Background and purpose: Although proton therapy is becoming increasingly common as a radiotherapy modality, facilities offering proton therapy are still scarce in comparison to photon therapy. Sweden's new proton therapy facility, Skandionkliniken, is scheduled to being operation during August 2015, employing the pencil beam scanning technique. Given Skandionklinikens unique stance as the only facility offering proton therapy in Sweden as of this writing, it is important to minimize the need for measurements during quality assurance to free up beam time for patients and other endeavors. It is the purpose of this work to create a foundation for a method whereby dose distribution verification is done via Monte Carlo simulation by developing and performing simple validation of a beam model. As input for simulating a dose distribution, log files storing a wide variety of data on how the dose distribution was delivered were used. Method: GATE, an open source Monte Carlo code and built on top of Geant4, was used for all simulations. A beam model parameterizing phase space at the nozzle exit was developed. The beam model development process made use of the beam data library and log file data. Using an in house developed code to convert log file data to treatment plans readable by GATE allowed simulation of delivered dose distributions. For validation, gamma index tests were performed comparing measured and simulated dose distributions. Results: The beam model was found able to predict the spot size in almost all cases within 0.2 mm. Likewise, the beam model was able to predict the proton range within 0.2 mm. The energy spread was found to be more difficult to estimate; comparisons of simulated and measured curves for at six points around the Bragg peak yielded a maximum deviation of 0.86 mm. Several difficulties prevented easy interpretation of the results of the gamma index tests. If allowance is made for certain data manipulation, pass rates of 90% or above using the global method can be achieved for all depths and for both treatment plans scanned. Conclusion: Although some complications arose during validation, the beam model performance appears capable of producing accurate results. To produce a full product suitable for routine patient specific quality assurance, further work will be necessary. Significant computing power would also be mandatory for routine use, necessitating the acquisition of a dedicated computer cluster or using GPUs.
| Identifer | oai:union.ndltd.org:UPSALLA1/oai:DiVA.org:su-123677 |
| Date | January 2015 |
| Creators | Almhagen, Erik |
| Publisher | Stockholms universitet, Fysikum |
| Source Sets | DiVA Archive at Upsalla University |
| Language | English |
| Detected Language | English |
| Type | Student thesis, info:eu-repo/semantics/bachelorThesis, text |
| Format | application/pdf |
| Rights | info:eu-repo/semantics/openAccess |
Page generated in 0.0228 seconds