #include "itkImage.h"
> #include "itkCastImageFilter.h"
>
> #include "itkCurvatureFlowImageFilter.h"
>
> #include "itkImageFileReader.h"
> #include "itkImageFileWriter.h"
>
> #include "itkGDCMImageIO.h"
>
> #include "itkVersion.h"
>
> #include "itkOrientedImage.h"
> #include "itkMinimumMaximumImageFilter.h"
>
> #include "itkGDCMImageIO.h"
> #include "itkGDCMSeriesFileNames.h"
> #include "itkNumericSeriesFileNames.h"
>
> #include "itkImageSeriesReader.h"
> #include "itkImageSeriesWriter.h"
>
> #include "itkResampleImageFilter.h"
> #include "itkShiftScaleImageFilter.h"
>
> #include "itkIdentityTransform.h"
> #include "itkLinearInterpolateImageFunction.h"
>
> #include <itksys/SystemTools.hxx>
>
> #include "gdcm/src/gdcmFile.h"
> #include "gdcm/src/gdcmUtil.h"
>
> #include <string>
>
> //added per attribute_values.cxx
> #include "itkImageFileReader.h"
>
> #include "itkShapeLabelObject.h"
> //#include "itkStatisticsLabelObject.h"
>
> #include "itkLabelMap.h"
>
> #include "itkBinaryImageToShapeLabelMapFilter.h"
> //#include "itkBinaryImageToStatisticsLabelMapFilter.h"
>
> //end of added code
>
> int main( int argc, char *argv[])
> {
>   if( argc < 7 )
>     {
>     std::cerr << "Missing Parameters " << std::endl;
>     std::cerr << "Usage: " << argv[0];
>     std::cerr << " inputImage  outputImage seedX seedY seedZ lowerThreshold
> upperThreshold" << std::endl;
>
>     return 1;
>     }
>
>   typedef   float           InternalPixelType;
>
>   const     unsigned int    Dimension = 3;
>
>   typedef itk::Image< InternalPixelType, Dimension >  InternalImageType;
>
>   typedef signed short OutputPixelType;
>
>   typedef itk::Image< OutputPixelType, Dimension > OutputImageType;
>   typedef itk::Image< float, Dimension > OutputImageType2;
>   typedef itk::CastImageFilter< InternalImageType, OutputImageType >
>     CastingFilterType;
>   CastingFilterType::Pointer caster = CastingFilterType::New();
>
>   const    unsigned int    ImageDimension = 3;
>   typedef  signed short    PixelType;
>
>   typedef itk::Image< PixelType, ImageDimension >  FixedImageType;
>   typedef itk::Image< float, ImageDimension >  FloatImageType;
>
>   typedef  itk::ImageFileReader< FixedImageType > ReaderType;
>   typedef  itk::ImageFileWriter<  OutputImageType  > WriterType;
>   typedef  itk::ImageFileWriter<  FloatImageType  > WriterType2;
>
>   ReaderType::Pointer reader = ReaderType::New();
>   WriterType::Pointer writer = WriterType::New();
>   WriterType2::Pointer writer2 = WriterType2::New();
>
>   typedef itk::GDCMImageIO           ImageIOTypefixed;
>   ImageIOTypefixed::Pointer gdcmImageIOfixed = ImageIOTypefixed::New();
>   reader->SetImageIO( gdcmImageIOfixed );
>
>   typedef itk::GDCMImageIO           ImageIOTypefixed2;
>   ImageIOTypefixed2::Pointer gdcmImageIOfixed2 = ImageIOTypefixed2::New();
>
>   reader->SetFileName( argv[1] );
>
>   reader->Update();
>
>   typedef itk::CurvatureFlowImageFilter< InternalImageType,
> InternalImageType >
>     CurvatureFlowImageFilterType;
>
>   CurvatureFlowImageFilterType::Pointer smoothing =
>                          CurvatureFlowImageFilterType::New();
>
>   typedef itk::ConnectedThresholdImageFilter< InternalImageType,
>                                     InternalImageType > ConnectedFilterType;
>
>   ConnectedFilterType::Pointer connectedThreshold =
> ConnectedFilterType::New();
>
>   typedef signed short InputAPixelType;
>   typedef float OutputBPixelType;
>
>   typedef itk::Image< InputAPixelType, 3 > InputAImageType;
>   typedef itk::Image< OutputBPixelType, 3 > OutputBImageType;
>
>   typedef itk::CastImageFilter< InputAImageType, OutputBImageType >
> CastFilterType;
>
>   CastFilterType::Pointer castFilter = CastFilterType::New();
>
>
>   castFilter->SetInput( reader->GetOutput() );
>
>
>   connectedThreshold->SetInput( castFilter->GetOutput() );
>
>   caster->SetInput( connectedThreshold->GetOutput() );
>
>
>   smoothing->SetNumberOfIterations( 20 ); //was 5
>   smoothing->SetTimeStep( 0.125 );
>
>   const InternalPixelType lowerThreshold = atof( argv[6] );
>   const InternalPixelType upperThreshold = atof( argv[7] );
>
>   connectedThreshold->SetLower(  lowerThreshold  );
>   connectedThreshold->SetUpper(  upperThreshold  );
>
>   connectedThreshold->SetReplaceValue( 255 );
>
>   InternalImageType::IndexType  index;
>
>   index[0] = atoi( argv[3] );
>   index[1] = atoi( argv[4] );
>
>   //added
>   index[2] = atoi( argv[5] );
>
>   std::cout << index << std::endl;
>
>   // Software Guide : BeginCodeSnippet
>   connectedThreshold->SetSeed( index );
>
>   //obtain a 5 x 5 bounding region of seeds
>   int ii, jj, kk;
>
>   ii = index[0];
>   jj = index[1];
>   kk = index[2];
>
>   for (int i = ii; i < ii + 5; i++)
>     for (int j = jj; j < jj + 5; j++)
>       for (int k = kk; k < kk + 5; k++)
>     {
>
>       index[0] = i;
>       index[1] = j;
>       index[2] = k;
>       connectedThreshold->AddSeed( index );
>     }
>
>   for (int i = ii; i > ii - 5; i--)
>     for (int j = jj; j > jj - 5; j--)
>       for (int k = kk; k > kk - 5; k--)
>     {
>
>       index[0] = i;
>       index[1] = j;
>       index[2] = k;
>       connectedThreshold->AddSeed( index );
>     }
>
>   connectedThreshold->Print(std::cout,17100);
>
>   typedef itk::MetaDataDictionary DictionaryType;
>
>   DictionaryType inputdict = reader->GetMetaDataDictionary();
>
>   writer->SetMetaDataDictionary( inputdict );
>
>   writer->SetFileName( argv[2] );
>
>   //added per attribute_values.cxx
>
>   // define the object type. Here the ShapeLabelObject type
>   // is chosen in order to read some attribute related to the shape
>   // of the objects (by opposition to the content of the object, with
>   // the StatisticsLabelObejct).
>   typedef unsigned long LabelType;
>
>   typedef itk::ShapeLabelObject< LabelType, 3 > LabelObjectType;
>
>   typedef itk::LabelMap< LabelObjectType > LabelMapType;
>
>   // convert the image in a collection of objects
>   typedef itk::BinaryImageToLabelMapFilter< InternalImageType, LabelMapType
>> ConverterType;
>
>   ConverterType::Pointer converter = ConverterType::New();
>
>   converter->SetInput( connectedThreshold->GetOutput() );
>
>   //converter->SetForegroundValue( atoi(argv[2]) );
>   //converter->SetForegroundValue( 255 );
>
>   // valuate the attributes with the dedicated filter: ShapeLabelMapFilter
>
>   typedef itk::ShapeLabelMapFilter< LabelMapType > ShapeFilterType;
>   ShapeFilterType::Pointer shape = ShapeFilterType::New();
>   shape->SetInput( converter->GetOutput() );
>
>   shape->Update();
>
>
>   // then we can read the attribute values we're interested in. The
> BinaryImageToShapeLabelMapFilter
>   // produce consecutive labels, so we can use a for loop and
> GetLabelObject() method to retrieve
>   // the label objects. If the labels are not consecutive, the
> GetNthLabelObject() method must be
>   // use instead of GetLabelObject(), or an iterator on the label object
> container of the label map.
>   LabelMapType::Pointer labelMap = shape->GetOutput();
>
>   std::cout << "Number of label objects = " <<
> labelMap->GetNumberOfLabelObjects() << std::endl;
>
>
>   for( unsigned int label=1; label<=labelMap->GetNumberOfLabelObjects();
> label++ )
>     {
>     // we don't need a SmartPointer of the label object here, because the
> reference is kept in
>     // in the label map.
>     const LabelObjectType * labelObject = labelMap->GetLabelObject( label );
>     std::cout << label << "\t" << labelObject->GetPhysicalSize() << "\t" <<
> labelObject->GetCentroid() << std::endl;
>     }
>
>   writer->SetInput( caster->GetOutput() );
>
>   try
>     {
>     writer->Update();
>     }
>   catch( itk::ExceptionObject & excep )
>     {
>     std::cerr << "Exception caught !" << std::endl;
>     std::cerr << excep << std::endl;
>     }
>
>   std::cout << "output from reader->GetOutput()->GetDirection()" <<
> std::endl;
>   std::cout << reader->GetOutput()->GetDirection() << std::endl;
>
>   std::cout << "output from castFilter->GetOutput()->GetDirection()" <<
> std::endl;
>   std::cout << castFilter->GetOutput()->GetDirection() << std::endl;
>
>   std::cout << "output from connectedThreshold->GetOutput()->GetDirection()"
> << std::endl;
>   std::cout << connectedThreshold->GetOutput()->GetDirection() << std::endl;
>
>   std::cout << "output from caster->GetOutput()->GetDirection()" <<
> std::endl;
>   std::cout << caster->GetOutput()->GetDirection() << std::endl;
>
>   return 0;
> }
>
>
> _____________________________________
> 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
>
>


More information about the Insight-users mailing list