A Geant4 based treatment plan verification tool: commissioning of the LINAC model and DICOM-RT interface

Transcription

1 A Geant4 based treatment plan verification tool: commissioning of the LINAC model and DICOM-RT interface Iwan Cornelius, Chris Poole. Stephen Warwick, Christian Langton Discipline of Physics, Faculty of Science and Technology Institute of Health and Biomedical Innovation Queensland University of Technology, Brisbane, Australia Brendan Hill, Nigel Middlebrook Department of Medical Physics, Premion, Brisbane, Australia Brad Oborn Illawarra Cancer Care Centre, Wollongong, Australia

2 Introduction Monte Carlo is a useful tool for the independent verification of treatment plans in radiotherapy more accurate than analytical methods of calculating dose in patient geometries MC verification tools created using BEAMnrc, PENELOPE Geant4 includes many features that make it attractive for use in radiotherapy Ability to model CAD geometries Accurate physics models (including photonuclear models) Time dependent geometries Multi-threaded version in beta testing Bush K, Townson R and Zavgorodni S 2008 Monte Carlo simulation of RapidArc radiotherapy delivery Jan S et al 2011 GATE V6: a major enhancement of the GATE simulation platform enabling modelling of CT and radiotherapy

3 Aims To create Monte Carlo tool for the simulation of radiotherapy LINACs using GEANT4 Validate Apply to research projects Combine this with a DICOM-RT interface to create an independent treatment plan verification tool Validate Use in a clinical setting

4 LINAC Geometry Varian 6X clinac Vendor documents incomplete TGT: target PC: primary collimator FF: CAD modelled flattening filter IC: ionisation chamber JAWSY: upper jaws in y direction JAWSX: lower jaws in x direction MLCs: CAD modelled multileaf collimators Constantin M et al, 2010 Linking computer-aided design (CAD) to Geant4-based Monte Carlo simulations for precise implementation of complex treatment head geometries Phys. Med. Biol. 55 N211 N220

5 CAD Interface Follows approach of Constantin et al Components created in SolidWorks Export to STEP format 5

6 CAD Interface Follows approach of Constantin et al Components created in SolidWorks Export to STEP format Import into 3 rd party software FASTRAD Export as GDML (Geometry Description Markup Language) 6

7 CAD Interface Follows approach of Constantin et al Components created in SolidWorks Export to STEP format Import into 3 rd party software FASTRAD Export as GDML (Geometry Description Markup Language) Import into simulation via GDML parser (tesselated solid) Step-by-step how-to document available for anyone who s interested Chris Poole s talk Alternate method 7

8 Geometry Varian ix clinac Degrees of freedom of LINAC components are reproduced Physics IEC compliant Standard parameterised electromagnetic physics models Photons: compton, PE, pair production Electrons: bremm, ionisation, multiple scatter Primary Beam Monoenergetic electron beam incident on target Gaussian beam profile. UI commands to modify geometry Treatment field can be simulated via macro file

9 Variance Reduction Range Cuts by region 100 micron in ROIs Uniform bremsstrahlung splitting (UBS) Splitting number Nr = 10 Tinslay et al, 2007 Binary phase space files Some aspects of geometry remain unchanged between runs 3 stage simulation Popescu et al, 2005 Popescu I A, Shaw C P, Zavgorodni S F and W A Beckham 2005 Absolute dose calculations for Monte Carlo simulations of radiotherapy beams Phys. Med. Biol Tinslay J, Perl J and Asai M 2007 Verification of Bremsstrahlung Splitting in Geant4 for Radiotherapy Quality Beams Med. Phys. 34(6)

10 Variance Reduction Phase 0 Simulation of electron beam incident on target Em cascade particles scored at plane below ionisation chamber Need only simulate once per linac

11 Variance Reduction Phase 1 phase space particles sampled and second scoring plane below MLCs does not change for a control point

12 Variance Reduction Phase 2 particles transported through patient / phantom geometry phase space files recycled Nr=25 times Kawrakow et al, 2006 Kawrakow I and Walters B R B 2006 Efficient photon beam dose calculations using DOSXYZnrc with BEAMnrc Med. Phys

13 Phantom / Dose Scoring phantom geometry water or solid water (Gammex) voxellised geometry G4PVReplica G4VNestedParameterisation adapted from example RE02 primitive dose scorer G4PSDoseDeposit

14 Phantom / Dose Scoring Dose in each voxel is accumulated after each event D(x,y,z) Dose per primary particle is calculated after each run error is calculated by S=

15 QUT High Performance Computing: LYRA PBS queuing system Each process unique random number generator seed based on process PBS_JOBID Approx GHz 64bit Intel Xeon processors per job Collation of results Copying result back to desktop for analysis 1-2 hours per field (run) for 5x10 8 primaries 15

16 Validation: water phantom measurements 16

17 Validation: water phantom measurements 10cmx10cm PDD Profiles Z=1.5cm Z=10cm Results used for beam tuning E = 6.0 MeV sigma = 1.1 mm Flattening filter density!!! Z=30cm 17

18 Validation: water phantom measurements 4cmx4cm PDD Profiles Z=1.5cm Z=10cm Z=30cm 18

19 Validation: water phantom measurements 1cmx1cm PDD Profiles Z=1.5cm Z=10cm Gamma criteria of 3 mm / 3% satisfied for > 98% of datapoints Z=30cm 19

20 MLC model validation: experiment For IMRT and RapidArc treatments, contribution to patient dose from MLC leakage can be significant Need to verify MLC CAD model EBT2 film at 5cm in solid water SSD 100cm, field size 10x10cm 2 MLCs fully closed Change in optical density proportional to dose Heath E and Seuntjens J 2003 Development and validation of a BEAMnrc component module for accurate Monte Carlo modelling of the Varian dynamic millennium multileaf collimator 20

21 MLC model validation: simulation 21

22 MLC leaf leakage: abutted Abutted leaf leakage profile (x=1cm) in direction of leaf motion Lower MUs to keep film from saturating 22

23 MLC leaf leakage: interleaf Interleaf leakage profile (y=4cm), perpendicular to leaf motion <%D sim >=1.27+/-0.07 vs <%D exp >=1.33+/

24 DICOM-RT interface TPS CT Treatment planning system (TPS) CT import Contouring Dose objectives

25 DICOM-RT interface TPS DOSE CT PLAN TPS produces PLAN file (with control points governing gantry, collimator angles, MLC positions, weighting of each beam) DOSE file With a 3D dose distribution calculated using analytical engine

26 DICOM-RT interface TPS CT G4 Need a DICOM-RT interface to import CT geometry and PLAN file based on Vega library of Zavgorodni et al, 2005 Wrapper class translates PLAN file into set of UI commands DOSE PLAN reads CT data and creates patient geometry using G4PVReplica volumes and G4VNestedParameterisati on class Work in progress Locke C and Zavgorodni S 2008 Vega library for processing DICOM data required in Monte Carlo verification of radiotherapy treatment plans

27 DICOM-RT interface TPS CT G4 Vega library enables production of DICOM- RT DOSE file PLAN DOSE DOSE

28 DICOM-RT interface TPS CT G4 Compare TPS dose distribution and Geant4 dose distribution using 3 rd party radiotherapy QA software I mrt PLAN MapCheck DOSE DOSE EVALUATE

29 Absolute Dose Calculation: Calibration Virtual monitor chamber TPS gives absolute dose distribution G4 dose per primary Absolute dose calculation possible by calibrating virtual LINAC monitor chamber analogous to calibration of actual LINAC monitor 10x10cm 2, SSD 100cm, calibration depth dmax Popescu et al, 2005 Popescu I A, Shaw C P, Zavgorodni S F and W A Beckham 2005 Absolute dose calculations for Monte Carlo simulations of radiotherapy beams

30 Absolute Dose Calculation general case given monitor units for treatment field, U monitor dose per primary particle to virtual monitor chamber calculate dose per primary particle calculate absolute dose Popescu I A, Shaw C P, Zavgorodni S F and W A Beckham 2005 Absolute dose calculations for Monte Carlo simulations of radiotherapy beams

31 DICOM-RT Validation: the chair test Used for QA of leaf leakage parameters and MLC drivers create plan to deliver chair shaped dose distribution deliver to phantom, compare with diode array measurements DICOM-RT PLAN file imported into Geant4 simulation Verify DICOM-RT interface, Control Points, MLC positions, absolute dose calculation Van Esch A et al 2002 Acceptance tests and quality control (QC) procedures for the clinical implementation of intensity modulated radiotherapy (IMRT) using inverse planning and the sliding window technique: experience from five radiotherapy departments 31

32 Results 32

33 DICOM-RT validation: simple treatment plan 33

34 Results 34

35 Conclusions A Geant4 based model of a 6MV Varian ix clinac has been developed Validated against experimental data Will form the basis of independent treatment plan verification tool for radiotherapy Potential usage for MC-based detector development / optimisation Paper available on arxiv.org 35

36 Work in Progress 18X filter Further validation studies RapidArc / IMRT plan delivered to homogeneous phantom Repeat for anthropomorphic phantom Further CAD modelling (Steve) Higher energy photons Flattening Filters Photonuclear reactions? Electron beams Christopher Poole s talk DICOM-RT ROIs, Elastic Cloud, alternate CAD interface MRI-LINAC merging Brad Oborn s MRI-LINAC work with application 36

37 UoW Geant4 workshop organisers High Performance Computing group, QUT T. Kairn, J. Kenny, Premion, discussion on EBT2 film and MLC modelling Acknowledgements

38 Questions? 38