<p>We investigated the use of multiplexing and an electro-optical coupling system in the design of magnetic resonance compatible positron emission tomography (PET) detectors. Reducing the number of output channels is an effective way to minimize cost and complexity and complements the substitution of coaxial cables for fiber optics. In this work, we first compared the system performance of two multiplexing schemes using both simulation and experimental studies. Simulations were performed using the LTSPICE environment to investigate differences in resulting flood histograms and rising edge slopes. Experiments were performed using Lutetium-Yttrium Oxyorthosilicate (LYSO) crystals of coupled to a SensL ArraySL-4 silicon photomultiplier (SiPM) connected to interchangeable circuit boards containing the two multiplexing schemes of interest. Three crystal configurations were tested: single crystal element (3x3x20 mm³), 2x2 array (crystal pitch: 3x3x20 mm³) and 6x6 array (crystal pitch: 2.1x2.1x20 mm³). Good agreement was found between the simulations and experiment results. The capacitive multiplexer is able to achieve improved time resolution of good uniformity (average of 1.11±0.01ns and 1.90±0.03ns for the arrays, respectively) and crystal separation, compared to the resistive multiplexing (average of 1.95±0.03ns and 3.33±0.10ns). The resistive multiplexing demonstrates slightly improved energy resolution (11±0.1% and 22±0.6%, compared to 12±0.1% and 24±0.4% for the capacitive array). The relevancy of this work to the PET block detector design using SiPM arrays is also discussed, including light sharing, edge compression and gain variation among SiPM pixels. This work also examines the effect of the electro-optical coupling system by comparing the system performance between cases with and without it. The coupling system is found to adversely affect performance, increasing global energy resolution by ~6%, average timing resolution by ~120% and distorting the flood histogram.</p> / Master of Applied Science (MASc)

Traditionally, medical imaging techniques such as PET (positron emission tomography) and SPECT (single-photon emission computed tomography) have relied on mechanical collimators to detect the sources of photons. This limits the image's resolution and field of view. To improve upon this, Compton cameras have emerged as a promising alternative. The principle is to measure the angle of a photon scattered in the detector, which indicates the likely sources in the form of a cone culminating in the position of the interaction. The cones from multiple events may be superimposed in order to generate an image. The object of this work is to use list-mode likelihood methods to better reconstruct the source image from the data recorded by a simulated Compton camera in the case of a solid detector volume with good spatial- and energy resolution.  The results demonstrate an improvement of image quality for reconstructions of single point source and multiple extended sources. In addition, the results indicate that our used algorithm converges for point sources. The minimum number of measured events for accurate reconstruction for different source distributions remains to be determined, and the algorithm's ability to resolve closely adjacent sources should be investigated more.