Receive image spacing from image messages

[olevs]

Current solution throws away spacing of incoming image messages and sets spacing to (1, 1, 1). The incoming values should be kept.

[adamrankin]

@lassoan is this somehow related to Slicer storing spacing info in MRML (or something like that? I remember it wasn't stored in the vtkImageData)

[adamrankin]

Right, I thought so.

Here:
https://github.com/openigtlink/SlicerOpenIGTLink/blob/master/OpenIGTLinkIF/MRML/vtkMRMLIGTLConnectorNode.cxx#L367

If the spacing is set both on the image data and in the ijk to ras, there will be a compounding effect. This PR will have to have an accompanying PR in there to properly manage the spacing on the vtkImageData.

@lassoan is it as simple as:

          volumeNode->SetIJKToRASMatrix(imageDevice->GetContent().transform);
          volumeNode->SetAndObserveImageData(imageDevice->GetContent().image);
          volumeNode->Modified();

becomes

          volumeNode->SetIJKToRASMatrix(imageDevice->GetContent().transform);
          volumeNode->SetAndObserveImageData(imageDevice->GetContent().image);
          volumeNode->GetImageData()->SetSpacing(1.0, 1.0, 1.0); // IGTL device sets spacing in image data, do not duplicate with IJKtoRAS
          volumeNode->Modified();

?

[adamrankin]

@olevs could you also add in setting of the origin as well? I can't send edits to your PR

  float spacing[3];       // spacing (mm/pixel)
  float origin[3];        // origin (mm)
  int   svsize[3];        // sub-volume size
  int   svoffset[3];      // sub-volume offset
  int   scalarType;       // VTK scalar type
  int   numComponents;    // number of scalar components
  int   endian;

  scalarType = IGTLToVTKScalarType( imgMsg->GetScalarType() );
  endian = imgMsg->GetEndian();
  imgMsg->GetDimensions(size);
  imgMsg->GetSpacing(spacing);
  imgMsg->GetOrigin(origin);
  numComponents = imgMsg->GetNumComponents();
  imgMsg->GetSubVolume(svsize, svoffset);

  // check if the IGTL data fits to the current MRML node  
  int sizeInNode[3]={0,0,0};
  int scalarTypeInNode=VTK_VOID;
  int numComponentsInNode=0;

  if (imageData.GetPointer()!=NULL)
    {
    imageData->GetDimensions(sizeInNode);
    scalarTypeInNode = imageData->GetScalarType();
    numComponentsInNode = imageData->GetNumberOfScalarComponents();
    }
    
  if (imageData.GetPointer()==NULL
      || sizeInNode[0] != size[0] || sizeInNode[1] != size[1] || sizeInNode[2] != size[2]
      || scalarType != scalarTypeInNode
      || numComponentsInNode != numComponents)
    {
    imageData = vtkSmartPointer<vtkImageData>::New();
    dest->image = imageData;
    imageData->SetDimensions(size[0], size[1], size[2]);
    imageData->SetExtent(0, size[0]-1, 0, size[1]-1, 0, size[2]-1);
    imageData->SetOrigin(origin[0], origin[1], origin[2]);
    imageData->SetSpacing(spacing[0], spacing[1], spacing[2]);

[lassoan]

@olevs could you also add in setting of the origin as well? I can't send edits to your PR

And if you detect VTK9 is used then you could set image directions, too.

[olevs]

@lassoan I don't have the time or setup to test/verify image directions and VTK 9. If we should do it now, that probably needs to be done in a separate pull request by someone else?