Variance and covariance of 82Rb kinetic parameters: computer simulations and dynamic PET studies

Cardiac ⁸²Rb PET can quantify absolute myocardial perfusion through tracer kinetic modeling. The aim of this study was to evaluate performance of the one-tissue-compartment model to estimate quantitative perfusion values with the tracer ⁸²Rb and dynamic PET. Myocardial tissue curves were simulated for a range of values for the model parameters k1 (0.1-4.0 ml/min/g), k2 (0.0-1.0/min), and blood volume Fa (0.1-0.9) with 20% Gaussian noise. Bias, variance and correlation of the parameters were investigated as measures of quality assurance for the quantitative perfusion parameter k1. PET studies were performed in dogs with and without occlusion of the LAD coronary artery to validate the simulations. The parameters k1, k2 and Fa were all estimated to within 10% of the true values when Fa was within 0.1-0.7. k1 estimates were unstable for Fa values larger than 0.9, which can occur in regions of severely diseased myocardium. Correlation of k1 with Fa increases with large values of k1, but does not introduce significant bias. Estimated k1 values and correlations ρ(k1,Fa) from the dog studies were in good agreement with the simulation values, indicating that the simulations were representative of measured ⁸²Rb PET data. Changes in image smoothness were compensated by Fa as expected, and did not bias k1 estimates in the range employed for typical cardiac ⁸²Rb PET studies. A one-tissue-compartment model of ⁸²Rb tracer kinetics generates accurate and precise estimates of the perfusion parameter k1. The k1 estimates are not sensitive to image smoothness or the correlation with blood volume