Myocardial blood flow (MBF) and coronary flow reserve (CFR) measurement with dynamic SPECT: Operator variability

Objectives MBF and CFR quantification with dynamic SPECT (dSPECT) could lead to widespread improvement in clinical practice with little financial overhead. This work evaluates the operator-dependent variability of FlowQuant MBF analysis of dSPECT images.

Methods 15 patients underwent rest and dipyridamole-stress imaging with SPECT/CT using dynamic image acquisition (≥10s time frames) <10s after MIBI injection. Quantitative images were iteratively reconstructed with all physical corrections and then processed using FlowQuant to semi-automatically segment myocardial and blood regions of interest (ROI). A 1-tissue-compartment kinetic model was applied to these ROI sampled time-activity-curves, and MIBI extraction correction was applied to derive MBF. CFR was estimated as stress/rest MBF ratios. Images were processed twice (≥7 days apart) by each of two operators independently. Intra- and inter-operator variability of MBF and CFR in three primary coronary territories was evaluated.

Results Mean rest MBF, stress MBF, and CFR were 1.28 mL/min/g, 2.54 mL/min/g, and 2.30. Operators 1 and 2 intervened during processing of 50% and 85% of images resulting in combined rest-stress regional MBF intra-operator correlation r²=0.98 and 0.96 (n=90, p=0.002), and Bland-Altman derived reproducibility coefficients (RPC)=0.33 and 0.53 mL/min/g (p<0.001). Inter-operator variability had r²= 0.94 and RPC=0.63 mL/min/g (n=180, p<0.001 vs. intra-operators). CFR had higher regional variability (r²=0.95, 0.88, 0.89 and RPC=0.59, 0.83, 0.83 for intra1, intra2, inter-operator).

Conclusions MBF and CFR measurements using quantitative dynamic MIBI SPECT are highly reproducible and warrant further investigation. A new human derived extraction function (compared with PET) may be necessary to calibrate MBF values.