Quantitative myocardial perfusion PET parametric imaging at the voxel-level.

摘要: Quantitative myocardial perfusion (MP) PET has the potential to enhance detection of early stages atherosclerosis or microvascular dysfunction, characterization flow-limiting effects coronary artery disease (CAD), and identification balanced reduction flow due multivessel stenosis. We aim enable quantitative MP-PET at individual voxel level, which allow enhanced visualization quantification blood (MBF) reserve (MFR) as computed from uptake parametric images. This framework is especially challenging for (82)Rb radiotracer. The short half-life enables fast serial imaging high patient throughput; yet, acquired dynamic images suffer noise-levels introducing large variability in and, therefore, estimates MBF MFR. Robust estimation requires substantial post-smoothing noisy data, degrading valuable functional information physiological pathological importance. present a feasible robust approach generate voxel-level that substantially reduces noise without significant loss spatial resolution. proposed methodology, denoted clustering, makes use similarity voxels penalize deviation kinetics partners. results were validated using extensive simulations (with transmural non-transmural defects) clinical studies. Compared post-smoothing, clustering depicted versus bias performance well superior recovery defects (as quantified by CNR) with minimal increase bias. Overall, obtained methodology presence high-noise levels manifested time-activity-curves.