Noninvasive Bi-graphical Analysis for Parametric Imaging of Slowly Reversible Neuroreceptor Binding with Dynamic Brain PET

Abstract

Tracer kinetic modeling in dynamic positron emission tomography (PET) has been widely used to investigate characteristic distribution pattern or dysfunction of neuroreceptors in brain diseases, by offering a unique tool for generating images of quantitative parameters (or parametric imaging) of neuroreceptor binding. Graphical analysis (GA) is a major technique of parametric imaging, and is based on a simple linear regression model that is linearized and further simplified from a more complex general compartment model. Although each simple model of various GA methods enables very desirable parametric imaging, it depends on several assumptions that are commonly hard to satisfy simultaneously in parametric imaging for slow kinetic tracers, leading to error in parameter estimates. A combination of two GA methods, a bi-graphical analysis, may improve such intrinsic limitation of GA approaches by taking full advantage of spatiotemporal information captured in dynamic PET data and diverse strengths of individual GA methods. This thesis focuses on a bi-graphical analysis for parametric imaging of reversible neuroreceptor binding. Firstly, I provide an overview of GA-based parametric image generation with dynamic neuroreceptor PET data. The associated basic concepts in tracer kinetic modeling are presented, including commonly used compartment models and major parameters of interest. Then, technical details of GA approaches for reversible and irreversible radioligands are described considering both arterial-plasma-input-based (invasive) and reference-region-input-based (noninvasive) models

their underlying assumptions and statistical properties are described in view of parametric imaging. Next, I present a novel noninvasive bi-graphical analysis for the quantification of a reversible radiotracer binding that may be too slow to reach relative equilibrium (RE) state during PET scans. The proposed method indirectly implements the conventional noninvasive Logan plot, through arithmetic combination of the parameters of two other noninvasive GA methods and the apparent tissue-to-plasma efflux rate constant for the reference region (k_2^'). I investigate its validity and statistical properties, by performing a simulation study with various noise levels and k_2^' values, and also evaluate its feasibility for [18F]FP-CIT PET in human brain. The results reveal that the proposed approach provides a binding-parameter estimation comparable to the Logan plot at low noise levels while improving underestimation caused by non-RE state differently depending on k_2^'. Furthermore, the proposed method is able to avoid noise-induced bias of the Logan plot at high noise levels, and the variability of its results is less dependent on k_2^' than the Logan plot. In sum, this approach, without issues related to arterial blood sampling if a pre-estimated k_2^' is given, could be useful in parametric image generation for slow kinetic tracers staying in a non-RE state within a PET scan.