| Summary: | The central task of radiotherapy is to deliver lethal radiation dose to diseased cells while sparing healthy tissue. To this end, much research in radiotherapy has focused either on improving our ability to deliver dose conformally, therefore improving the shape of dose distributions in the patient, or on improving our knowledge of the location of the tumor during radiotherapy, therefore allowing us to prescribe more conformal contours to treat and to improve our confidence in the treatment we deliver. Much of the research on improving patient positioning relies on imaging, thus motivating the field of image guided radiation therapy. In this dissertation, we develop a method for treatment planning in the context of real-time image guidance from kV x-ray sources. To that end, we describe a set of experiments to develop this MV+kV image guided planning optimization.
First, the theoretical basis for MV+kV optimization for treatment with real-time fluorscopy was developed, and a treatment planning system built for MV+kV optimization was programmed. The treatment plans from a set of 10 retrospective lung cancer patients were used to test the effect of MV+kV optimization on plan quality. It was found that by using MV+kV optimization as opposed to using MV optimization and performing non-optimized kV imaging, it was possible to reduce the imaging time by up to 55%. The MV+kV optimization program was then modified, first to consider the case of cone beam CT (CBCT) imaging (rather than real-time fluoroscopy), and then to use a more recent optimization technique of direct aperture optimization. With CBCT, MV+kV optimization did not provide as large a reduction in imaging dose as in the case of real-time fluoroscopy, though skin dose reductions on the order of 1-2% were still possible. However, preliminary work shows that combining MV+kV optimization with direct aperture optimization may provide further skin dose sparing.
The three experiments just described form the core of this dissertation, and were supported by a set of four secondary experiments. First, a measurement of the spectral qualities of the kV imaging beam was performed and compared to Monte Carlo derived spectra. It was found that the Monte Carlo spectra satisfactorily represented the measured spectra. Second, a method of imaging parameter optimization was developed in order to motivate the selection of imaging parameters used in the core of the dissertation. The method affirms the choice of imaging parameters for the specific task chosen, and provides a method of choosing imaging parameters for different tasks chosen in the future. Third, the dose delivered by the kV imaging beam was measured using thermoluminscent dosimeters (TLDs) in order to verify the dose calculations performed earlier. The TLD measurements confirmed that the MC dose calculation is accurate for well-modeled patients, but can struggle if there are errors in the modeling of the patient. Finally, a calibration method was developed for the alignment of the kV and MV imaging reference frames with the radiation therapy reference frame. This method was found to produce a calibration accurate to within .25 mm.
In summary, this work developed a method of MV+kV optimization, and the framework required for the optimization and the imaging to occur. The method of MV+kV optimization may allow for real-time imaging with improved skin sparing.
|
|---|
