Computational and visualization techniques for Monte Carlo based SPECT

Abstract

Nuclear medicine imaging systems produce clinical images that are inherently noisier and of lower resolution than images from such modalities as MRI or CT. One method for improving our understanding of the factors that contribute to SPECT image degradation is to perform complete photon-level simulations of the entire imaging environment. We have designed such a system for SPECT simulation and modelling (SimSPECT), and have been using the system in a number of experiments aimed at improving the collection and analysis of SPECT images in the clinical setting. Based on Monte Carlo techniques, Sim- SPECT realistically simulates the transport of photons through asymmetric, 3-D patient or phantom models, and allows photons to interact with a number of different types of collimators before being collected into synthetic SPECT images. We describe the design and use of SimSPECT, including the computational algorithms involved, and the data visualization and analysis methods employed.