Abstract
1799
Objectives Routine quantitation of myocardial blood flow (MBF) and coronary flow reserve (CFR) with PET has the potential to provide new clinically relevant data. However, to be useful, the parameter estimation algorithms and software must be sufficiently robust to provide reproducible results across both users and software choice. To test robustness of method, three pharmacokinetic (PK) software packages, each with Rb-82 MBF quantitation models, were exercised and compared across a population of 38 cardiovascularly normal subjects.
Methods Three PET PK software packages (PMOD-PMOD Technologies, Syngo MBF-Siemens Molecular Imaging, and FlowQuant-Ottawa Heart Inst.) were tested. 38 normal subjects underwent rest/dipyridamole stress Rb-82 myocardial perfusion imaging using standard clinical techniques (40 mCi injection, 6 minute acquisition) using a Siemens Biograph 40 PET/CT scanner with high-count-rate detectors. List mode data was histogrammed into 26 dynamic frames (12x5s, 6x10s, 4x20s, 4x40s). MBF and CFR were calculated with all three software programs using similar Rb-82 single compartment models. MBF and CFR were calculated for the three vascular territories and 17 segments using each program's default parameters and approaches.
Results Parametric polar flow and CFR maps from all three software platforms generated images concordant in appearance (higher and lower flow observed in the same anatomical region) and with similar, but not identical perfusion values. PMOD stress/rest flows were systematically 10%/20% higher than Syngo MBF, which were, in turn 10%/20% higher than FlowQuant estimates. It appears that subtle differences in spillover corrections and blood pool definition drive these differences.
Conclusions Quantitative MBF and CFR measurement can be reliably performed on any of the 3 software platforms tested. Results are internally consistent. However, systematic differences in calculated MBF and CFR between platforms are large enough to suggest that harmonization of PK model assumptions would aid in clinical adoption of this potentially useful quantitative methodology.
Research Support Siemens Molecular Imagin