Wk2 - Advanced Contouring

Question 1

the different contour comparison metrics

Answer 1

• volume
• centre of volume
• volume overlap
• 2D shape and dimension
• 3D shape and dimension

Question 2

contour comparison metrics - volume + limitations

Answer 2

• the absolute volume (cc)
• can be used to calculate the % volume different from the reference, mean volume and SD

limitation
- no spatial correlation of volumes

Question 3

contour comparison metrics - centre of volume (COV) + limitation

Answer 3

• very easy to work out in TPS
  
  • autoplace POI in ROI

limitations
- differing volumes can have the same COV

Question 4

contour comparison metrics - volume overlap + limitation

Answer 4

• dice similarity coefficient
• concordance index (CI) = volume of intersection/volume of consensus x 100%

limitation
- do not measure distances between volume edges

Question 5

contour comparison metrics - 2D shape and dimension + limitation

Answer 5

• measure maximum distance in a particular dimension
• can measure slice by slice
• can use BEV display with contours displayed

limitations
- can have same maximum in a particular dimension with different volumes and COVs

Question 6

contour comparison metrics - 3D shape and dimension + limitation

Answer 6

• hausdorff distance is commonly used
• mean distance to agreement is also commonly used
• calculates differences between surfaces for multiple points around the volume

limitation
- maximum distance, average distance and 95% CI measurements only
- no directional information
- irregular surfaces can result in errors
- particularly for radial methods (eg. polar coordinates)

Question 7

StructSure

Answer 7

• contour analysis software
• integrated into ProKnow

Question 8

ProKnow

Answer 8

• contour and plan review software
• calculates the sensitive and sophisticated StructSure accuracy score as well as simple metrics (dice coefficient, total volume, matching/missing/extra volumes, etc)
• displays and analyses variability

Question 9

MIM Maestro

Answer 9

• contouring, image registration, plan adaptation software
• calculates a wide range of metrics

Question 10

Matlab

Answer 10

• calculates dice, Hausdorff distance, STAPLE

Question 11

3D Slicer-Slicer RT

Answer 11

• calculates dice and Hausdorff distance

Question 12

factors affecting manual contouring outcomes

Answer 12

• windowing
• image interpretation skills
• limitations due to image quality

Question 13

deep neural networks (DNNs)

Answer 13

• a type of artificial neural network (ANN)

Question 14

Describe the rationale for auto-contouring

Answer 14

With the advancement of RT techniques came the need for detailed contouring of datasets required to drive optimisation. Manual contouring is time consuming and is prone to intra and inter-observer error.

Question 15

How does the threshold auto-contouring tool work?

Answer 15

Quick and effective contouring tool where upper and lower CT numbers are selected and only those will be outlined

Question 16

What is the challenge with fully automatic segmentation methods?

Answer 16

To determine how much user interaction is necessary
When is it suitable to implement this into clinical practice
Contours need to be given to build the model - what is the gold standard?

Question 17

Describe the body atlas based method

Answer 17

Reference image associated with atlas contours is matched to CT via deformable registration. Atlas contours are then morphed to match patients image.

Question 18

Define AI

Answer 18

Models algorithms or computer programs to tackle certain tasks requiring human intelligence

Question 19

Define machine learning

Answer 19

Numerical algorithms and models established to analyse data and derive or learn decision-making capabilities to achieve certain tasks

Question 20

Define deep learning

Answer 20

Large scale hierarchal models with multi-layer architectures to automatically generate comprehensive representations and learn complicated inherent patterns of data

Question 21

Explain 3 learning styles of DL

Answer 21

1. Supervised - learns relationship between input data and labels (requires labelled data) good for diagnosis, image segmentation
2. Unsupervised - learns patterns in input data (unlabelled) good for anomaly detection in QA
3. Reinforcement - learns to perform actions in response to environment to maximise a reward function good for decision making in ART

Question 22

List 4 different contouring tools

Answer 22

1. Manual
2. Grey scale interrogation
3. Body atlas based
4. Statistical shape modelling

Question 23

2 types of greyscale interrogation - CT

Answer 23

1 - threshold
2 - model based segmentation

Question 24

body atlas based contouring process

Answer 24

1. reference image associated with atlas contouring is “matched” to the patients’s image (CT or CBCT) via a deformable registration algorithm
2. the resulting deformation field is used to morph atlas contours to match the patient image

Question 25

manual contouring is

Answer 25

time consuming and prone to intra and inter observer error

Question 26

analysis of contouring methods and tools

Answer 26

- manual contours of an expert is used as the gold standard - clinical expertise - knowledge of spatial relationships, normal anatomy and patterns of disease progression - multiple experts preferred (due to intra and inter observer error) - consensus contours pooling the expertise of multiple clinicians - within a discipline and across disciplines (radiologist, ROs and RTs) - assessment of intra and inter observer variability required

Question 27

threshold techniques - contouring CT

Answer 27

- most commonly applied to segment anatomy on individual 2D slices of the 3D data set in radiotherapy treatment planning systems - upper and lower limits for the CT numbers are selected, essentially applying thresholds for CT data to be included in the ROI - a start point is identified on the image proximal to the edge of the ROI to be outlined. the edge ROI is detected/tracked and the ROI is outlined - successful auto-outlining is dependent on image resolution, significant contrast between corresponding structures and a continuous surface

Question 28

threshold technique - PET

Answer 28

- can use count or SUV voxel data - very contentious issue

Question 29

emerging solutions

Answer 29

- artificial intelligence - machine learning - deep learning models

Question 30

common types of DNNs

Answer 30

- convolution neural networks (CNNs)/ fully convolutional network (FCN) - widely used to extract image features for classification - U-nets commonly used for segmenting/contouring images