Abstract
Radiation therapy techniques were considerably enhanced in the last two decades and became more sophisticated, which calls for improved treatment verification and quality assurance methods. Many researchers have investigated alternative dosimetry solutions to answer the need for complex radiotherapy treatment verifications. Electronic portal imaging devices are among the plausible solutions. There are numerous published articles discussing and characterising EPID dosimetry in the literature. However, published results are not consistent, in particular when describing EPID dose linearity, and it is often difficult to compare the results from different publications due to a lack of common terminology to describe the EPID response. Many EPID dosimetry studies have focused on the so-called image lag and ghosting effects, and yet it is not always possible to relate different studies due to the inconsistent definitions of image lag and ghosting used in these studies. In addition, EPID performance characteristics are often underreported. In this dissertation, which characterises EPID response, I propose a novel dosimetry system for an amorphous-Silicon based EPID signal that fully incorporates the radiation beam characteristics and EPID parameters. Two Elekta iViewGT devices were used to study the EPID performance and response to radiation. Both devices are equipped with Perkin Elmer amorphous silicon flat panel detectors and attached to an Elekta Synergy linear accelerator. EPID response was assessed over the range of 1 - 500 MU with 6 MV X-ray beams. The field size of 10 x 10 cm2 was used to quantify the EPID signal with different radiation doses, dose rates and integration times. The effect of field size on the EPID signal was assessed using all possible rectangular fields equivalent to square fields between 1 and 25 cm2. Transient dosimetry measurements were carried out using solid water phantom, RMI457. Monte Carlo simulations were used to study the 6 MV beam energy spectrum variations due to phantom attenuation using the EGSnrc code. In addition, a MatLab computer model for the iViewGT was designed to study the effect of the readout technique on the EPID signal. Experimental results show that the readout technique has a clear effect on the signal profile, with two distinct signal profiles observed depending on the irradiation time and integration time. With short irradiations the EPID signal profile has a Gaussian shape. Long irradiations however result in readout equilibrium and have a different signal profile. EPID performance is influenced by electronic gain and readout technique as well as delivered dose, dose rate, integration time and field size. The EPID signal was found to be linear with the delivered radiation dose independent of the dose rate and integration time. However, frame signal during equilibrium is proportional to dose rate and integration time independent of the delivered radiation dose. It is not feasible to measure the residual signal accurately because of the readout technique. The residual signal is the most considerable result of the image lag and ghosting effects, and is most significant at small radiation doses (<50 MU), with a maximum value of less than 2. 5% of the total integrated signal. EPID signal could be interpreted to reflect the delivered radiation dose or to measure phantom physical dimensions. The EPID has a uniform response to the increased phantom thickness in both measurement and Monte Carlo simulation and was not influenced by the energy spectrum of the incident radiation. Field size has a detectable effect on the EPID measurements and the field size relative output factors follow a uniform pattern as a function of the field size. The EPID response depends on the radiation field size yet is independent of the field shape. This dissertation provides a frame work for EPID dosimetry that accommodates both radiation beam and EPID characteristics including integration time and frame readout technique.