Interstudy reproducibility of biplane cine nuclear magnetic resonance measurements of left ventricular function

Cine nuclear magnetic resonance (NMR) imaging, as a noninvasive and high-resolution imaging modality, has been shown to be reliable for determining absolute left ventricular (LV) volumes and ejection fraction. A relatively new gradient echo cine NMR approach using 2 orthogonal long-axis planes (2- and 4-chamber) aligned with the true axes of the left ventricle has been previously developed and validated against radiographic biplane LV cineangiography. The aim of the present investigation was to determine the reproducibility of this more rapid cine NMR approach for the measurement of LV volumes and ejection fraction. Eighteen normal subjects underwent 2 cine NMR studies, on different days, using a 1.5-tesla clinical imaging system. Studies were analyzed on-line and blindly by 2 independent observers. Intraobserver error was also determined in a blinded manner. Mean values of measurements determined by this method in this group of normal subjects were end-diastolic volume (120 ± 20 ml), end-systolic volume (39 ± 9 ml) and ejection fraction (67 ± 4%). Paired analysis of data revealed no significant bias between interstudy, interobserver or intraobserver measurements, except for interobserver end-diastolic volume, where the first observer measurements were slightly elevated (5.6 ± 7.8 ml) compared with the second. This resulted in a small difference in ejection fraction (1.7 ± 2.3%) between observers. The absolute variation between measurements (square root of variance components) was low for all interstudy, interobserver and intraobserver comparisons: end-diastolic volume was