| The aims of this dissertation were to develop a method with improved accuracy for various heterogeneous geometries and to evaluate the feasibility of using EPID transit dosimetry for error detection. This dissertation examined phantom geometries with an emphasis on a tumor-in-lung geometry that may occur in hypo-fractionated SBRT treatments. For dose investigation in heterogeneous geometries, a multi-planar film measurement system was used with a measurement accuracy (within 3%) by using specific procedures to reduce the film perturbation in a low density medium. Measurements were used to validate a Monte Carlo (MC) method, and the results indicated that this MC method can be used as a reference to validate other calculation algorithms or to evaluate the doses delivered to patients for lung treatment. For developing an error detection method, a commercial electronic portal imaging device (EPID) composed of amorphous silicon was characterized for dosimetry application. A general calibration method was explored to use this device as a water-equivalent dosimeter, allowing for direct comparison to calculated doses from treatment planning systems. The calibration method was validated for a range of situations, field shapes, and intensities. The EPID transit dosimetry was sensitive to delivery errors, such as variations in treatment field shape, machine output and patient setup. The correlation between in-patient and transit dose variations may be established and used to determine acceptance or rejection criteria when the error is found. This dissertation showed the potential of using EPID dosimetry during treatment for on-line error correction and for estimating the in-patient dose error. |
