Improved PET-Based Voxel-Resolution Myocardial Blood Flow Analysis

In this paper, we describe a new method for estimating the myocardial blood flow (MBF) of each voxel in a myocardium using data consisting of a dynamic sequence of cardiac positron emission tomography images. First, we incorporate the factor analysis of dynamic sequences model into a certain two-compartment model with physiological constraints. Then, using the majorize-minimize optimization technique, we develop an algorithm, which we call the improved voxel-resolution MBF (IV-MBF) algorithm, that estimates the parameters of the resulting model, including the MBF parameters, by iteratively minimizing a certain least-squares objective function. We perform a preliminary evaluation of the IV-MBF algorithm using real patient data.