Clinical Dosimetry in Photon Radiotherapy – a Monte Carlo Based Investigation

Practical clinical dosimetry is a fundamental step within the radiation therapy process and aims at quantifying the absorbed radiation dose within a 1-2% uncertainty. To achieve this level of accuracy, corrections are needed for calibrated and air-filled ionization chambers, which are used for dose measurement. The procedures of correction are based on cavity theory of Spencer-Attix and are defined in current dosimetry protocols. Energy dependent corrections for deviations from calibration beams account for changed ionization chamber response in the treatment beam. The corrections applied are usually based on semi-analytical models or measurements and are generally hard to determine due to their magnitude of only a few percents or even less. Furthermore the corrections are defined for fixed geometrical reference-conditions and do not apply to non-reference conditions in modern radiotherapy applications. The stochastic Monte Carlo method for the simulation of radiation transport is becoming a valuable tool in the field of Medical Physics. As a suitable tool for calculation of these corrections with high accuracy the simulations enable the investigation of ionization chambers under various conditions. The aim of this work is the consistent investigation of ionization chamber dosimetry in photon radiation therapy with the use of Monte Carlo methods. Nowadays Monte Carlo systems exist, which enable the accurate calculation of ionization chamber response in principle. Still, their bare use for studies of this type is limited due to the long calculation times needed for a meaningful result with a small statistical uncertainty, inherent to every result of a Monte Carlo simulation. Besides heavy use of computer hardware, techniques methods of variance reduction to reduce the needed calculation time can be applied. Methods for increasing the efficiency in the results of simulation were developed and incorporated in a modern and established Monte Carlo simulation environment. The efficiency of ionization chamber calculations could be improved by several orders of magnitude. Using the developed methods, current clinical dosimetry protocols for the determination of absorbed dose to water under reference conditions in photon beams were reviewed. Calculations of correction factors were performed and compared to the currently existing data. It could be shown that the calculated values are in agreement with recent data, mainly based on calorimetric measurements, but partially deviate from currently used data in dosimetry protocols by _1%. Reason for these discrepancies are outdated theories and measurments for the single underlying perturbations. Sources of uncertainties in the calculated results based on Monte Carlo simulations were investigated, also considering uncertainties in underlying cross sections as input for these calculations. It could be shown that following a conservative estimation, systematic uncertainties of ~1% might be adherent to the calculated results, a fact that is barely considered in recent works. Ion chambers under non-reference conditions were investigated with the use of a virtual model of a clinical linear accelerator. Besides developing a procedure for commissioning the model i.e. adapting it to measurements with respect to primary electron characteristics, these calculations aimed at answering the question how ionization chambers behave in non-reference geometrical conditions. It turned out that commonly used ionization chambers show only small changes in response under non-reference conditions when fulfilling the condition of charged particle equilibrium. In contrast, whenever charged particle disequilibrium and high dose gradients exists, i.e. in the penumbra of a small radiation field, a strong change in detector response might occur. The applicability of the Spencer-Attix theory under these severe conditions was tested. It could be shown that, within a 1% uncertainty, the application of the Spencer-Attix theory with corresponding perturbation factors is valid. A further investigation of these conditions when measuring dose profiles was used to determine the type of detector with minimal change in response for regions of charged particle dis-equilibrium and high dose gradients. In terms of penumbra broadening, radiochromic film shows the smallest deviation from dose to water. Monte Carlo simulations will replace or at least extend the existing data in clinical dosimetry protocols in order to reduce the uncertainty in radiotherapy. For corrections under non-reference conditions as occuring in modern radiotherapy techniques, Monte Carlo calculations will be a crucial part. This work and the developed methods accordingly form an important step towards reduced uncertainties in radiotherapy for cancer treatment.