MRFImageFilter vector image input


I have a few questions about the inputs to the MRFImageFilter. I understand that the filter takes both the original image and a labeled image. I am, however, confused about how to input both of these images into the filter. Since the Markov Random Field algorithm is defined for vector type images, are we to use composeImageFilter to combine the original image and the labeled image and then input this resultant vector image into the MRF filter? I provided what I imagine this combination step to look like:

filter->SetInput( 0, image0 );
filter->SetInput( 1, image1 );
itk::VectorImage<PixelType, dimension>::Pointer = filter->GetOutput();

I followed the example within Examples/Statistics/ScalarImageMarkovRandomField1.cxx, but could not find where this image combination step occurs. It seems to me that the only input into the ScalarToArrayFilterType is the unlabeled image. Any guidance is much appreciated.

Thank you!