1 Gean4 medical_linac advanced example: A Geant4 accelerator model Dr. S. Guatelli Geant4 Collaboration member, Lecturer, Centre of Medical Radiation Physics,

1

Gean4 medical_linac advanced example: A Geant4 accelerator model

Dr. S. GuatelliGeant4 Collaboration member,

Lecturer, Centre of Medical Radiation Physics, Engineering Physics Department, University of Wollongong, NSW, Australia

[email protected]

Geant4 School, 12-14 October 2009, Catania, Italy2

Development of the medical_linac G4 application

Simulation of a typical linear accelerator used for radiation therapy, in particular IMRT

M. Piergentili, Master Thesis, Dipartimento di Fisica, Universita’ di Genova, Italy

Supervisors: F. Foppiano (IST, Genova) and M. G. Pia (INFN, Genova)

Code available in: Geant4/examples/advanced/medical_linac


Content Context User requirements Architecture Implementation details

Geometry component Primary particles Physics component User Interface Visualisation

Results Summary and conclusions


Context

Problem! The most used treatment planning systems

do not accurately predict dose in small fields where lateral equilibrium is not achieved This is crucial for IMRT

Solution! Monte Carlo simulation


Modeling a medical accelerator

Crucial issues: Complex geometries to be

modelled Accurate description of

the radiation field Accurate physics modeling Simple user interfaceSolution: Geant4!

6

Why Geant4?

Versatile and powerful MC Toolkit in: Physics modelling

Beam line descriptionDosimetry

User interface and visualisation


Multiple scattering BremsstrahlungIonisationAnnihilationPhotoelectric effect Compton scattering Rayleigh effect conversione+e- pair productionSynchrotron radiationTransition radiationCherenkovRefractionReflectionAbsorptionScintillationFluorescenceAuger

Electromagnetic physics

Low energy extensions are fundamental for Geant4 Medical Physics applications

Model based on evaluated data libraries

Penelope processes completely re-engineered in Geant4 thanks to OO Technology

• electrons and positrons• , X-ray and optical photons• charged hadrons• ions• muons


Detailed detector description and efficient navigation

CSGCSG (Constructed Solid Geometries)- simple solids

BREPSBREPS (Boundary REPresented Solids)- volumes defined by boundary surfaces- polyhedra, cylinders, cones, toroids etc.

Boolean solidsBoolean solids- union, subtraction…

Geometry

ATLASBaBar


Radiation field definition

ParticleGun beam of selectable particle type, energy etc.

GeneralParticleSource provides sophisticated facilities to model a

particle sourcehttp://reat.space.qinetiq.com/gps/

It is possible to input radiation field phase space files, to generate primary particles in a simulation

10

Aim of the G4 medical_linac

Perform dosimetry of a medical linac for IMRT in a

water phantom


Software development of the G4 medical_linac example

User requirements

Architecture of the software

Implementation details

Use of the G4 medical_linac

12

Main user requirements of the G4 medical_linac

What should the software application do

Geometry component

Primary particle

Physics

User interface and visualisation


Geometry UR:beam line description

The beam line components to be described are: Primary collimator The target The vacuum window The flattening filter The mirror The secondary movable collimators (jaws) The multileaf collimator

The user should be able to change the position of the jaws

The user should be able to change the position of every single leaf

The water phantom is set at the end of the beam line


Geometry UR:Phantom definition

The detector is a phantom

The phantom is: made of water is divided in voxels

The distance between the upper surface of the target and the upper surface of the phantom is 100 cm


Primary particle UR:Definition of the radiation field

The primary particles are electrons generated in a point source with random direction (0 deg<Θ<30

deg) the energy has a gaussian distribution the momentum distribution is gaussian


Physics component UR

Gamma, electrons and positrons should be defined in the experimental set-up

The electromagnetic processes should be defined

The user shall be able to change the threshold of production of secondary particles interactively


Analysis UR

The user shall be able to retrieve the energy deposition in the phantom and store it in analysis objects



User requirements





Design of medical_linac

Flexibility Extensibility

Decorator design pattern for geometry component


Decorator design patternTo model geometries in the medical_linac

Decorator design pattern - attach additional responsibilities to an object dynamically

Advantage: develop a general, flexible, extensible software with the capability of choosing a specific experimental set-up

Dynamically loadable geometries: Position of the collimators Set the target and flattering filter in the

experimental set-up



User requirements





Geant4/examples/advanced/medical_linac


Beam line model +

phantom

Flattening filter

MLC

Primary collimators and target

M. Piergentili


Primary particle component:MedLinacPrimaryGeneratorAction class (1)

MedLinacPrimaryGeneratorAction::MedLinacPrimaryGeneratorAction() :pEnergy(99.0*MeV), sigma(12.7*MeV)

{…….

particleGun -> SetParticleDefinition(particleTable -> FindParticle("e-"));

particleGun -> SetParticleEnergy(pEnergy);

particleGun -> SetParticlePosition(G4ThreeVector(0.0*cm,

0.0*cm, 123.0*cm));

}

MedLinacPrimaryGeneratorAction::~MedLinacPrimaryGeneratorAction() {….}


void MedLinacPrimaryGeneratorAction::GeneratePrimaries(G4Event* anEvent){

// Angular distribution of primary particlesG4double cosTheta = CLHEP::RandGauss::shoot(-1.,0.00003);

G4double phi = twopi * G4UniformRand();

G4double sinTheta = std::sqrt(1. - cosTheta*cosTheta); G4double ux = sinTheta*std::cos(phi); G4double uy = sinTheta*std::sin(phi); G4double uz = cosTheta;

particleGun -> SetParticleMomentumDirection(G4ThreeVector(ux,uy,uz));

// Energy distribution of primary particles

energy = CLHEP::RandGauss::shoot(pEnergy,sigma); particleGun -> SetParticleEnergy(energy);

particleGun -> GeneratePrimaryVertex(anEvent);}

Primary particle component:MedLinacPrimaryGeneratorAction class (2)


Physics component: particle definition

MedLinacPhysicsList class (1)

MedLinacPhysicsList::MedLinacPhysicsList(): G4VUserPhysicsList(){ // Instantiate a messenger to change the parameters of the physics

list physicsListMessenger = new MedLinacPhysicsListMessenger(this);}

MedLinacPhysicsList::~MedLinacPhysicsList(){;}

void MedLinacPhysicsList::ConstructParticle(){ G4Gamma::GammaDefinition(); G4Electron::ElectronDefinition(); G4Positron::PositronDefinition();}


Physics component: definition of

processesvoid MedLinacPhysicsList::ConstructProcess(){ AddTransportation(); ConstructEM();}void MedLinacPhysicsList::ConstructEM(){…..if (particleName == "gamma") { pmanager->AddDiscreteProcess(new G4LowEnergyRayleigh); pmanager->AddDiscreteProcess(new G4LowEnergyPhotoelectric); pmanager->AddDiscreteProcess(new G4LowEnergyCompton); pmanager->AddDiscreteProcess(new G4LowEnergyGammaConversion);}

else if (particleName == "e-") { pmanager->AddProcess(new G4MultipleScattering, -1, 1,1); pmanager->AddProcess(G4LowEnergyIonisation, -1, 2,2); pmanager->AddProcess(G4LowEnergyBremsstrahlung, -1,-1,3);} else if (particleName == "e+") {

pmanager->AddProcess(new G4MultipleScattering,-1, 1,1); pmanager->AddProcess(new G4eIonisation, -1, 2,2); pmanager->AddProcess(new G4eBremsstrahlung, -1,-1,3); pmanager->AddProcess(new G4eplusAnnihilation, 0,-1,4);} }



void MedLinacPhysicsList::SetCuts(){…defaultCutValue = defaultCut;

// Production thresholds for detector regions G4String regionName = "PrimaryCollimatorLow"; G4double cutValue = 8.*cm; G4Region* region = G4RegionStore::GetInstance()-

>GetRegion(regionName); G4ProductionCuts* cuts = new G4ProductionCuts; cuts -> SetProductionCut(cutValue); region -> SetProductionCuts(cuts);……}

Physics component: definition of

cuts



Dosimetry in medical_linac (1)

The result of the simulation is the energy deposition in the phantom

The phantom is A water box, with size = 30 cm

The sensitive detector of the experimental set-up

The phantom is divided in voxels (size = 2 mm) G4MedLinacReadOutGeometry

The ReadOut geometry is a virtual, parallel geometry to describe the read-out configuration of the detector


Dosimetry in medical_linac (2) Hits

Hit is a user-defined class derived from G4VHit

You can store various types information by implementing your own concrete Hit class:

position and time of the step

momentum and energy of the track

energy deposit of the step

geometrical information

etc.

Hit objects of a concrete hit class must be stored in a dedicated collection, which is instantiated from G4THitsCollection template classG4THitsCollection template class


Analysis component

The energy deposition in the voxels of the phantom is Step 1: stored in a G4Hit collection Step 2: stored in analysis objects, in a xml file

MedLinacAnalysisManager.cc

AIDA and PI used


User interface

Definition of messenger classes to change parameters of the simulation experimental set-up:

Mean energy of the electron beam Position of the leafs Rotation of the jaws Phantoms sizes Threshold of production of secondary particles

Easy to define simulation parameters in a macro file

Geant4 School, 12-14 October 2009, Catania, Italy

G4MedLinacPhysicsMessenger.cc

MedLinacPhysicsListMessenger::MedLinacPhysicsListMessenger(....){ PhysicsDir = new G4UIdirectory("/PhysicsList/"); PhysicsDir -> SetGuidance("physics parameters"); CutCmd = new G4UIcmdWithADoubleAndUnit("/PhysicsList/cut",this); CutCmd -> SetGuidance("Set cut length (mm)"); CutCmd -> SetParameterName("defaultCut",false); CutCmd -> SetDefaultUnit( "mm" ); CutCmd -> SetUnitCategory("Length"); CutCmd -> AvailableForStates(G4State_PreInit);}..........void MedLinacPhysicsListMessenger::SetNewValue(G4UIcommand* command,G4String

newValue){ if( command == CutCmd ) pMedLinacPhysicsList -> SetCut(CutCmd -> GetNewDoubleValue(newValue));} 33


/PhysicsList/cut 0.2 mm

/run/initialize

/Phantom/dimension 15. cm

/Phantom/Nvoxels 150

/Phantom/maxStep 0.2 mm

/MLC/leaf_selection a1 # Multi Leaf

/MLC/position 0. cm

/Jaws/X1/DistanceFromAxis -4.2 cm

/Jaws/X2/DistanceFromAxis 5.2 cm

/Jaws/Y1/DistanceFromAxis -6.6 cm

/Jaws/Y2/DistanceFromAxis 6.8 cm

/primary/energy 6.0 MeV

/primary/sourceType 0.127 MeV

/run/beamOn 10

Macro file

Geometry

Physics

Primary particles


How to run the medical_linac example

Install Geant4 Install analysis tool Source the environment Compile the code

Run the executable $G4WORDIR/bin/Linux-g++/MedLinac

Guidance options:OpenGL visualisation: $G4WORDIR/bin/Linux-g++/MedLinac vis.mac

DAWN visualisation: $G4WORDIR/bin/Linux-g++/MedLinac dawnvis.mac

VRML visualisation: $G4WORDIR/bin/Linux-g++/MedLinac vrmlvis.mac

without visualization: $G4WORDIR/bin/Linux-g++/MedLinac macro.mac

36

Simulation execution


Initialisationmain Run manager user detector

const ructionuser physics

list

1: initialize2 : const ruct

3: material const ruct ion

4: geometry construct ion5: world volume

6 : const ruct

7 : physics process const ruction

8: set cuts

Describe the geometrical set-up

Activate electromagnetic/hadronic

processes appropriate to the energy range of the experiment


Beam Onmain Run Manager Geometry

managerEvent

generatorEvent

Manager

1: Beam On2: close

3: generate one event

4: process one event

5: open

Generate primary events


Event processing

Event manager

Stacking manager

Tracking manager

Stepping manager

User sensitive detector

1: pop

2: process one track3: Stepping

4: generate hits

5: secondaries

6: push

Record the energy deposit and the position associated


Analysis dynamic flow

41

Results

Comparison experimental measurements and simulation

results

Authors: F. Foppiano (IST, Genova), B. Mascialino, M. G. Pia (INFN, Genova), M.

Piergentili


Test Important and crucial issue

Always with respect to experimental experimental datadata

Levels of tests:: Microscopic test: Test on components of the

application (i.e. Physics models adopted)

Macroscopic test: Test on the experimental set-up (physics, geometry, etc. together)


Percent Depth Dose

Experimental measurements with ion chamber

Squared fields 5x5 cm, 10x10 cm, 40x40 cm

PTW MP3


Comparison with experimental data

range D p-value

-84 -60 mm 0.385 0.23

-59 -48 mm 0.27 0.90

-47 47 mm 0.43 0.19

48 59 mm 0.30 0.82

60 84 mm 0.40 0.10

range D p-value

-56 -35 mm 0.26 0.89

-34 -22 mm 0.43 0.42

-21 21 mm 0.38 0.08

22 32 mm 0.26 0.98

33 36 mm 0.57 0.13

Kolmogorov-Smirnov test

1010 events

100 CPU dayson Pentium IV 3 GHz


Comparison with experimental data

D = 0.005; p-value = 1

range D p-value

0 14 mm 0.548 0.09

15 300 mm

0.144 0.123

Kolmogorov-Smirnov test

Voxels 5mmVoxels 1mm

46

Summary

Software development of medical_linac

Experimental validation of the simulation


Modeling a beam line Accurate modeling of a complete beam line in terms of:

Geometry: Materials Geometry components Magnetic fields Moving leafs

Particle beam line: Define the particle type Define the energy, angular distribution of the beam with a defined

algorithm

Complete tracking of primary and secondary particles in the experimental set-up


Problem: how to obtain quick response!

Optimise the simulation Use of G4Regions Variance reduction techniques Split the simulation in two

subcomponents Step 1: Tracking of the electron beam down to the phantom

interface Step2: Store radiation field phase space at the phantom

interface (input of Step 3) Step3: Repeat simulations, tracking the radiation field just in

the phantom, to accumulate significant statistics


Parallelisation of the G4 simulation

Use a cluster or distributed computing resources (GRID)

Possible solution: http://cern.ch/ganga http://cern.ch/diane Simulation and production of analysis objects e.g. histograms, in parallel mode

- Medical Physics – brachytherapy - Radioprotection

- Detector Simulation

50

Thank you !

Questions?

Documents

1 Gean4 medical_linac advanced example: A Geant4 accelerator model Dr. S. Guatelli Geant4 Collaboration member, Lecturer, Centre of Medical Radiation Physics,