Monte Carlo modelling and optimisation of flattening filters for a novel Co-60 total body irradiation unit

Abstract

Total body irradiation (TBI) is used in the treatment of certain types of blood cancers. It involves irradiating the whole body of the patient with high energy x-rays or gamma rays in order to kill cancer cells and suppress the immune system in preparation for a stem cell or bone marrow transplant. The challenge in TBI is delivering a uniform dose throughout the whole patient. Most conventional cancer treatment units are not able to provide a sufficiently large uniform radiation field to enable TBI without complex movement of the beam or the patient. Recently, a dedicated Co-60 TBI unit was developed in Canada by BEST Theratronics (Kanata, ON). It uses gamma radiation from a Co-60 source housed in an extended gantry to irradiate a stationary patient lying on a couch below the source. The radiation beam of the source needs to be modified by a filter between the source and the patient to ensure uniform irradiation over the whole patient length as the distance from source varies considerably over the treatment volume. The amount of filtration required may also change over time to ensure an appropriate dose rate is maintained as the source decays. To assist BEST Theratronics with filter design, Monte Carlo (MC) simulations for the TBI unit were run to test the effectiveness of filter shapes that they had calculated by hand. This was a time consuming, trial-and-error process which required specialised training. It was the motivation for the development of a filter optimisation program which uses a radiation transport model based on primary and first scatter dose contributions that is quicker and easy to use. It was found that the relative dose distributions calculated by this program matched the MC-calculated distributions within about 3%. The absolute dose calculated was accurate for a small field and phantom, but for the TBI unit and a large phantom, the absolute dose was underestimated by 24-38% because of the missing higher order scatter. However, the amount of underestimation is predictable, so it is possible to use the program to calculate acceptable beam flattening filter shapes for a particular desired dose rate.