Strategies for reducing noise level of parametric images in emission tomography

Abstract

High noise level of pixel-wise kinetic measurements in emission tomography requires spatial constraints to be applied to suppress the noise on parametric images. Since the relationship between the kinetic response and the model parameters of compartmental models commonly used to describe the tracer kinetics in emission tomography is nonlinear, application of spatial constraint at different steps in the parametric image generation is expected to give different results. In this study, we use computer simulation to examine the characteristics of various approaches and constraint choices in order to develop a proper processing strategy to reduce noise level in parametric images.