Image Quality, Modeling, and Design for High-Performance Cone-Beam CT of the Head
MetadataShow full item record
Diagnosis and treatment of neurological and otolaryngological diseases rely heavily on visualization of fine, subtle anatomical structures in the head. In particular, high-quality head imaging at the point of care mitigates patient risk associated with transport and decreases time to diagnosis for time-sensitive diseases. Cone-beam computed tomography (CBCT) systems have found widespread adoption in diagnostic and image-guided procedures. Such systems exhibit potential for adaptation as point-of-care systems due to relatively low cost, mechanical simplicity, and inherently high spatial resolution, but are generally challenged by low contrast imaging tasks (e.g., visualization of tumors or hemorrhages). This thesis details the development and design of a CBCT imaging system with performance sufficient for high-quality imaging of the head and suitable to deployment at the point of care. The performance of a commercially available head-and-neck CBCT scanner was assessed to determine the potential of such systems for high-quality head imaging. Results indicated low-contrast visualization was challenged by high detector noise and scatter. Photon counting x-ray detectors (PCDs) were identified as a potential technology that could improve the low-contrast visualization, and an imaging performance model was developed to quantify their imaging performance. The model revealed important implications for energy resolution, noise, and spatial resolution as a function of energy threshold and charge sharing rejection. A new CBCT system dedicated to detection of low-contrast contrast intracranial hemorrhage was designed with guidance from an imaging chain model to optimize the system configuration (geometry, detector, x-ray source, etc.). The results indicated flat panel detectors (FPDs) were favorable due to a large field of view, but benefited from detector readout gain adjustments. Dual-gain detector readout was compared with use of bowtie filter in high-gain readout mode to investigate potential improvements to noise performance in FPDs. Finally, technical assessment of the prototype CBCT head scanner (with design based on guidance from the image quality model) indicated performance suitable for translation to clinical studies in the neurosciences critical care unit.