In this study, factor analysis of dynamic structures (FADS) was used to extract the "pure" blood-pool time-activity curves (TACs) and to generate parametric myocardial blood flow (MBF) images (pixel unit: ml/min/g).
Methods: Ten dynamic 13N-ammonia dog PET studies (three baseline, five hyperemia and two occlusion) were included. Three factors (TACs) and their corresponding factor images (the right ventricular and left ventricular blood pools and myocardial activities) were extracted from each study. The left ventricular factors matched well with the plasma TACs. The factor images of myocardium were then converted to a parametric images of MBF using a relationship derived from a two-compartment model.
Results: MBF estimates obtained from FADS correlated well with MBF estimates obtained with the two-compartment model (r = 0.98, slope = 0.84) and microsphere techniques (r = 0.96, slope = 0.94). FADS-generated MBF parametric images have better image quality and lower noise levels compared to those generated with Patlak graphical analysis.
Conclusion: Regional MBF can be measured accurately and noninvasively with 13N-ammonia dynamic PET imaging and FADS. The method is simple, accurate and produces parametric images of MBF without requiring blood sampling and spillover correction.