Master's Thesis Defense in Physics: Ryan Williams 3/20

Abstract:
This study proposes the utilization of positron annihilation photons, a by-product of the pair-production induced during MV X-ray radiotherapy treatment, to monitor radiation distribution in tissue for in-vivo dosimetry and patient anatomy for image-guided gating. The Monte Carlo code GATE was used to simulate a variety of phantom and detector geometries. First, a monoenergetic, isotropic positron source inside a large phantom was used to gather data on positron lifetime and range in various clinically relevant materials. Following this, a 6MV/15 MV LINAC photon beam was simulated to irradiate various phantom geometries. Two sets of detector geometries were used. First, two PET partial-arc arrays of LSO detectors were designed such that their central axis was at the LINAC isocenter, and the second was modeled after a realistic GE Discovery 690 PET-detector ring. Locations of positron annihilations within the phantom, singles and coincidences within the detector array, absorbed dose inside the phantom, and photon energy spectra for discrete detectors were scored. The photon energy spectra recorded at different detector angles highlight 511keV annihilation peaks with a significantly large signal-to-background ratio. Additionally, the Compton background for detector angles larger than 90° is negligible, which leads to more efficient filtering of annihilation photons via coincidence measurements. The annihilation peak for different materials scales both with effective atomic number and density, with different anatomical structures producing a different number of positrons/annihilation photons, leading to contrast between materials in a PIGRT reconstructed image. Finally, image reconstructions with a simplified lung geometry show the ability to track tumor movement during radiotherapy via PIGRT with a dose deposition of 0.6cGy per image.