|Institution:||University of Victoria|
|Keywords:||3D dosimetry; polymer gel dosimeters; radiation dosimetry; optical CT; radiation therapy; cancer|
|Full text PDF:||http://hdl.handle.net/1828/5982|
New radiation therapy (RT) techniques for treating cancer are continually under development. Our ability to demonstrate the safe and accurate implementation of new RT treatment techniques is dependent on the information provided by current dosimetric tools. Advanced dosimetric tools will become increasingly necessary as treatments become more complex. This work examines the readout of an advanced dosimeter – the polyacrylamide, gelatin, and tetrakis (hydroxymethyl) phosphonium chloride (PAGAT) dosimeter – using a prototype fan-beam optical computed tomography (CT) scanner. A number of developments sought to improve the performance of the optical CT device. A new fan-creation method (laser diode module) and new matching tank were introduced. Artefact removal techniques were developed to remove flask seam artefacts and ring artefacts via sinogram space. A flask registration technique was established to achieve reproducible placement of flasks in the optical CT scanner. A timing-correction technique was implemented to allow for the scanning of continuously rotating samples. A number of experiments examined factors related to the PAGAT dosimeter. Comparisons of post-irradiation scans to pre-irradiation scans improved dosimeter readout quality. Changes to the PAGAT dosimeter cooling/scanning routine provided further improvements to dosimeter readout. Evaluations of calibration curves showed that a linear calibration curve was less capable of describing PAGAT dose response than a quadratic calibration curve. Intra-gel calibration using another dose distribution was shown to be no less accurate than self calibration, but inter-gel calibrations saw a statistically significant increase in absolute readout errors. A set of investigations examined how optical CT scanning protocols affected readout quality for PAGAT dosimeters. Doubling the dose delivered to the dosimeter doubled the signal-to-noise ratio. Acquiring and averaging additional light profiles at each projection angle provided only slight reductions in readout noise. Sampling a higher number of projection angles provided substantial reductions in readout noise. Those reductions in readout noise were not lost when sinograms with many projections were encapsulated into sinograms of fewer projection angles. Detector element binning (sinogram space) and pixel binning (image space) also provided substantial reductions in readout noise. None of these elements of the scanning protocol had statistically significant effects on readout errors. Finally, distinct imaging artefacts seen throughout this work were shown to be caused by radiation-induced refractive index changes in PAGAT dosimeters. Radiation-induced refraction (RIR) artefacts result when dose gradients caused the refraction of fan-beam raylines towards high dose regions. A filtering technique was developed to remove RIR artefacts in sinogram space, but this technique caused substantial blurring to the measured dose distribution.