Date of Award

Winter 12-15-2017

Author's School

Graduate School of Arts and Sciences

Author's Department


Degree Name

Doctor of Philosophy (PhD)

Degree Type



Monte Carlo (MC) simulation is generally considered to be the most accurate method for dose calculation in radiation therapy. However, it suffers from the low simulation efficiency (hours to days) and complex configuration, which impede its applications in clinical studies. The recent rise of MRI-guided radiation platform (e.g. ViewRay’s MRIdian system) brings urgent need of fast MC algorithms because the introduced strong magnetic field may cause big errors to other algorithms. My dissertation focuses on resolving the conflict between accuracy and efficiency of MC simulations through 4 different approaches: (1) GPU parallel computation, (2) Transport mechanism simplification, (3) Variance reduction, (4) DVH constraint. Accordingly, we took several steps to thoroughly study the performance and accuracy influence of these methods. As a result, three Monte Carlo simulation packages named gPENELOPE, gDPMvr and gDVH were developed for subtle balance between performance and accuracy in different application scenarios. For example, the most accurate gPENELOPE is usually used as "golden standard" for radiation meter model, while the fastest gDVH is usually used for quick in-patient dose calculation, which significantly reduces the calculation time from 5 hours to 1.2 minutes (250 times faster) with only 1% error introduced. In addition, a cross-platform GUI integrating simulation kernels and 3D visualization was developed to make the toolkit more user-friendly. After the fast MC infrastructure was established, we successfully applied it to four radiotherapy scenarios: (1) Validate the vender provided Co60 radiation head model by comparing the dose calculated by gPENELOPE to experiment data; (2) Quantitatively study the effect of magnetic field to dose distribution and proposed a strategy to improve treatment planning efficiency; (3) Evaluate the accuracy of the build-in MC algorithm of MRIdian’s treatment planning system. (4) Perform quick quality assurance (QA) for the “online adaptive radiation therapy” that doesn’t permit enough time to perform experiment QA. Many other time-sensitive applications (e.g. motional dose accumulation) will also benefit a lot from our fast MC infrastructure.


English (en)

Chair and Committee


Committee Members

Zohar Nussinov, Harold Li, Anders Carlsson, Li Yang,


Permanent URL:

Included in

Physics Commons