Monte Carlo evaluation of small field dosimetry for low KV beams

Abstract

Background: Obtaining accurate measurements for absorbed dose calculations in water for kilovoltage x-rays is a very challenging task, especially at low energies and very small field sizes (≤ 3cm). This is due to the complex scatter conditions that exist, and that are mostly pronounced at the surface of the phantom. The uncertainties associated with the measurements for absorbed dose result in uncertainties in output factor calculations. Objectives: The only viable option to obtain this information is through Monte Carlo (MC) simulation. The main objective was therefore to produce a MC model for a superficial x-ray unit that would allow the simulated output information to be used in order to calculate output factors for very small field sizes. Research Design: This thesis presents the findings of the investigation into the calculation of percentage depth dose curves and beam profiles via Monte Carlo simulation of a superficial x-ray unit. The EGSnrc Monte Carlo simulation codes for the transport of electrons and photons through an artificial phantom representative of a water phantom were used. The D3225 from Gulmay Medical Ltd is simulated using the EGSnrc system of codes for 80kV photon fields. Percentage depth dose curves and beam profiles were compared against those measured from a PTW MP3 water phantom. The final simulated model is in turn utilised to calculate output factors for field sizes equal to and smaller than 3cm in diameter. Results: Good agreement was achieved between actual and simulated data, mostly in the depth dose curves where the most significant deviation registered was for the 10cm applicator with an average percentage error of 0.3%. Electron contamination and limited simulation histories caused noise to be induced across the beam profiles, thus yielding a maximum average percentage error of 2.2%. Recommendations: Benchmarking with more data sets, that is, performing the matching procedure for more energies and field sizes would further test the model’s accuracy. Also, increasing the number of histories would result in less uncertainty in the simulated data.