Clinical Treatment Planning Algorithms Flashcards
What are the types of planning algorithms available?
Types A, B, C - type A doesn’t vary penumbra
Non-kernel based, line kernel based, point kernel based
What are the issues with correction based algorithms?
Radiation scatters and interacts again depending on anatomy
Patient is inhomogeneous and irregular
It’s impossible to measure all fields to be used, can’t just interpolate
How are model based algorithms validated?
In non-standard conditions
Is a correction based or model based algorithm better in complex situations?
Model based
What does the primary fluence represent?
Radiation source/target
Collimation
Energy of photons/particles
Reative number of photons/particles
What does the patient model represent?
The patient
Interaction of beam and patient
How energy (dose) is deposited
What are the components of a TPS?
Patient database RTP main module Functionalities Beam data database Commissioning module Dose engine IMRT module Optimisation engine
What is the energy fluence?
2D array at the isocentre plane, describing how energy of the beam is distributed across the plane for a particular beam. Uses 2 energy fluence maps - 1 for primary energy, 1 for head scatter energy fluence
How is the energy fluence calculated?
Beam model + treatment plan = energy fluence
How is direct energy fluence determined?
At isocentre for an open beam
Beam divergence is considered by inverse square scaling from reference distance using target as beam source origin focus
Obtain by modulating open beam fluence with attenuation from head components. Determine the modulation at a given position by ray tracing from each beam source array element down through the head to each element in the 2D fluence matrix
Back project for each fluence pixel and integrate over visible part of source
How is head scatter energy fluence determined?
Matrix of indirect fluence calculated at an estimated effective depth of treatment
Scaled using ISL but using an effective focus at the flattening filter
All the head scatter is back projected as if it comes from the flattening filter as that is more computationally efficient
The main source of photons
What are the components of the energy fluence?
Number of particles - matrix element
Position of matrix element
Direction - as if particles were coming from their respective source to the matrix element
Energy - given by beam spectrum common to all elements
At what distance is the open beam fluence matrix stored?
At reference distance - usually isocentre
How is the energy fluence grid aligned?
To the beam limiting device - fixed resolution = 1mm
How can the energy fluence be determined from measurement?
Measure unobstructed fluence with largest field size
Take star profiles every 10 degrees
Usually done in water but best in air
How is the energy fluence constructed in the TPS?
Use the beam energy (effective spectrum) and dose deposition kernels to perform iterative fitting to deduce the energy fluence matrix
Perform parameter fitting to find contribution from head scatter from various sources, with the flattening filter dominating
How are the modelling beams defined during planning?
Size/shape selected by dosimetrist
Planning system defines energy fluence arriving at the patient
Achieved by ray tracing from the source to each point in the calculation plane for each source of primary and scattered radiation
Gives 2D fluence map
What is the equation for the 2D fluence map?
= Energy fluence reference value.modulation.open beam + flattening filter correction + collimator correction + moduator correction
How is the source modelled?
An oval with width in-plane and cross-plane
Fixed in space
Size of source found emperically by fitting calculations to measured profiles
Model as gaussian in both directions
Discretised and modelled as a 2D distribution in TPS
Represents the focal spot of the electron beam on the target
How are the characteristic of the collimating devices obtained to find the fluence at the isocentre plane?
From the manufacturers
How is the model of the linac head completed once the fluence is calculated?
Project the source through the limiting devices to the isocentre plane
What components are affect the penumbra most for large and small fields?
Large: Flattening filter size
Small: Source size
What are the sources of the scattered fluence map?
Flattening filter, wedge, collimator edges
How is the scattered fluence map constructed?
Find visible parts of each source and integrate over them to get fluence magnitude
Project through point like source, so the fluence gets sharp edges
How do the fluence engine and dose engine combine?
Calculate photon fluence - calculate TERMA - perform convolution
Energy fluence + attenuation information - TERMA and PSF - dose
How is the patient modelled?
Using CT data
HU converted to mass density through linear interpolation in a CT to density table
Follows conversions: Photon attenutation - HU - mass density - elemental composition - electron density - effective density
What is the equation for HU?
HU = 1000. ((u-uH2O)/uH2O)
What is the definition of radiological depth?
Assume all matter is water - estimate the path length in water that gives same attenuation as tissue, based on mass density
lw = li.pi/pw
What are the issues of using radiological depth?
Overestimate of ~5% for air/bone for 10cm equivalent path length, get ~1% overestimate for other tissues
As compton scatter dominates electron density is more relevent
How is pei/pew found?
From tables in ICRP 23, ICRU 44
What is the error on the CT to effective density conversion?
<1% for 0.5-20MeV
What is TERMA?
Total energy released by primary photons to secondary particles and scattered photons
What doe the calculation of TERMA require?
Attenuation coefficients
Energy spectrum
Gaussians describing penumbrae
What is the equation for TERMA?
u/p . phi.e^(-uz)
How is TERMA calculated?
Compute radiological intersection length using voxel effective density
Compute average radiological depth at each voxel
Compute TERMA from average radiological depth in each voxel with primary source energy fluence
Project each voxel onto fluence plane and scale TERMA with fluence
How is SCERMA - the energy released to scattered photons calculated?
= (u-uen)/p . phi.e^(-uz)
How is dose deposition calculated?
Use predetermined energy deposition kernels to describe energy spread around a primary interaction - PSF or point kernel
Model using MC for scattering cross section, interaction types
What is a pencil kernel?
A point kernel preconvolved with depth dimension - use if only changing in 2D
What is a planar kernel?
Point kernel preconvolvd over 2D - only use if changing in 1D
What is a broad beam kernel?
Point kernel preconvolved over 3D to create a dose distribution
When are point kernels used
Complex situations where a 3D dose distribution is needed
What are the 2 types of algorithm?
Superposition/convolution - collapsed cone Simple convolution (FFT) - pencil beam
How does a pencil beam algorithm work?
Use pencil kernel, so can only change in 1D
Direct and indirect 2D fluence maps are found at 2 depths in patient - with the summation of the direct and indirect fluences making the total fluence
Convolve with pencil kernel at each depth
Linearly interpolate to find full dose distribution using the radiological path length
What are the issues with pencil beam algorithms?
Assume pencil kernel is invarient through patient
Doesn’t consider changes in lateral scatter with density
Only accounts for heterogeneities in depth
How does a collapsed cone algorithm work?
Mono-energetic point spread kernels simulated using EGSnrc
Build polyenergetic kernels for up to 600 depths (binned according to max radiological depth in patient)
Kernel depth dependence needed for depth hardening and off axis softening
PSF discretisd into cones travelling from interaction site
All energy emitted in a solid angle
Kernels can adapt to local environment
What direction are the majority of bins in CC?
Forward
What is accuracy compromised with in CC?
Speed
How is the dose deposition scaled in CC?
Radiological path length
How is the time of calculation reduced?
Calculate the dose to M, rather than dose from NxNxN points
How is the issue of changing energy at depth overcome?
Generate spectrums at 0cm and 10cm, generating tables. Interpolate between the depths
How is the change in energy by the flattening filter accounted for?
TERMA is corrected for different energy at different depths
