BUG: Fix 4DCT vrg pipeline not properly centering sequences

[NicerNewerCar]

Regression introduced in e036c22

Closes #105

Also resolves this comment #102 (comment)

[jcfr]

We are getting close 🙏

Few nitpicks:

• move fix into its own commit
• add the performance change into its own commit
• move the removal of the threadexecutor into its own pull request

[jcfr]

in onGenerateVRG, after line 512, I suggest to simplify doing something like this:

            # center the volume
            transformOutputDir = os.path.join(mainOutputDir, tfmPath)
            self.logic.createPathsIfNotExists(transformOutputDir)
            
            
            if not self.logic.IsSequenceVolume(volumeNode):
                IO.exportOrigin2DicomTransform(volumeNode, transformOutputDir)
            else:
                # Just export the first frame
                proxyNode = self.logic.getItemInSequence(volumeNode, 0)[0]
                IO.exportOrigin2DicomTransform(proxyNode, transformOutputDir)

                if tfmNode is not None:

                    # if still needed, explain here why calling "AddCenteringTransform" and reverting its effect is still needed

And here is the function exportOrigin2DicomTransform:

def exportOrigin2DicomTransform(volumeNode, exportDir):
    """If volume is not already centered, save corresponding transform as `Origin2Dicom.tfm`"""

    if volumeNode.IsCentered():
        return False

    volumeNode.AddCenteringTransform():

    # Get reference to the centering transform (Volume to Center)
    tfmNode = volumeNode.GetParentTransformNode()

    # Revert effect of "AddCenteringTransform"
    volumeNode.HardenTransform()
    volumeNode.SetAndObserveTransformNodeID(None)

    # Update from "Volume to Center" to "Center to Volume"
    tfmNode.Inverse()

    # Save as "Origin2Dicom.tfm"
    tfmPath = os.path.join(exportDir, "Origin2Dicom.tfm")
    slicer.util.saveNode(tfmNode, tfmPath)

    slicer.mrmlScene.RemoveNode(tfmNode)

    return True

Ideally, we should also look into adding support to call GetCenterPositionRAS from Python, this would simplify the logic.

[NicerNewerCar]

Ideally, we should also look into adding support to call GetCenterPositionRAS from Python, this would simplify the logic.

This doesn't seem to be exposed in the Python API:

>>> x.GetCenterPositionRAS()
Traceback (most recent call last):
  File "<console>", line 1, in <module>
AttributeError: 'MRMLCorePython.vtkMRMLScalarVolumeNode' object has no attribute 'GetCenterPositionRAS'

It also seems to be a protected function for the vtkMRMLVolumeNode. Was this exposed somewhere else?

[amymmorton]

Suggest with the depreciation of vrgs- also close this