Monte Carlo Simulations in Diagnostic Radiology Greg Stevens Clinical and Radiation Physics Plymouth Hospitals NHS Trust, Derriford Road, Plymouth, PL6 8DH [email protected]

Assumptions I have assumed

you have not: 1. learned a programming language 2. had experience with non-Microsoft OS’s 3. used MC codes I you do have: 1. access to a PC with administrative rights 2. You have network access 3. some computer literacy I chose EGSnrc because: 1. EGS is not export restricted 2. EGS is free for research activity 3. well supported 4. lots of peer reviewed papers (in diagnostic work) 5. easy to set up on Microsoft Windows

Monte Carlo in Diagnostic Physics Greg Stevens

Introduction Assumptions 1/1 Intro to MC CH technique Var Reduction

Installation what you need one warning

Building the Sim specify tube specify params

Running the sim no VR increasing kVp increasing filtration

Examples Education Publications Personal

Monte Carlo in Diagnostic Physics

Introduction

Greg Stevens

Photons

PE

electrons

Introduction

inelastic

Compton

Installation

Brem Rayleigh

Assumptions Intro to MC 1/3 CH technique Var Reduction what you need one warning

elastic

Building the Sim specify tube specify params

Running the sim

1. photon interaction: Compton, Rayleigh, Thompson, PE 2. PE and Compton: 



e-

3. electron interactions: elastic, inelastic and Brem 4. bremsstrahlung: e-





5. mathematically we have a coupling of the e- - field

no VR increasing kVp increasing filtration

Examples Education Publications Personal

Monte Carlo in Diagnostic Physics

Introduction

Greg Stevens

Photons

PE

electrons

Introduction

inelastic

Compton

Installation

Brem Rayleigh

what you need one warning

elastic

1. particles created: 1. 2.

known distribution randomly sampled

2. particles move a distance: 1. 2.

according to a probability distribution highly dependent on total cross section

3. particles interact: 1.

according to the differential cross section

Assumptions Intro to MC 2/3 CH technique Var Reduction

results are highly dependent on the Xsection data

Building the Sim specify tube specify params

Running the sim no VR increasing kVp increasing filtration

Examples Education Publications Personal

Monte Carlo in Diagnostic Physics

Introduction

Greg Stevens

Photons

PE

electrons

Introduction

inelastic

Compton

Installation

Brem Rayleigh

what you need one warning

elastic

• All  interactions dominate: 1. 2.

depending on material Depending on energy

• Many elastic and inelastic e- interactions occur 1. 2.

Little direction change Loose small amounts of energy/step

Assumptions Intro to MC 3/3 CH technique Var Reduction

Very difficult to run single interaction computations due to limited computing power

Building the Sim specify tube specify params

Running the sim no VR increasing kVp increasing filtration

Examples Education Publications Personal

Monte Carlo in Diagnostic Physics

CH Technique (Berger 1963) B A

Greg Stevens

Introduction Assumptions Intro to MC CH technique 1/3 Var Reduction

Installation what you need one warning

Building the Sim

• e- interactions tend to: • •

loose energy semi continuously, do not have large direction changes

• large numbers of transport/collision processes condensed into a single step • cumulative effect accounted for by sampling over relevant scatter distributions at end of step artificial “step length”

specify tube specify params

Running the sim no VR increasing kVp increasing filtration

Examples Education Publications Personal

Monte Carlo in Diagnostic Physics

CH Technique (Berger 1963) B A

Greg Stevens

Introduction Assumptions Intro to MC CH technique 2/3 Var Reduction

Installation what you need one warning

Building the Sim specify tube specify params

• Condensed History approach fails when a particle travels parallel to a boundary:

Running the sim no VR increasing kVp increasing filtration

Examples Education Publications Personal

Monte Carlo in Diagnostic Physics

CH Technique (Berger 1963) B A

Greg Stevens

Introduction Assumptions Intro to MC CH technique 3/3 Var Reduction

Installation

E≠E

what you need one warning

Building the Sim specify tube specify params

• Condensed History approach fails when a particle travels perpendicular to a boundary: • •

Finite probability of particle in other medium Energy of particle along alternate route will be different

• boundary crossing algorithm: single step if D < Dthreshold

Running the sim no VR increasing kVp increasing filtration

Examples Education Publications Personal

Variance Reduction

Monte Carlo in Diagnostic Physics Greg Stevens

Family of techniques used to improve the efficiency of the calculation by increasing the importance of particles with a large influence on the final result, and rejecting those that have little effect. Thus reducing statistical variance

Introduction Assumptions Intro to MC CH technique Var Reduction 1/2

Installation what you need one warning

Building the Sim specify tube specify params

Running the sim

Quantities of interest: average over histories (N) Statistical uncertainty generally goes as N-0.5

reduce uncertainty by artificially increasing N using 3 variance RT here: PCUT, ECUT and Bremsstrahlung Splitting

no VR increasing kVp increasing filtration

Examples Education Publications Personal

Variance Reduction Electron Range

Monte Carlo in Diagnostic Physics Greg Stevens

Introduction Assumptions Intro to MC CH technique Var Reduction 2/2

Installation

ECUT/PCUT Energy of particle followed until it reaches PCUT (for photons) of PCUT thresholds Energy deposited locally at this point Bremsstrahlung Splitting • Result of Brem event is N photons each having fraction of the e- energy – energy of last brem photon • Only conserves energy on “average”

what you need one warning

Building the Sim specify tube specify params

Running the sim no VR increasing kVp increasing filtration

Examples Education Publications Personal

Installation TCL/TK • • • •

http://www.activestate.com/activetcl Required for all BEAMnrc GUIs downloadable as a single “Program” community addition freely downloadable

Monte Carlo in Diagnostic Physics Greg Stevens

Introduction Assumptions Intro to MC CH technique Var Reduction

Installation

EGSnrc • www.nrccnrc.gc.ca/eng/solutions/advisory/egsnrc_index.html • handles coupled electron photon transport (1keV to 1MeV)

what you need 1/2 one warning

Building the Sim specify tube specify params

Running the sim no VR increasing kVp increasing filtration

BEAMnrc • www.nrccnrc.gc.ca/eng/solutions/advisory/beam_index.html • code generator for EGSnrc specifically designed for Linac simulation

Examples Education Publications Personal

Installation

Monte Carlo in Diagnostic Physics Greg Stevens

Introduction Assumptions Intro to MC CH technique Var Reduction

Installation what you need 2/2 one warning

Building the Sim specify tube specify params

Running the sim no VR increasing kVp increasing filtration

Examples Education Publications Personal

Monte Carlo in Diagnostic Physics

Installation -Warning

Greg Stevens

Introduction

Enable this

Assumptions Intro to MC CH technique Var Reduction

Installation what you need one warning 1/1

Building the Sim specify tube specify params

Running the sim no VR increasing kVp increasing filtration

Examples Education Publications Personal

Specify the Tube File -> Specify new accelerator need XTUBE and SLABS save and close (keep in default directory) choose PEGS file (choose 521icru.pegs4dat in HENHOUSE)

Monte Carlo in Diagnostic Physics Greg Stevens

Introduction Assumptions Intro to MC CH technique Var Reduction

Installation what you need one warning

Building the Sim specify tube 1/2 specify params

Running the sim no VR increasing kVp increasing filtration

Examples Education Publications Personal

Specify the Tube • • • • • • • • • • •

Half width = z thickness =2cm distance of CM too ref plane = 3cm anode angle = 9 number of slabs = 2 W embedded in Cu target S1: Be of 0.0025cm thickness 1 layer with central extra region S2: Al of 3mm thickness central region 1mm x 1cm x 1cm for Al: dose scoring = 1 no dose scoring

Monte Carlo in Diagnostic Physics Greg Stevens

Introduction Assumptions Intro to MC CH technique Var Reduction

Installation what you need one warning

Building the Sim specify tube 2/2 specify params

Running the sim no VR increasing kVp increasing filtration

Examples Education Publications Personal

Monte Carlo in Diagnostic Physics

Specify the Parameters

Greg Stevens

EGSnrc Parameters

Introduction Assumptions Intro to MC CH technique Var Reduction

Parameter

Value

electron impact ionisation

off → penelope

Brems angular sampling

simple → KM

Brems cross section

BH → NRC

photoelectron angular sampling

BH → NRC

rayleigh scattering

off → on

specify tube specify params 1/2

atomic relaxation

off → on

Running the sim

photon cross section

xcom → epdl

no VR increasing kVp increasing filtration

Installation what you need one warning

Building the Sim

Examples Education Publications Personal

Monte Carlo in Diagnostic Physics

Specify the Parameters

Greg Stevens

Main Inputs

Introduction

Parameter

Value

number histories

107

Brem splitting

Uniform

ECUT (MeV)

0.512

PCUT (MeV)

0.01

Scoring planes

1

Incident particle

electron

Source number

10 (-1,0,0)

Assumptions Intro to MC CH technique Var Reduction

Installation what you need one warning

Building the Sim specify tube specify params 2/2

Running the sim no VR increasing kVp increasing filtration

Examples Education Publications Personal

Running the Simulation • • • • • • •

Execute -> compile (fail) -> compile (pass) Run -> execute Wait a long time Will then have a .egsphsp1 file Run beamdp Select “derive spectral distribution” – use .agr for o/p file Analyse data as you’d like - (qtgrace, sciDAVis, excel etc)

Monte Carlo in Diagnostic Physics Greg Stevens

Introduction Assumptions Intro to MC CH technique Var Reduction

Installation what you need one warning

Building the Sim specify tube specify params

Running the sim no VR 1/1 increasing kVp increasing filtration

Examples Education Publications Personal

Running the Simulation • • • • • • •

Execute -> compile (fail) -> compile (pass) Run -> execute Wait a long time Will then have a .egsphsp1 file Run beamdp Select “derive spectral distribution” – use .agr for o/p file Analyse data as you’d like - (qtgrace, sciDAVis, excel etc)

Monte Carlo in Diagnostic Physics Greg Stevens

Introduction Assumptions Intro to MC CH technique Var Reduction

Installation what you need one warning

Building the Sim specify tube specify params

Running the sim no VR increasing kVp 1/1 increasing filtration

Examples Education Publications Personal

Running the Simulation • • • • • • •

Execute -> compile (fail) -> compile (pass) Run -> execute Wait a long time Will then have a .egsphsp1 file Run beamdp Select “derive spectral distribution” – use .agr for o/p file Analyse data as you’d like - (qtgrace, sciDAVis, excel etc)

Monte Carlo in Diagnostic Physics Greg Stevens

Introduction Assumptions Intro to MC CH technique Var Reduction

Installation what you need one warning

Building the Sim specify tube specify params

Running the sim no VR increasing kVp increasing filtration

Examples Education Publications Personal

Monte Carlo in Diagnostic Physics

Education

Greg Stevens

Introduction Assumptions Intro to MC CH technique Var Reduction

Installation what you need one warning

Building the Sim specify tube specify params

Running the sim no VR increasing kVp increasing filtration

Examples Education 1/1 Publications Personal

PHOTOELECTRIC

COMPTON

Publications David Platten: A MC study of the energy spectra and transmission characteristics of scattered radiation from CT- June publication JRP.

Monte Carlo in Diagnostic Physics Greg Stevens

Introduction Assumptions Intro to MC CH technique Var Reduction

Installation what you need one warning

Building the Sim specify tube specify params

Running the sim no VR increasing kVp increasing filtration

Examples Education Publications 1/1 Personal

Personal

Improve on this • Humans not cylindrical • Use MC to model where beam subtends skin • Electronic phantom or CT?

Monte Carlo in Diagnostic Physics Greg Stevens

Introduction Assumptions Intro to MC CH technique Var Reduction

Installation what you need one warning

Building the Sim specify tube specify params

Running the sim no VR increasing kVp increasing filtration

Examples Education Publications Personal 1/1