Background: Oxygen 15 water and nitrogen 13 ammonia are widely used for the quantitative measurement of myocardial perfusion with positron emission tomography. However, blood flow obtained with N-13 ammonia by use of the conventional 2-compartment model frequently underestimates flow by 30% to 50% compared with O-15 water. We hypothesized that this discrepancy is a result of the model configuration of N-13 ammonia and investigated changes to the mathematical model to determine whether more accurate measurements of perfusion could be obtained.
Methods and results: Twelve healthy volunteers were sequentially studied with O-15 water and N-13 ammonia at rest and during maximal coronary vasodilation with adenosine. Perfusion measurements obtained with the conventional and modified models were compared with values obtained with O-15 water. The conventional N-13 ammonia model underestimated flow by 37% +/- 16% at rest and by 20% +/- 24% with stress when compared with flows obtained with O-15 water. The modified model yielded flow values closer to the line of identity than the conventional model (y = 1.07x + 0.04 vs y = 0.69x + 0.08; respectively; P < .01).
Conclusions: Model changes made N-13 ammonia myocardial blood flow estimates more comparable to those obtained with O-15 and may allow for better comparison of flows obtained with these two tracers in the future. Further efforts are warranted to evaluate the accuracy of flow models in human subjects.