SUBJECT: Ph.D. Dissertation Defense
   
BY: Caleb Price
   
TIME: Thursday, April 2, 2015, 9:00 a.m.
   
PLACE: Boggs, 3-47
   
TITLE: A Coarse Mesh Transport Method with Novel Source Treatment
   
COMMITTEE: Dr. Farzad Rahnema, Co-Chair (NRE/MP)
Dr. Eric Elder, Co-Chair (NRE/MP)
Dr. Chris Wang (NRE/MP)
Dr. Timothy Fox (NRE/MP)
Dr. Thomas Morley (MATH)
 

SUMMARY

Treatment planning algorithms for use in the radiotherapeutic treatment of cancer have progressively evolved since the earliest attempts to develop automated dose calculation software in the mid-1950s. Modern algorithms use advanced techniques such as convolution superposition or grid-based Boltzmann solvers to perform external beam radiotherapy calculations. A new method of dose calculation was developed at the Georgia Institute of Technology based on transport theory called COMET-PE. The method combines stochastic pre-computation with a deterministic solver to achieve high accuracy and precision. For the COMET-PE method to be implemented clinically it needs a practical source model that closely mimics the physical characteristics of a typical radiation beam from a linear particle accelerator. The COMET-PE method should also be validated against a known benchmark. A novel linear accelerator source model is presented that models the geometry, angular distribution, spectrum, energy, and electron contamination of a 6 MV photon beam from a Varian c-series clinac. Of note is the use of a hemispherical harmonic expansion with the functional expansion tally method to model photonic fluence. The source was implemented in the COMET-PE radiation transport code and calculations performed with various field sizes and phantoms. The results are benchmarked against Monte Carlo reference solutions and compared with calculations performed with two popular commercially available treatment planning algorithms. The results indicate that the proposed source model when coupled with the COMET-PE method is capable of dosimetric calculations that in many cases more closely match Monte Carlo solutions than the commercially available options.