Parametric Imaging of Myocardial Blood Flow with 82Rb PET: An Accuracy and Image Quality Analysis

We aimed to develop a framework for generating 3D myocardial blood flow (MBF) images, computing their accuracy against clinically validated 2D polar MBF maps of the left ventricle, and evaluating their improvements in image quality over relative myocardial perfusion imaging (MPI). N=40 patients with a wide range of defect severities and uptake dynamics were retrospectively studied. The FlowQuantTM software was used to generate reference MPI and polar MBF maps and was adapted for voxel-wise MBF mapping. We evaluated agreement between parametric vs. polar values for MBF at rest and stress and for reserve (stress/rest MBF). We also assessed improvements in image quality, assessed by signal-to-noise ratio, contrast-to-noise ratio, tissue-to-blood ratio, and defect severity, from relative MPI to MBF. There was excellent agreement between 3D parametric and 2D polar-maps for all flow parameters (ICC>0.96) albeit with minimal bias (<8%) for rest and stress MBF at the patient level. Image quality substantially improved from MPI to MBF in every patient for all image quality metrics (p<0.0001) We developed a robust methodology for producing highly accurate 3D MBF images exhibiting considerably improved image quality compared to relative MPI commonly used in clinical practice.