Date of Award
Doctor of Philosophy (PhD)
Physics & Astronomy
Joel S. Karp
The detector is the functional unit within a Positron Emission Tomography (PET) scanner, serving to convert the energy of radiation emitted from a patient into positional information, and as such contributes significantly to the performance of the scanner. While modern whole-body scanners use detectors composed of very many (i.e., 20000-30000) small pixels, typically ~4x4x20mm3 in size, several groups are actively investigating the performance of continuous crystals coupled to position sensitive photodetectors as an alternative detector design with a number of potential advantages, including improved spatial resolution and position sampling. This work in particular focuses on thick (≥14mm) continuous crystals in order to maintain the sensitivity of modern scanners. Excellent spatial resolution in continuous detectors that are thick, however, has proven difficult to achieve using simple positioning algorithms, leading to research in the field to improve performance. This thesis aims to investigate the effect of modifications to the scintillation light spread within the bulk of the scintillator to improve performance, focusing on the use of laser induced optical barriers (LIOBs) etched within thick continuous crystals, and furthermore aims to translate the effect on detector performance to scanner quantitation in patient studies.
The conventional continuous detector is first investigated by analyzing the various components of the detector as well as its limitations. It is seen that the performance of the detector is affected by a number of variables that either cannot be improved or may be improved only at the expense of greater complexity or computing time; these include the photodetector, the positioning algorithm, and Compton scatter in the detector. The performance of the detectors, however, is fundamentally determined by the light spread within the detector, and limited by the depth-dependence of the light spread and poor performance in the entrance region, motivating efforts to modify this aspect of the detector.
The feasibility and potential of LIOBs to fine-tune this light spread and improve these limitations is then studied using both experiments and simulations. The behavior of the LIOBs in response to optical light is investigated, and the opacity of the etchings is shown to be dependent on the parameters of the etching procedure. Thick crystals were also etched with LIOBs in their entrance region in a grid pattern in order to improve the resolution in the entrance region. Measurements show an overall improvement in spatial resolution: the resolution in the etched region of the crystals is slightly improved (e.g., ~0.8mm for a 25mm thick crystal), though in the unetched region, it is slightly degraded (e.g., ~0.4mm for a 25mm thick crystal). While the depth-dependence of the response of the crystal is decreased, the depth-of-interaction (DOI) performance is degraded as well. Simulation studies informed by these measurements show that the properties of the LIOBs strongly affect the performance of the crystal, and ultimately further illustrate that trade-offs in spatial resolution, position sampling, and DOI resolution are inherent in varying the light spread using LIOBs in this manner; these may be used as a guide for future experiments.
System Monte Carlo simulations were used to investigate the added benefit of improved detector spatial resolution and position sampling to the imaging performance of a whole-body scanner. These simulations compared the performance of scanners composed of conventional pixelated detectors to that of scanners using continuous crystals. Results showed that the improved performance (relative to that of 4-mm pixelated detectors) of continuous crystals with a 2-mm resolution, pertinent to both the etched 14mm thick crystal studied as well as potential designs with the etched 25mm thick crystal, increased the mean contrast recovery coefficient (CRC) of images by ~22% for 5.5mm spheres.
Last, a set of experiments aimed to test the correspondence between quantification in phantom and patient images using a lesion embedding methodology, so that any improvements determined using phantom studies may be understood clinically. The results show that the average CRC values for lesions embedded in the lung and liver agree well with those for lesions embedded in the phantom for all lesion sizes. In addition, the relative changes in CRC resulting from application of post-filters on the subject and phantom images are consistent within measurement uncertainty. This study shows that the improvements in CRC resulting from improved spatial resolution, measured using phantom studies in the simulations, are representative of improvements in quantitative accuracy in patient studies.
While unmodified thick continuous detectors hold promise for both improved image quality and quantitation in whole-body imaging, excellent performance requires intensive hardware and computational solutions. Laser induced optical barriers offer the ability to modify the light spread within the scintillator to improve the intrinsic performance of the detector: while measurements with crystals etched with relatively transmissive etchings show a slight improvement in resolution, simulations show that the LIOBs may be fine-tuned to result in improved performance using relatively simple positioning algorithms. For systems in which DOI information is less important, and transverse resolution and sensitivity are paramount, etching thick detectors with this design, fine-tuned to the particular thickness of the crystal and application, is an interesting alternative to the standard detector design.
Panetta, Joseph, "Development Of A Scintillation Detector And The Influence On Clinical Imaging" (2017). Publicly Accessible Penn Dissertations. 2512.