Parallelised Monte Carlo simulation strongly refines radiotherapy treatment planning
Barcelona 30 November 1999The Monte Carlo simulation constitutes a very efficient technique to describe the radiation transport in complex physical systems, such as the human body. The Department of Physics at the University of Barcelona has designed the Penelope programme to implement the Monte Carlo simulation in the clinical area of radiotherapy. Together with a team of researchers from the Hospital Clinic, the Polytechnic University of Catalonia, and the National University of Cordoba in Argentina, scientists of the UB Physical Department have ported the Penelope programme to a parallel system architecture, as to accelerate the Monte Carlo simulation.
Advertisement
Penelope stands for Penetration and Energy Loss of Positrons and Electrons. The programme incorporates a mixed simulation algorithm for the interaction of charged particles, which is far more accurate and flexible than the alternative existing applications, like ITS or EGS4. The adopted sections of effective interaction are based on physical models with a general validation for materials of arbitrary composition and energy radiations, which oscillate between a few keV and hundreds of MeV. Penelope also includes a package of geometrical subroutines, that allows to perform the simulation in systems with a complex geometry.
Although the Monte Carlo simulation is the obvious technique to resolve all problems related to transport of matter, the simulation results as well as the experimental measurements always are affected by statistical uncertainties. In order to reduce the amount of those uncertainties to an acceptable, very small number, it is necessary to increase the calculation time but this limits the practical application of the method. In this regard, the use of fast parallel machines enables the simulation of many problems which were inaccessible in the past. The parallelisation also allows to incorporate interactive models which are far more accurate.
An important part of the Penelope parallelisation project involves simulation of radiation doses for radiotherapy and adequate dose estimation. To date, already two electron accelerators have been simulated, namely a Saturn 43 for photon generation and a Siemens KDS. Also other radiation sources have been studied, such as external radiotherapy, referred to as brachytherapy. The common characteristic of all these problems is their dependency on a relatively small number of variables. This allows to run the simulation in a way to possibly have the values of the variables coincide.
Currently, there exists a considerable interest in utilising the Monte Carlo simulation as a basis for radiotherapy planning in order to proportion the information for dose estimation in real time. In the future, this will turn out to be an alternative, superior to conventional planning, which is based on analytical approximations with only limited medical validation. During the treatment planning phase, the simulation allows to consider each patient's specific geometry, described through axial computed tomography (ACT). The objective of this part in the Penelope project is to develop a calculation code, yielding accurate results.
In the end, the research team aims to apply these results in their capacity of reference values, serving as simplified codes which finally are incorporated in the actual planning. In order to generate the reference code, the structure of the simulation algorithm needs to be altered to minimise the total volume of characteristic data within the material, and to optimise the geometrical parts of the algorithm, as to accelerate the simulation. Thus, the parallelisation of Penelope provides the computational power to perform the simulation which enables to accurately calculate the exact dose map, corresponding to the ACTbased geometry of each individual patient.
As the news source for this article, VMW Magazine refers to the Penelope project contribution in the November 1999 issue of the Catalonian supercomputer magazine Teraflop.
Leslie Versweyveld
[Medical IT News][Calendar][Virtual Medical Worlds Community][News on Advanced IT]
