3D list-mode cardiac PET for simultaneous quantification of myocardial blood flow and ventricular function

Current PET instrumentation and performance metrics have been developed primarily to optimize whole-body imaging with
¹⁸
FDG. Dynamic 3D-mode PET imaging for quantification of myocardial blood flow (MBF) with
⁸²
Rb requires high count-rate capability and correction accuracy maintained over a wide range of activity. Therefore, we propose a new method to evaluate dynamic range for 3D cardiac PET imaging and evaluate the performance of a high count-rate PET system (GEHC Discovery Rx) for single-scan simultaneous quantification of MBF and left ventricular ejection fraction (EF) using list-mode PET imaging. Dynamic imaging was performed over a wide range of activities using
⁸²
Rb and
¹³
NH
3
in a heart phantom, within and without an anthropomorphic torso, by FORE-FBP and a 12 mm 3D Hann filter. Time-activity curves (TAC) in the liver, myocardium and ventricle were analyzed to determine the operating range where quantitative accuracy is maintained in the reconstructed images. Dynamic rest-stress list-mode imaging was also performed in 21 patients with
⁸²
Rb PET to compare MBF and EF quantification between 2D and 3D-modes. Based on the phantom results, injected activity was targeted at 10 MBq/kg to permit accurate measurement of the bolus first-pass activity in the LV cavity. The phantom studies indicated that activity concentrations were measured accurately (≪15% deviation) with prompt count rates ≪10 Mcps and dead-time losses ≪35%. There was no count-rate-dependent loss of resolution observed in the myocardium:liver TACs, even above these limits. Residual scatter in the ventricle cavity:liver was 3.4% and consistent across the whole dynamic range. For the patient studies, pseudo-NEC rates were 30% higher in 3D (p≪0.001), resulting in improved image quality. There were no significant differences in segmental myocardium uptake distribution, LVEF or MBF between 2D and 3D-modes (P=NS). Conclusion: Quantitative 3D cardiac imaging appears to be accurate with
⁸²
Rb activity administered in the range of 9±1.5 MBq/kg used in this study. If the dynamic range of the PET system was increased further, higher injected activity and improved ECG-gated image quality may be obtained, while still retaining the quantitative accuracy of the first-pass data for accurate MBF quantification.