Abstract
Background: The choice of software package (SP) for image processing affects the reproducibility of myocardial blood flow (MBF) values in [(13)N]NH(3) PET/CT scans. However, the impact of motion correction (MC) tools-integrated software motion correction (ISMC) or data-driven motion correction (DDMC)-on the inter-software reproducibility of MBF has not been studied. This research aims to evaluate reproducibility among three commonly used SPs and the role of MC. Methods: Thirty-six PET/CT studies from patients without myocardial ischemia or infarction were processed using QPET, Corridor-4DM (4DM), and syngo.MBF (syngo). MBF and coronary flow reserve (CFR) values were obtained without motion correction (NMC) and with ISMC and DDMC. Intraclass correlation coefficients (ICC) and Bland-Altman (BA) plots were used to analyze agreement. Results: Good or excellent reproducibility (ICC ≥ 0.77) was found for rest-MBF values, regardless of the SPs or use of MC. In contrast, stress-MBF and CFR values presented mostly a moderate agreement when NMC was used. The RCA territory consistently had the lowest agreement in stress-MBF and CFR in the comparisons involving QPET. The use of MC, particularly DDMC, enhanced the reproducibility of most of the stress-MBF and CFR values by improving ICCs and reducing bias and limits of agreement (LoA) in BA analysis. Conclusions: MBF quantification agreement between SPs is strong for rest-MBF values but suboptimal for stress-MBF and CFR values. MC tools, especially DDMC, are recommended for improving reproducibility in stress-MBF assessments, although differences in SP reproducibility up to 0.77 mL/g/min in global stress-MBF and up to 0.88 in global CFR remain despite the use of MC.