Improvement of the Monte Carlo Simulation Efficiency of a Proton

Therapy Treatment Head Based on Proton Tracking Analysis and

Geometry Simplifications

Miguel A. Cortés-Giraldo*, José M. Quesada, M. Isabel Gallardo

(Universidad de Sevilla)

Harald Paganetti(Massachusetts General Hospital - Boston, MA, USA)

6th DITANET Topical Workshop on Particle Detection TechniquesSeville (Spain)

November 8th, 2011

(*) e-mail: miancortes@us.es

Contents

Introduction Methods Results Conclusions

Introduction Methods Results Conclusions

Motivation Monte Carlo (MC) simulations are:

A precise technique to calculate dose in patients… but expensive in terms of CPU time.

The aim of this work is: To decrease the CPU time needed to create a

phase-space file in the MC simulation of a passive scattering proton therapy treatment head.

To develope techniques capable of increasing the computational efficiency in the simulation of nozzles with similar geometry.

The MC code (phase-space files)

Geant4.9.0.p01

Only proton tracking is taken into account in detail in order to create a phase-space file as fast as possible.

Secondary radiation is evaluated separately

Monte Carlo treatment head model:Paganetti et al. Med. Phys. 31:2107-18

(2004)

Physics settings (Geant4 physics list):Zacharatou and Paganetti IEEE-TNS 55:1018-25

(2008)

Francis H Burr Proton Therapy Center (Boston, MA, USA)

Introducton Methods Results Conclusions

Methodology

The efficiency improvement is evaluated for various nozzle set-ups:

Covering the energy range of the proton beam.

Output efficiency: 25-cm (maximum) and 12-cm diameter snout (most typical case in proton therapy).

Validation with published results.

Identical computational conditions.

(Paganetti et al. Med. Phys. 31:2107-18, 2004.)

Time spent along the nozzle

IC2

2nd scattererRMW

IC1

Modulator Wheels PatientAperture

Compensator

Double scattering system

Modulator Wheels PatientAperture

Compensator

Double scattering system

Proton tracking filtering

The basic idea is to terminate the tracking of protons which, very likely, will not reach the aperture

Modulator Wheels PatientAperture

Compensator

Double scattering system

Modulator Wheels PatientAperture

Compensator

Double scattering system

Proton tracking filtering

There is a strong correlation of the protons reaching the nozzle exit and their dynamical conditions at the exit of the scatterer.

An example… A tolerance margin is taken into account.

Open field conditions.

Simplifications of the monitor chambers

A detailed geometry model of the monitor chambers slows down the MC simulation.

Considering all the layers grouped together simplifies the tracking of particles, improving the efficiency.

Production cuts per region

Production cut: key parameter in Geant4 simulations.

The secondary production cut value is higher in regions filled by air (magnets, jaws…)

The scattering and modulation devices require a lower value of the production cuts.

Modulator Wheels PatientAperture

Compensator

Double scattering system

Modulator Wheels PatientAperture

Compensator

Double scattering system

Introduction Methods Results Conclusions

Proton tracking filtering

The efficiency increases by about 30% with a 12 cm snout. In the worst case scenario (25

cm), it improves by about 5%.

Simplifications of the monitor chambers

The efficiency improvement varies between 5% and 15%. The improvement increases with the

proton beam range

Production cuts per region

0.2 mm for devices filled by air (jaws, aperture…); the CPU time decreases by about 5%.

For scatterers and modulators the production cut value is 0.05mm.

Modulator Wheels PatientAperture

Compensator

Double scattering system

Modulator Wheels PatientAperture

Compensator

Double scattering system

Using a global production cut value too high may change the energy distribution at the exit of the nozzle.

(Geant4.9.0.p01)

Output fluence verification

12 cm diameter snoutRange = 12.00 cm

Modulation width = 4.0 cm

Output fluence verification

12 cm diameter snoutRange = 17.19 cm

Modulation width = 6.78 cm

New time profiles

12 cm diameter snout 25 cm diameter snout

Introduction Methods Results Conclusions

Conclusions We have developed techniques to increase the

computational efficiency of Geant4 simulations to obtain phase-space files of a passive scattering proton therapy nozzle.

For the most typical case in the facility, the efficiency increases by about 35%; in the worst case scenario, it improves by about 15%.

These techniques can be applied to other treatment heads, simulated either with Geant4 or another MC transport code.

Acknowledgements

Ministerio de Ciencia e Innovación (P07-FQM-02894 y FIS2008-04189).

Junta de Andalucía (FPA2008-04972-C03-02).

PO1 Grant.

Physics Research group at Dep. Radiation Oncology (Massachusetts General Hospital, Boston, MA, USA).