Treatment Planning Algorithms

Question 1

What are the 2 main types of algorithms?

Answer 1

Model based and correction based algorithms

Question 2

Whats the main difference between photons and electrons (in relation to filters)?

Answer 2

With photons we have a flattening filter (bow-tie shaped) that removes low energy x-rays.

With electrons we have a scattering foil (circle shaped) that scatters electrons more

Question 3

What are the 2 parts of model based algorithms?

Answer 3

(1) a part of the algorithm that MODELS THE BEAM and represents the fluence distribution BEFORE THE BEAM ENTERS THE PATIENT. (e.g. models electrons striking Linac, models propagation through Linac head)
(2) A part that MODELS THE PATIENT (e.g. models interactions within the patient).

Question 4

What is the most common algorithm?

Answer 4

Convolution/Superposition algorithms
(All 3 algo’s on Pinnacle are convolution algorithms)
Have very good accuracy.

Question 5

What are the main photon algorithms?

Answer 5

Pencil Beam
Convolution
Superposition
Monte Carlo

Question 6

What are the components of Convolution Superposition algorithms?

Answer 6

Component 1: TERMA - total energy released per unit mass (represents the energy of primary photon interactions)

Component 2: Kernel - represents the energy deposited by a primary
photon interaction site.

REMEMBER: TERMA + KERNEL = DOSE

Question 7

What are the 2 parts of a kernel?

Answer 7

Primary kernel (calculates primary dose) and the scatter kernel

Question 8

What is the simple definition of a kernel?

Answer 8

The summation of all the interactions that gives a dose.

Question 9

What are the limitations of pencil beam algorithms?

Answer 9

It uses measurement based approaches (i.e Batho and ETAR) to account for imhomogeneties (not accurate).

Doesn’t account for the increased electron scatter contribution in air.

Question 10

The shape of the convolution kernel is wider than than the pencil beam kernel - what does this mean?

Answer 10

It is broader because the convolution algorithm accounts for the scatter contribution when it interacts with air. (This is why it more accurate)

Question 11

What is a superposition algorithm?

Answer 11

It is a type of convolution algorithm. It has been modified to take into account radiobiological path length.

It is the same shape as the convolution kernel and is also called convolution-superposition.

Question 12

What are advantages of convolution-superposition algorithms?

Answer 12

Models the interactions between the beam and the LINAC head.

Accounts for beam shaping and modifiers

It has efficient dose calculation

Question 13

What are disadvantages of convolution-superposition algorithms?

Answer 13

Relatively time consuming (especially for CNS/ large volumes )

Need to conduct beam commissioning measurements.

Question 14

What is the AAA algoritm?

Answer 14

anisotropic analytical

algorithm. It has a short computation time but it has accuracy close to Monte Carlo

Question 15

What are the types of electron algorithms?

Answer 15

Clarkson
Pencil Beam
Electron Monte Carlo

Question 16

Why are photons and electron beams different?

Answer 16

Electron beams have more pronounced side scatter (lateral equilibrium) - this influences dose distribution

So we need an algorithms that measures scatter

Question 17

What is the Clarkson algorithm?

Answer 17

A measurement based algorithm. It is not accurate because it is based on a phantom than was measured in water - so it does not accurately account for side scatter.

Question 18

What is a pencil beam (electron) algorithm?

Answer 18

Its good at predicting dose.

However has problems in regions of inhomogeneity

Pinnacle uses PB for electrons.

Question 19

What is the Monte Carlo algorithm?

Answer 19

It picks events from a probability distribution using random generated numbers

Uses probability to estimate the amount of photoelectric, compton scatter and pair production interactions. - guesses all the possible interactions the x-rays could have.

More accurate than using a water phantom.

Question 20

What are limitations of Monte Carlo?

Answer 20

It is subject to statistical uncertainty in dose.

This is proportional to the number of histories (the # of particles generated by the source model)

Question 21

How do you increase the speed of a Monte Carlo algorithm?

Answer 21

You must decrease the uncertainty.
E.g. If
you decrease the uncertainty by a factor of 2, you need 4
times more particle histories. This increases the calculation
time by a factor of 4.

However this can reduce.
See lecture: 1:00:00

Question 22

watch the lecture

Answer 22

??