Development and evaluation of Monte Carlo-based SPECT reconstruction

Single Photon Emission Computed Tomography (SPECT) is one of the most applied molecular imaging techniques to diagnose human diseases, e.g., of the heart, the brain or in oncology. For example, cardiac SPECT imaging plays a central role in diagnosing coronary heart diseases by providing clinicians with vital information of myocardial perfusion, viability, and functionality. The quality of reconstructed SPECT images is degraded by a number of factors, such as noise, photon attenuation, collimator and detector blurring, and the interference of scatter photons. These degradations can have a large impact on quantitative accuracy and on the outcome of clinical diagnosis. An accurate way to correct for attenuation, camera blurring and scatter is to employ a physical model of the formation of the projections during iterative SPECT reconstruction. Monte Carlo simulation is a general and accurate method to model the photon transport phenomena, and thereafter corrects for degradations, but has the disadvantage of being computationally demanding. In Chapter 2, we used convolution forced detection to accelerate reconstruction by two orders of magnitude. In Chapter 3, we compared statistical reconstruction based on Monte Carlo modeling of scatter with the more commonly used triple-energy-window scatter correction. Our results indicate that the imaging performance of Tc-99m SPECT can be improved more by Monte Carlo based scatter correction than by window-based scatter correction. In Chapter 4, our goal was to improve and evaluate the effectiveness of 3D Monte Carlo–based scatter correction in Tl-201 imaging. Compared to the other algorithms, statistical reconstruction based on Monte Carlo modeling of scatter had better lesion contrast, were less sensitive to anatomical variations and had better image uniformity in the homogeneously perfused myocardium. In Chapter 5, we investigate to what extent truncation impacts attenuation correction and model-based scatter correction in the cases of Tc-99m, Tl-201, and simultaneous Tc-99m/Tl-201 studies. In addition, we evaluate a simple correction method to mitigate effects of truncation. Our results indicate that, for single isotope studies, using small FOV systems has little impact on attenuation correction and model-based scatter correction. For simultaneous Tc-99m/Tl-201 studies, artificial projection extension almost fully eliminates adverse effects of projection truncation. In Chapter 6, we developed and validated a method for quantitative SPECT of Ho-166 that involves correction for several types of scatter. The combined attenuation, photopeak-scatter and down-scatter correction framework proposed here, greatly enhanced the quantitative accuracy of Ho-166 imaging, which is of the utmost importance for image guided therapies