Automated evaluation of aortic valve stenosis from phase-contrast magnetic resonance data

Accurate quantification of aortic valve stenosis (AVS) is crucial for relevant patients management. We hypothesized that automated analysis of phase-contrast cardiovascular magnetic resonance (PC-CMR) data might provide accurate AVS evaluation in agreement with the well established transthoracic echocardiography (TTE). We studied 74 subjects (53 AVS patients, 21 controls) who had TTE and CMR on the same day. PC-CMR analysis included dynamic segmentation of left ventricular outflow tract (LVOT) and aortic valve, as well as hemodynamic parameters extraction. We performed 3 estimates of aortic valve area (AVA): AVA
_CMR1
based on Hakki's formula, AVA
_CMR2
based on continuity equation, AVA
_CMR3
simplified continuity equation=LVOT peak flow-rate/aortic peak velocity. Our analysis was reproducible (inter-operator variability<;4.56±4.40%). Strong correlations were found for comparisons between CMR and TTE aortic peak velocities and mean gradients (r=0.92, r=0.86 respectively, p<;0.0001) and between CMR and TTE AVA (r>;0.90, p<;0.0001). PC-CMR was able to detect severe AVS (accuracy>;92%) as defined by TTE. Our automated AVS evaluation would enhance CMR usefulness for comprehensive evaluation of AVS, while combining this valuable information with myocardial functional and structural findings.