Factor analysis for quantitation of myocardial blood flow (MBF) using /sup 13/N-ammonia dynamic PET imaging

Abstract

Regional myocardial blood flow (MBF) can be measured by /sup 13/N-ammonia PET dynamic imaging using the conventional modeling approach that requires blood sampling, region-of-interest (ROI) drawing and a time-consuming nonlinear regression on each time activity curve (TAC). In this study, the factor analysis of dynamic structures (FADS) was used to extract the pure blood pool TAC and generate a parametric image of MBF (pixel unit: ml/min/g) which can map the myocardial perfusion accurately, Ten dynamic /sup 13/N-ammonia dog PET studies (3 baseline, 5 hyperemia, and 2 occlusion) were included. Three factors (TACs) and their corresponding factor images (the tight and left ventricular (RV and LV) blood pools and myocardial activities) were extracted from each study. The LV factors matched well with the plasma TACs, The factor image of myocardium was then converted to parametric image of MBF using a relationship derived from a two-compartment model, The results showed that the MBF obtained from PADS correlated well with MBF by two-compartment model fitting (correlation coefficient (r)t 0.98, slope=0.83) and by microsphere technique (r=0.98, slope=0.95). The FADS generated MIBF images have good image quality and lower noise levels compared to those generated by Patlak graphical analysis (PGA). It is concluded that regional myocardial blood flow can be measured accurately and noninvasively from /sup 13/N-ammonia dynamic PET imaging and FADS technique. FADS provides a simple method to map the distribution and magnitude of myocardial perfusion accurately and generate parametric image of MBF without requiring blood sampling and spillover correction. >