Aortic valve timing is critical for accurate estimation of MRI-derived ejection fraction

Background We aim to investigate the effect of utilizing aortic valve timing in the measurement of cardiac magnetic resonance (CMR)-derived ejection fraction (EF). Although CMR-derived left ventricular (LV) EF is the gold-standard for volumetric evaluation of the heart, it always reports higher values when compared to other modalities such as echocardiography and radionuclide ventriculography. The higher EF has often been attributed to better endocardial definition by CMR. In the ACCF/ACR/ AHA/NASCI/SCMR 2010 Expert Consensus Document on CMR, a 1987 paper is cited with utilizes the “maximal and minimal left ventricular cross-sectional area at the mid-ventricle” to identify end-systole and end-diastole. The 2013 CMR Guideline suggests phases with the largest and smallest global LV blood volume be utilized to identify end-diastole and end-systole. The same guideline also suggests aortic valve timing be utilized when arrhythmia or mitral regurgitation is encountered. We investigated the effect of different schemes on measured LVEF. Methods Standard retrospectively EKG-gated bSSFP short axis and LVOT cine images in 46 clinical patients in normal sinus rhythm with a range of EF (5.8 75.8 %) were acquired and analyzed. Scan parameters were as follows: in-plane resolution: 1.25 2.08 mm, slice spacing: 8 mm with a gap of 2 mm, reconstructed temporal resolution: 18.2 58.8 ms. Aortic valve opening and closing as a percentage of cardiac cycle was identified on LVOT images. Volumetric evaluation of short axis images was performed via semi-automated segmentation for all phases and all left ventricular slices (ITK-SNAP, Philadelphia PA). The schemes (shown in Table 1) were used to quantify LVEF. Scheme G was considered “true EF” and usied aortic valve opening and closing. Scheme F is intended to mimic real-time methods, which acquire slice-by-slice data and do not incorporate aortic valve timing or other external measures such as EKG.