Advanced Contouring With Pinnacle3D

I find that when I talk to colleagues about contouring, they seem to be unaware of the more advanced and useful tools existing inside Pinnacle#D to assist with our tasks.

There are two disclaimers that I would start with. Firstly I don't know all that there is to know about Pinnacle3D and if you know something that is also useful but I haven't mentioned, then drop a line. Secondly, my contouring over the years a=has become much more pixel-value based and less manual where possible. This leads to some circumstances where the process looks excessively convoluted, but through the use of automated steps is actually slightly quicker and exceedingly more accurate.

The basis of this approach is that I take the definitions in ICRU documents very seriously as they have embedded in them a large amount of onco-logic and knowledge.

So GTV is marked up as if it is GROSS, that is, what is seen, without an overlay of suspicion about what a tumour is doing. You might have seen that I call this 'anti-anatomy', the bit with no name.

And CTV is marked as an anatomical volume against soft tissue planes where the estimated risk of tumour transgression falls to 0%. The exception is where a tumour sits in a large bed of different tissue (node in fat, tumour in the lung, tumour in muscle) and the nearest anatomical boundary is a ridiculous distance from a tumour. (You will recognise that this is the exact opposite of how most ROs think about the CTV, generally adopting the expansion fro GTV and then excluding a few tissues like bone and also air).

I prefer to think that my method actually reflects the definition of CTV, that is, the anatomical area of subclinical disease risk. But let's not argue about that!

So the functions that I shall point out and discuss are listed below and I'll eventually include pictures to make it a photo essay.

  • MUCOSA definition

Once again I am unusual in that I define an OAR called MUCOSA in all H&N patients. Pinnacle3D has some automated contours such as ORAL CAVITY, and some of my colleagues add MIDLINE STRUCTURES. My critiques of these are that the ORAL CAVITY is actually a tongue volume including mucosa and striated muscles which have very different acute and late effects, and that there is no anatomical structure called MIDLINE STRUCTURES. Rather MIDLINE STRUCTURES is being used to cover the mucosa of the aerodigestive tract, and the larynx. SO why not contour the MUCOSA and LARYNX?

The process use pixel-based contours, and auto-contouring, both of which are standard Pinnacle3D features, and uses 3 contours, none of which need be drawn.

The process starts with a plain CT.


The process follows the following steps:

  1. create the AIRTISSUE ROI. This interface exists between the air in the aerodigestive tract and the mucosa.
    1. set the CT autocontour threshold number range to 0 and 800.
    2. set the view to poly do that you can see what you have completed on each panel.
    3. position the crosshairs in the air of the tract around the top of the epiglottis.
    4. click on the Autocontour Options… and then Apply Bounding Box Yes button. Press the Create Bounding Box icon, and then in each 3D window click and drag the cursor to fashion a dashed cube of the required size around the aerodigestive tract.
    5. draw the box to sit on the hard palate superiorly, the teeth anteriorly, the buccal mucosa laterally, the vertebral column posteriorly, and the superior epiglottis inferiorly
    6. click on the Autocontour Structure on All Slices button (this effectively will define any air cavity which sits contiguously above or below where you click within the whole volume. You will see the AIRTISSUE contour being created within the bounding box.1
    7. If the volume produced is not complete, move up and down through the bounded volume and click to the left of any air pocket.
    8. When complete, press the Adjust Bounding Box button and pull the inferior, superior, posterior and anterior planes into positions that cover the middle portion of the tract. I usually stop at the thoracic inlet. Use the Autocontour Structure on All Slices button again on the upper slice. Move down to discover places where air is not contoured. It is usually a little complex around the epiglottis and hypopharynx.
    9. When complete, readjust the bounding box to go behind the sternum. The danger here is that it is reasonably common for some lung to be included on some slices which will require manual correction.
  2. create the TEETH ROI, set the CT number range to 0-2200, and set to poly.
    1. set the window view to BONE.
    2. scroll up to the maxilla, on each slice moving downwards, use the Autocontour Structure on Current Slice button by clicking to the left of each tooth seen outside the maxila and mandible.
  3. create the MANDIBLE ROI either by using MBS (if you have it), or by setting the CT number range to 0-1150, and then move from the chin up to the TMJ using the Autocontour Structure on Current Slice and then clicking to the left of each segment of the mandible. Once you go past the teeth level, you can do every 2nd or 3rd slice and interpolate. The actual TMJ doesn't need contouring unless it's in field.
    1. review any place where TEETH and MANDIBLE overlap and decide it there is actually any mandible in the slice. If not proceed on.
  4. using the ROI combination/subtraction tool, use AIRTISSUE and TEETH as the source, and MANDIBLE as the avoid, set the distance to 0.2cm and then create ring ROI to produce MUCOSA.
    This will create a 2mm annulus around the TEETH and AIRTISSUE but will not go into MANDIBLE.
    1. scroll up to the top of the maxillary teeth and you will see a solid mucosa contour, delete it

That's it, the MUCOSA is no present. Verify for yourself that the mucosa which sits against the teeth and lines the aerodigestive tract where it contact air, is actually contoured.

[as an equation MUCOSA = annulus2mm (AIRTISSUE include TEETH exclude MANDIBLE)]

The one place where there is mucosa but it is not contoured by this method is around the area of the posterior tongue/anterior soft palate interface where there is rarely any air. To correct this, after the annulus creation use a 3mm brush to add MUCOSA in these places.

  • PAROTID definition (I will demonstrate this on one slice, but you do the autocontour step first, the following steps introduce more automation:
    1. start with the CT again
    2. create the PAROTID_R ROI, set the CT thresholds in the range of 950-1050, and set the 2D display to poly.2
    3. use the single slice pixel-based spike button ["Autocontour Structure on Current Slice"] and click to the left of the parotid in the fat or in the muscle, and click in many places until you have defined the parotid.3
    4. use the smooth contours option from the dropdown list, and click Erode/Dilate ROI with a margin of 0.5 pixel.

      You will see that and excellent PAROTID_R contour now appears.4

    5. use the Clean ROI option from the Options dropdown list and click the Rescan button, then the Delete Curves button to remove extraneous contours to leave the parotid.56

The sequence displayed here produced a PAROTID_R contour with a volume of 0.938434 cm^3. I repeated the process and produced another contour with a volume of 0.938434 cm^3. The sem-iautomated process produces much more reproducible result than manual contouring.7

It takes a little time to play with the pixel-based selection to get all of the non-parotid tissues included. This makes the contour of the parotid very conformal. Don't forget to look if your parotid actually expends down towards the aerodigestive tract.

  • Name sorting
  • Definition of CTV in LUNG
  • Definition of CTVn0 in H&N
  • Viewing the CT scan at weird contrast levels (and why I do it!)
Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-NonCommercial-NoDerivs 3.0 License