Computational Analysis and Statistical Methods in Spatial Transcriptomics and Neuroimaging

Abstract

This dissertation explored the viability of using 3D position-sensitive CdZnTe (CZT) detectors for use in medical imaging, specifically for conducting simultaneous single-photon emission computed tomography (SPECT) and positron emission tomography (PET) multimodality imaging for better medical diagnoses. The CZT detectors used in this work operate at room temperature and have excellent position and energy resolutions of ≤ 500 µm and ≤ 1% FWHM at 662 keV, respectively, making CZT detectors suitable candidates for medical imaging. The two main areas of concern when utilizing CZT for medical imaging, specifically for coincidence imaging, are the limited count rate and timing resolution compared to currently used scintillation technologies. To examine the feasibility of using CZT for simultaneous PET and SPECT measurements, a “parallel plane” CZT detector system that could operate in coincidence was investigated. This system had the ability to place pinhole masks between the detectors and the source for single-photon emission imaging. The performance of the system was first characterized separately for coincidence imaging and single-photon emission imaging. When measuring only coincidence events, the system was able to resolve the 1.5 mm diameter rods of a Derenzo phantom and demonstrated a peak count rate for imageable coincidence pairs of 450 cps. When measuring only single-photon emission events, the resolution of the system was determined to be approximately 7 mm FWHM in the entire FOV. This limited resolution was due to the size of the 2 mm diameter pinholes in the mask. A new mask with pinholes that have a 0.79 mm diameter was designed that should improve this resolution to 2.6 mm. Following this, the performance of simultaneous coincidence imaging and single-photon emission imaging was investigated. This was completed by measuring a line source comprised of Na-22 and Co-57. The resolution from the reconstructed images matched the results observed from the individual measurements for both coincidence and single-photon emission events. The coincidence results from this simultaneous measurement had a decrease in efficiency of 60% with respect to the coincidence-only mode and the single-photon emission measurement had a decrease in efficiency of 90%. Successful small-animal simultaneous measurements were not completed in this work. The design of a pseudo-random mask-antimask coded aperture was completed as another section of this work. This mask was designed to replace redundant arrays that are commonly used in coded aperture measurements. The pseudo-random mask was designed to keep the desirable mask-antimask feature that redundant masks can have, while eliminating ghost artifacts that occur with redundant arrays when a source is in the partially-coded field of view (PCFOV). In simulation, with a detector-to-mask distance of 22 cm and mask-to-source distance of 38 cm, this random mask showed a resolution of 1.3 mm FWHM when imaging a 1 mm diameter point source in the fully-coded field of view (FCFOV). This was compared to the simulation results for a rank 101 modified uniformly redundant array (MURA), which had a resolution of 1.9 mm FWHM. Additionally, the ability for the random mask to be used to reconstruct sources in the PCFOV without ghost artifacts was demonstrated by simulation as well.