Development of a fast Monte Carlo code for dose calculation in treatment planning and feasibility study of high contrast portal imaging.

摘要

A fast and accurate treatment planning system is essential for radiation therapy and Monte Carlo (MC) techniques produce the most accurate results for dose calculation in treatment planning. In this work, we developed a fast Monte Carlo code based on pre-calculated data (PMC, Pre-calculated Monte Carlo) for applications in radiation therapy treatment planning. The PMC code takes advantage of large available memory in current computer hardware for extensive generation of pre-calculated data. Primary tracks of electrons are generated in the middle of homogeneous materials (water, air, bone, lung) and with energies between 0.2 and 18 MeV using the EGSnrc code. Secondary electrons are not transported but their position, energy, charge and direction are saved and used as a primary particle. Based on medium type and incident electron energy, a track is selected from the pre-calculated set. The performance of the method is tested in various homogeneous and heterogeneous configurations and the results were generally within 2% compared to EGSnrc but with a 40-60 times speed improvement.;The PMC is also extended for proton transport in radiation therapy. The pre-calculated data is based on tracks of 1000 primary protons using general purpose MCNPX code. The proton energy range was 20, 40, ...100, 110, ...200 MeV with ECUT=200 keV. Protons produce many different secondary particles such as neutrons, deuterons, tritons, alphas, secondary protons, etc and they are handled in three categories: (1) Secondary protons: treated like a primary protons and transported using a track picked up from pre-calculated tracks; (2) Neutrons: The energy of the neutron are deposited far from the initial point and neglected. (3) All other secondaries: Since other secondaries have a very short range their energy is deposited locally. In comparison of the code with MCNPX as the reference the difference is generally between 2-4% and it runs 100 times faster than MCNPX.;Pre-calculated Monte Carlo codes are accurate, fast and physics-independent and therefore applicable to different radiation types including heavy-charged particles.;In another project, we worked on Monte Carlo feasibility study to use orthogonal bremsstrahlung beams for imaging in radiation therapy. The basic characteristics of orthogonal bremsstrahlung beams are studied and the feasibility of improved contrast imaging in linear accelerator with such a beam is evaluated. In the context of this work orthogonal bremsstrahlung beams represent the component of the bremsstrahlung distribution perpendicular to the electron beam impinging on an accelerator target. In this set up the bending magnet of the linac is turned off and the primary electron beam directly hits a target from the side and the orthogonal beam in downward direction is used for imaging purposes. Monte Carlo modeling (BEAM code) is used to design the shape of different targets and to obtain the energy spectrum and the relative intensity of the orthogonal beams. After optimizing the shape of the target, two different target and a collimator was designed and built. The CLINAC 18 in Montreal General Hospital was used for the experiments. The simple lucite objects one of which with 1 cm steps was placed in the way of the orthogonal beams to verify the contrast. The simulations showed that in the orthogonal direction 80% of the CSDA range is enough to stop all of the scattered electrons. The intensity of the orthogonal beam for high-z targets is larger compared to low Z targets i.e., by a factor 20 for W/Be. The energy spectrum of the photon spectrum for low-z targets energy is lower (330 KeV for Al and 170 keV for Be) compare to higher z targets (900 KeV for Pb). In the experimental setup as well as Monte Carlo simulation it was illustrated that the contrast of the images created with the orthogonal beam is better than that of the forward beam.;The limitations of various techniques for the improvement of speed and accuracy of particle transport have been evaluated. We studied the obstacles for further increased speed ups in voxel based geometries by including ray-tracing and particle fluence information in the pre-generated track information. The latter method leads to speed-increases of about a factor of 500 over EGSnrc for voxel-based geometries. In both approaches, no physical calculation is carried out during the runtime phase after the pre-generated data has been stored even in the presence of heterogeneities. The pre-calculated data is generated for each particular material and this improves the performance of the pre-calculated Monte Carlo code both in terms of accuracy and speed.