A Geant4/TOPAS Monte Carlo System for Validating Proton Therapy Treatment Plan Calculations at Mayo Clinic
H Wan Chan Tseung*, C Beltran, Mayo Clinic, Rochester, MNSU-E-T-499 Sunday 3:00PM - 6:00PM Room: Exhibit Hall
To build a Geant4-based Monte Carlo (MC) dose verification system for proton therapy with a scanning nozzle. This system will be used routinely at the upcoming Mayo Clinic proton center for both research and verification of pencil beam dose calculations.
Geant4 is a well-verified MC toolkit that is applied extensively in many branches of physics, including medical physics. We implemented our dose verification system using TOPAS, a proton MC package that facilitates the use of Geant4. The system consists of: scripts that read in data from DICOM files and create TOPAS input parameter files; TOPAS geometry extensions to accommodate our scanning nozzle, and TOPAS code additions to input 'parameterized' phase space data for beam spots at the nozzle exit. Typically, a phase space file containing source events for a grid of beam spots covering the scanning area at the nozzle exit is about 500 Gb. To reduce the size of this file down to ~30 Mb, we fit MC-generated beam spot position and direction distributions to extract Twiss parameters. The beam energy distributions are fitted with skewed Gaussians. The phase space is then represented by a set of skewed Gaussian and Twiss parameters for each beam spot in the grid.
Using our MC dose calculation system, we show MC and pencil beam comparisons for: (1) water phantom, (2) prostate and (3) head and neck cases. Good agreement is obtained. To reach 2% statistical error on the MC dose calculations, a tumor 200 cm³ in volume would require around 1x10⁷ protons. On a 100-node cluster, such a calculation takes ~1 hour to complete.
A TOPAS-based dose verification system for proton treatment plans has been built at Mayo Clinic. The typical computation time per plan is expected to be a few hours on a dedicated cluster.
Funding Support, Disclosures, and Conflict of Interest: This work is supported by a grant from Varian Medical Systems.
Add this talk to vcal | ical | Contact Email: