Development of an Efficient Algorithmic Framework for Deterministic Patient Dose Calculation in MRI-guided Radiotherapy

  • Author / Creator
    Yang, Ray
  • Accurate and efficient patient dose calculation for Magnetic Resonance Imaging (MRI)-guided Radiotherapy is pursued based on a deterministic solution to the Linear Boltzmann Transport Equation (LBTE) with magnetic fields, as a technique which does not suffer from statistical uncertainty, and presents an emerging alternative to Monte Carlo. In this framework, magnetic fields are modeled by an angular advection operator which introduces unique challenges and a new frontier to devise accurate and efficient solution techniques.

    Key innovations in this work include, (i) the development of conventions to discretize the 6-dimensional phase-space permitting harmonious interplay between space and angle, while retaining an acyclic space-angle discontinuous finite element solution sweep graph for all magnetic field orientations, (ii) the development of a novel angular advection upwind stabilization framework for curvilinear finite elements on the unit-sphere with flexibility energy adaptive forward-peaked angular meshing for parallel and perpendicular magnetic fields, and (iii) a novel runtime approach which ray-traces primary fluence using underlying continuous densities, while secondary scatter is reasonably approximated using a limited set of bulk material densities parameterized by k-means clustering. This enables an efficient transport sweep architecture leveraging batched multiplication by pre-inverted matrices and hierarchical batched assembly of the iterative scatter source.

    Through the development of these novel mathematical frameworks and algorithms, overall computational complexity is greatly reduced, with the flexibility to compute multi-beam treatment plans on patient anatomies in the presence of strong magnetic fields parallel or perpendicular to the radiation beam. At anatomical sites including lung, liver, and brain, over 99% (94%) of points pass a stringent 2%/2 mm (1%/1 mm) gamma criterion validated against GEANT4 reference Monte Carlo calculations in the presence of clinical magnetic field configurations. Runtimes of approximately 10 minutes per beam were achieved on a non-parallelized workstation implementation.

    The algorithmic building blocks and prototype code developed in this work demonstrate feasibility for highly accurate patient dose calculations in clinical magnetic field configurations, and serves as a robust launching point for further investigation towards realtime adaptive MRIgRT.

  • Subjects / Keywords
  • Graduation date
    Spring 2020
  • Type of Item
  • Degree
    Doctor of Philosophy
  • DOI
  • License
    Permission is hereby granted to the University of Alberta Libraries to reproduce single copies of this thesis and to lend or sell such copies for private, scholarly or scientific research purposes only. Where the thesis is converted to, or otherwise made available in digital form, the University of Alberta will advise potential users of the thesis of these terms. The author reserves all other publication and other rights in association with the copyright in the thesis and, except as herein before provided, neither the thesis nor any substantial portion thereof may be printed or otherwise reproduced in any material form whatsoever without the author's prior written permission.