LAUSR

Background We tested the repeatability of myocardial blood flow (MBF) quantified using 82 Rb with and without motion correction (MC) and with arterial input functions estimated from left ventricle (LV) and atrium (LA). Methods Twenty-one patients referred for clinical 82 Rb PET/CT underwent repeated rest scans in a single imaging session. Global MBF was quantified using three different assessments by two operators: (1) automatic processing without MC and LV arterial input function (AIF), (2) with MC and LV-AIF, and (3) with MC and LA-AIF. Inter-scan and inter-operator repeatability were tested using coefficient of variation (CV). Results MC with LV-AIF did not change MBF (no MC: 1.01 ± 0.30 mL/min/g vs MC with LV-AIF: 1.01 ± 0.29, P  = 0.70), whereas MC with LA-AIF showed significantly lower MBF assessments (0.95 ± 0.28 mL/min/g, P  = 0.0006). We report significant improvement for test-retest reproducibility for global MBF following MC (CV; No MC: 16.0, MC (LV-AIF): 9.2, MC (LA-AIF): 8.8). Good inter-operator repeatability was observed for LV-AIF (CV = 4.7) and LA-AIF (CV = 5.6) for global MBF assessments. Conclusions MC significantly improved the test-retest repeatability between operators and between scans. MBF obtained after MC with LV-AIF were comparable, whereas MBFs after MC and LA-AIF were significantly reduced.