An imaging performance standard for MBF quantification with 3D PET

Objectives Quantification of myocardial blood flow (MBF) is used increasingly in clinical research and management of CAD patients. New 3D list-mode PET scanners can reconstruct dynamic images for MBF together with standard ECG-gated perfusion images. High-count-rate correction accuracy is critical during the injected bolus first-pass, but this not adequately assessed by current imaging standards. We propose a method to characterize 3D PET systems for quantitative MBF imaging.

Methods A Data Spectrum anthropomorphic torso phantom is used with liver and myocardium compartments filled with equal concentration of N-13 or C-11 solution (50 mCi total), and scanned over 8 half-lives in 40 equal time frames. Dynamic images are reconstructed with all corrections and minimal smoothing. Time-activity curves are generated for myocardial wall, cavity and liver compartments to measure reconstructed correction accuracy over a wide dynamic range. Bias is assessed as a function of activity and dead-time factors for 2 state-of-the-art 3D PET systems.

Results On the Discovery 690(&RX) PET/CT system, the bias in reconstructed activity was less than 15% with 25(&20) mCi in the field of view, and a corresponding dead-time correction factor of 1.7(&1.7). Myocardium wall-to-liver ratios varied by less than 2%(&4%) indicating minimal loss of resolution due to high-count-rate pile-up effects. Myocardium cavity-to-liver ratios varied by less than 6%(&3%) indicating low and stable residual scatter.

Conclusions Dynamic range was assessed for two 3D PET systems to determine suitability for quantitative MBF imaging. On both systems, activity should be administered to limit the dead-time to <40% for accurate measurement of the bolus first-pass data. This corresponds to ~12(&10) MBq/kg on the 690(&RX) scanner