Simulation of four-dimensional effects in radiation therapy: the dose in intensity modulated beam delivery under organ motion

Abstract

Purpose/Objective: In radiation therapy there are geometrical uncertainties due to patient setup errors and internal organ motion or deformation, which can lead to differences between the planned dose distribution and the delivered dose distribution. This is true in particular with dose distribution showing sharp edges, like in IMRT and proton beam therapy. The traditional approach to these effects is to estimate the uncertainties and add margins around the clinical target volume, which ensure full tumor coverage. This method obviously increases the dose to the surrounding healthy tissue. A different approach is the implementation of the expected uncertainty into the dose calculation algorithm. For analytical dose calculation methods this leads to convolution methods (hereby organ deformation is usually disregarded). With the introduction of multi-leaf motion for conventional radiation therapy or magnetic beam scanning for charged particle therapy treatment has become dynamic therapy as far as the geometrical setup is concerned. To study the dosimetric effects of moving parts in the beam delivery systems or of a moving beam spot Monte Carlo simulations are a perfect while accurate tool. Each 4D, i.e. motion, problem can be simulated by combining a set of 3D simulations. There is also an increased interest in research concerning the dynamic behavior of the patient. Here in particular effects of setup uncertainties or organ motion. Again, to theoretically study these effects a set of 3D calculations can be performed adding up the results to answer questions of 4D behavior. The problem becomes more complicated for double-dynamic systems, i.e. cases where one is interested in the interplay of time-dependent beam delivery and moving target. To study the dosimetric behavior of a double-dynamic system true four-dimensional calculations are required. The problem with standard Monte Carlo routines is however, that input geometries, i.e. beam delivery system and patient geometry, cannot be changed easily during a Monte Carlo run. Our goal was the implementation of multi-leaf motion in IMRT and beam spot movement in proton beam scanning combined with organ movement via 4D CT information into a Monte Carlo dose calculation program.