How to convert a 3D vtkDataSet file to a numpy array?

Question

I am trying to read a series of DICOM files and turn them into a 3D model. I did it. However, I also want to remove some fragments before rendering. Here I am lost. I can’t find a way to access each downloaded file and delete the ones I want. I read that I can convert a dataset to a numpy array. However, I am not sure how I can do this.

    reader = vtkDICOMImageReader()
    reader.SetDirectoryName('loads a directory of DICOM files')
    reader.Update()

    imageDataGeometryFilter = vtkImageDataGeometryFilter()
    imageDataGeometryFilter.SetInputConnection(reader.GetOutputPort())
    imageDataGeometryFilter.Update()

    mapper = vtkPolyDataMapper()
    mapper.SetInputConnection(imageDataGeometryFilter.GetOutputPort())

    '''Here is where I want to take my vtkDataSet and convert it into a numpy array.
       Once in array form I can remove data that I do not want (ie DICOM slices).
       Then I want to convert it back into a vtkDataSet to continue program.
    '''



    actor = vtkActor()
    actor.SetMapper(mapper)

    ren = vtkRenderer()
    ren.AddActor(actor)
    ren.SetBackground(0.1,0.3,0.1)

    renWin = vtkRenderWindow()
    renWin.AddRenderer(ren)

    interactor = vtkRenderWindowInteractor()
    interactor.SetRenderWindow(renWin)

    renWin.Start()

    interactor.Initialize()
    interactor.Start()

Answer 1

    pixelspace = reader.GetPixelSpacing()
    spacing = image.GetSpacing()

    vtk_data = image.GetPointData().GetScalars()

    numpy_data = numpy_support.vtk_to_numpy(vtk_data)

    numpy_data = numpy_data.reshape(dims[0], dims[1], dims[2])
    numpy_data = numpy_data.transpose(2,1,0)

    dims = numpy_data.shape

    dataImporter = vtkImageImport()
    dataImporter.SetDataScalarTypeToFloat() 
    dataImporter.SetNumberOfScalarComponents(1)
    dataImporter.SetDataExtent(0, dims[2]-1, 0, dims[1]-1, 0, dims[0]-1)
    dataImporter.SetWholeExtent(0, dims[2]-1, 0, dims[1]-1, 0, dims[0]-1)
    dataImporter.SetDataSpacing(spacing[0],spacing[1],spacing[2])
    dataImporter.CopyImportVoidPointer(numpy_data, numpy_data.nbytes)