ITK-Image processing, calculate shape features for 3D image

Question

I'm trying to calculate shape features for a 3D image using ITK library with C++. So I followed the example given in the ITK documentation. The example takes a 2D image, and extracts different shape features. In my program, I just want for a given 3D image to get shapes attributes and stock them in an array <double>.

Here is what I have so far :

//principal declarations
const unsigned int Dimension = 3;
typedef unsigned char                                 PixelType;
typedef unsigned short                                LabelType;
typedef itk::Image<PixelType, Dimension>              InputImageType;
typedef itk::Image< LabelType, Dimension >            OutputImageType;
typedef itk::ShapeLabelObject< LabelType, Dimension > ShapeLabelObjectType;
typedef itk::LabelMap< ShapeLabelObjectType >         LabelMapType;
typedef itk::ImageFileReader<InputImageType> ReaderType;
typedef itk::ConnectedComponentImageFilter <InputImageType, OutputImageType   > ConnectedComponentImageFilterType;
typedef itk::LabelImageToShapeLabelMapFilter< OutputImageType, LabelMapType> I2LType;

typedef itk::Array< double > MeasurementVectorType;

MeasurementVectorType formes(9);
InputImageType::Pointer image;

//read the 3Dimage
ReaderType::Pointer reader = ReaderType::New();
reader->SetFileName(file);
reader->Update();
image = reader->GetOutput();

ConnectedComponentImageFilterType::Pointer connected =      ConnectedComponentImageFilterType::New ();
connected->SetInput(image);
connected->Update();

// create the shape label map filter
I2LType::Pointer i2l = I2LType::New();
i2l->SetInput( connected->GetOutput() );
i2l->SetComputePerimeter(true);
i2l->Update(); 

LabelMapType *labelMap = i2l->GetOutput();

//calculate shape attributes for the first label
ShapeLabelObjectType *labelObject = labelMap->GetNthLabelObject(0);
            //stock the attributes in the array
            formes[0]=labelObject->GetBoundingBox();
            formes[1]=labelObject->GetNumberOfPixels();
            formes[2]=labelObject->GetPhysicalSize();
            formes[3]=labelObject->GetElongation();
            formes[4]=labelObject->GetPerimeter();
            formes[5]=labelObject->GetRoundness();
            formes[6]=labelObject->GetEquivalentSphericalRadius();
            formes[7]=labelObject->GetEquivalentSphericalPerimeter();
            formes[8]=labelObject->GetFlatness();

I was able to read the 3D images and calculate its shape attributes. However I have this problem : i can't stock them in the array <double> because labelObject methodes returns a const type. I get this error : IntelliSense: no suitable conversion function from const itk::ImageRegion<3U> to "double" exists

Does any one used ITK to do this ? If there is any other method to achieve that, could anyone point me to a solution please ?

Any help will be much apreciated

Answer 1

The problem isn't the const type--you are allowed to assign from consts, you are just not allowed to assign to them.

The problem is that GetBoundingBox() returns an itk::ImageRegion object, which is a more complicated object that contains Index and Size objects as members. There is no simple way to cast this to a double that ITK understands, so trying to cast it to a double doesn't make sense.

You should go through the ShapeLabelObject class documentation and see which methods return types that can be cast to double.

For example, the GetNumberOfPixels() method returns a SizeValueType, which is just a typedef of unsigned long, so this could be cast to a double. The GetElongation() and GetRoundness() methods both have return type const double so those will be fine too.