CT Brain vessel segmentation using ITK and VTK

Question

I am doing my own project in which i need to segment brain vessels of CT dicom images. I have 2 sets of DICOM images .

1)CT dicom images of head with DSA (888 slices)
2)CT dicom images of head without DSA(same 888 slices),

So i read these two series seperately using vtkDICOMImageReaderand subtracted these two sets using vtkImageMathematics and stored the result in a variable matand then copied it to vtkImageReslice * reslice; Then I imported this reslice to ITK using VKImageToImageFilter *filter_toitkimage and used this code:

using HessianFilterType = itk::HessianRecursiveGaussianImageFilter<ImageType>;
HessianFilterType::Pointer hessianFilter = HessianFilterType::New();
hessianFilter->SetInput(filter_toitkimage->GetOutput());
hessianFilter->SetSigma(.6);

using VesselnessMeasureFilterType = itk::Hessian3DToVesselnessMeasureImageFilter<PixelType>;
VesselnessMeasureFilterType::Pointer vesselnessFilter = VesselnessMeasureFilterType::New();
vesselnessFilter->SetInput(hessianFilter->GetOutput());
vesselnessFilter->SetAlpha1(1);
vesselnessFilter->SetAlpha2(2); 

Then i converted it back to VTK using ImageToVTKImageFilter and used this :

       vtkSmartPointer<vtkGPUVolumeRayCastMapper> volumeMapper = vtkSmartPointer<vtkGPUVolumeRayCastMapper>::New();
volumeMapper->CroppingOff();
volumeMapper->RemoveAllInputs();
volumeMapper->SetInputData(conv->GetOutput());   //conv is the converted output   
volumeMapper->SetBlendModeToMaximumIntensity();
volumeMapper->AutoAdjustSampleDistancesOff();
volumeMapper->SetSampleDistance(.1);

volumeMapper->Update();

//-----volume color
vtkSmartPointer<vtkColorTransferFunction> volumeColor = vtkSmartPointer<vtkColorTransferFunction>::New();
volumeColor->RemoveAllPoints();

////-----volume color

//---air
volumeColor->AddRGBPoint(0, 0.0, 0.0, 0.0);
volumeColor->AddRGBPoint(220, 1, 1, 1);
//---skin
volumeColor->AddRGBPoint(250, 0, 0, 0);
volumeColor->AddRGBPoint(300, 0, 0, 0);
//---muscle
volumeColor->AddRGBPoint(800, 1, 1, 1);
volumeColor->AddRGBPoint(1000, 1, 1, 1);
volumeColor->AddRGBPoint(1320, 1, 1, 1);
//---soft bone
volumeColor->AddRGBPoint(1350, 0, 0, 0);
volumeColor->AddRGBPoint(1500, 0, 0, 0);
//---hard bone
volumeColor->AddRGBPoint(1550, 0, 0, 0);
volumeColor->AddRGBPoint(2000, 0, 0, 0);
//---hardest bone
volumeColor->AddRGBPoint(2050, 0, 0, 0);
volumeColor->AddRGBPoint(4096, 0, 0, 0);

vtkSmartPointer<vtkPiecewiseFunction> volumeScalarOpacity =
    vtkSmartPointer<vtkPiecewiseFunction>::New();
volumeScalarOpacity->AddPoint(0, 0.00);
volumeScalarOpacity->AddPoint(500, 0.15);
volumeScalarOpacity->AddPoint(1000, 0.15);
volumeScalarOpacity->AddPoint(1150, 0.85);

vtkSmartPointer<vtkPiecewiseFunction> volumeGradientOpacity =
    vtkSmartPointer<vtkPiecewiseFunction>::New();
volumeGradientOpacity->AddPoint(0, 0.0);
volumeGradientOpacity->AddPoint(90, 0.5);
volumeGradientOpacity->AddPoint(100, 1.0);

vtkSmartPointer<vtkVolumeProperty> volumeProperty = vtkSmartPointer<vtkVolumeProperty>::New();
volumeProperty->SetColor(volumeColor);
    volumeProperty->SetScalarOpacity(volumeScalarOpacity);
    volumeProperty->SetGradientOpacity(volumeGradientOpacity);
    volumeProperty->SetInterpolationTypeToLinear();
    volumeProperty->ShadeOn();
    volumeProperty->SetAmbient(0.5);
    volumeProperty->SetDiffuse(0.3);
    volumeProperty->SetSpecular(0.3);

    vtkSmartPointer<vtkVolume> volume =vtkSmartPointer<vtkVolume>::New();
    volume->SetMapper(volumeMapper);
    volume->SetProperty(volumeProperty);
    renderer->AddVolume(volume);

I am getting the vessels in the lungs But i didn't get the clear image of brain vessels. Can anybody give suggestions?

Answer 1

Your transfer function assumes CT intensities, but the output of vesselness filter is nothing like it. It might be normalized to 0.0-1.0 range, or 0-255 range, or something else. You should examine that.