Advanced Contouring

Question 1

What are the classes of contouring tools?

Answer 1

Manual
Image greyscale interrogation
Body atlas based methods
Statistical shape modelling

Question 2

What factors affect manual contouring?

Answer 2

Windowing
Image interpretation skills
Limitations due to image quality

Question 3

What is grey-scale interrogation?

Answer 3

Uses model based segmentation (Pinnacle) and threshold techniques.

• Upper and lower limits for the CT numbers are selected, essentially applying thresholds for CT data to be included in the region of interest (ROI)
• A start point is identified on the image proximal to the edge of the ROI to be outlined. The edge of the ROI is detected / tracked and the ROI is outlined.

Successful auto-lining depends on image resolution, a continuous surface and significant contrast between corresponding structures.

Question 4

Describe body atlas based methods of contouring?

Answer 4

Two step process: Reference image associated with atlas contours is “matched” to the patient’s image (CT or CBCT) via a deformable registration algorithm.

The resulting deformation field is used to morph atlas contours to match the patient image.
Examples: Eclipse Smart segmentation, Elekta ABAS, Velocity AI, Brainlab iPlan

Question 5

What are statisitcal shape based methods of contouring?

Answer 5

Do not rely solely on deformable registration maps.Creates a probabilistic model of organs.

Combines several deformable registration algorithms with model based segmentation and probabilistic refinement e.g. Pinnacle SPICE - Smart Probabilistic Image Contouring Engine

Question 6

Currently what is seen as the gold standard of contouring?

Answer 6

Manual contours of an expert. This is because they have clinical expertise and reasoning. Also have knowledge of spatial relationships, normal anatomy and patterns of disease progression.

Multiple experts are preferred (within a discipline or across disciplines - radiologists, oncologists, RT-). Important for assessing intra and inter-observer variability.

Question 7

What are limitations of manual contouring and What are motivations for auto-segmentation?

Answer 7

The increased advances in treatment planning and delivery involves complex tumour volumes and numerous OAR tolerances. This requires detailed contouring of datasets to drive optimisation.

Manual contouring is time consuming and prone to intra and inter observer error.
So auto-segmentation may be the answer.