Methods for dose calculation and beam characterization in external photon beam radiotherapy

Abstract

Treatment planning in external photon beam radiotherapy requires fast and accurate methods for the calculation of an absorbed dose distribution in a patient. Modern dose calculation algorithms also require characterization of the radiation beam produced by a medical linear accelerator. In this thesis, a dose calculation method based on the superposition of Monte Carlo simulated pencil-beam kernels was developed. The pencil-beam is divided into a depth-directed and a lateral component, which are separately scaled according to the radiological path-length information to account for tissue heterogeneities. The scatter along the plane is computed efficiently using incremental methods. In addition, a physics-based multiple-source model was developed in order to model the radiation beam. The free parameters of the model are derived using an automatic process, which minimizes deviations between the dose computations and the water-phantom measurements. The beam model was also incorporated with a Monte Carlo (MC) based dose calculation method. The accuracy of developed kernel-based dose calculation method and beam model were verified by performing comparisons to measurements in a wide range of conditions including irregular, asymmetric, wedged and IMRT fields. In heterogeneous phantoms containing lung and bone inserts, the accuracy was also investigated using MC simulations. The deviations between the dose calculations and measurements or MC simulations were within the clinical acceptability criteria in most of the studied cases with the exception of a high-energy beam with small dimensions in a low-density material. However, the obtained accuracy in the problematic cases was still significantly better than that of a currently widely used semi-empirical method. The dose calculations with the developed MC based system also agreed with water-phantom measurements within 2%/2 mm for open fields and within 3%/3 mm for wedged fields. Thus, the dose calculation algorithm and beam model developed in this thesis were found to be applicable for clinical treatment planning of megavoltage photon beams. In addition, it was demonstrated that the beam model can be successfully used as an input for other modern dose calculation methods, such as the Monte Carlo method.