Synopsis

Keywords: Atherosclerosis, Perfusion

Recently, a fully automated QP CMR workflow has been established, which provides measures of stress MBF according to a 32-segmentation model with subdivision into endo- and epicardial subsegments. We compared the diagnostic accuracy of QP CMR according to the standard AHA 16-segment model (16M-QP) and the newly developed automated 32-segment model (32M-QP) with conventional visual assessment in patients who underwent adenosine stress perfusion CMR imaging followed by invasive coronary angiography and/or coronary computed tomography angiography. Our preliminary data have not shown superiority of diagnostic accuracy of 16M-QP or 32M-QP in comparison to visual assessment of adenosine stress first-pass perfusion imaging.

Introduction

In daily clinical routine, stress perfusion cardiovascular magnetic resonance (CMR) is generally assessed by visual interpretation of first-pass perfusion images. Alternatively, introduction of quantitative perfusion CMR (QP CMR) enables absolute estimation of myocardial blood flow (MBF). Recently, a fully automated QP CMR workflow has been established, which provides measures of stress MBF according to a 32-myocardial segmentation model. This model is based on the standard 17-segment American Heart Association (AHA) model excluding apex, and further subdivision into endo- and epicardial subsegments. Currently, scarce data have shown, that QP CMR has high diagnostic accuracy for detecting obstructive CAD, but it is not superior to conventional visual assessment¹. Some studies suggested that myocardial segments subdivision improves diagnostic accuracy of myocardial perfusion measurements. However, this has not been tested using full quantification of myocardial blood flow (MBF) by automated QP CMR². This study, therefore, aims to compare the diagnostic accuracy of fully automated 32-segment QP CMR and conventional visual assessment in detection of obstructive coronary artery disease (CAD).

Methods

The retrospective analysis included 23 patients who underwent adenosine stress perfusion CMR imaging at a 3T whole body scanner (Vida, Siemens, Erlangen, Germany) according to a dual-bolus scanning protocol followed by invasive coronary angiography and/or coronary computed tomography angiography (CCTA) within 6 months. Qualitative assessments of first-pass perfusion images were made by level 3 CMR experts. QP was analysed off-line using cvi42 software (Circle Cardiovascular Imaging Inc, Calgary, Canada) equipped with a newly updated, fully automated pixel-wise QP module. Stress MBF in ml/g/min was measured for automatically determined transmural myocardial segments according to the AHA 16-segment model (16M-QP) and the newly developed automated 32-segment model with epi- and endocardial subdivision (32M-QP). A receiver operating characteristics (ROC) curve per-vessel analysis was performed to evaluate and compare the diagnostic accuracies of 16M-QP, 32M-QP and conventional visual assessment for detection of obstructive CAD. Coronary territories with presence of late gadolinium enhancement were excluded from the analysis.

Results

In total, 60 vessels were analyzed. Areas under the curve (AUCs) of stress MBF for 16M-QP, 32M-QP and visual assessment were 0.702 (p=0.04), 0.732 (p=0.005) and 0.744 (p=0.008), respectively. Comparison of AUCs showed no significant differences in diagnostic accuracy of 16M-QP, 32M-QP and conventional visual assessment in detection of obstructive CAD (p=NS).

Discussion

Both 16M-QP and 32M-QP show high diagnostic accuracy for detecting obstructive CAD. However, our preliminary data have not shown superiority of diagnostic accuracy of 16M-QP or 32M-QP in comparison to visual assessment of adenosine stress first-pass perfusion imaging.

Conclusions

In fully automated QP CMR, subdivision of myocardial segments into endo- and epicardial layers does not improve the overall diagnostic accuracy. Although this preliminary data show comparable diagnostic performance of QP CMR and expert visual assessment, further studies are needed to evaluate the clinical utility of QP CMR in a daily clinical routine.

Acknowledgements

No acknowledgement found.