Semi-automated estimation of left ventricular ejection fraction by two-dimensional and three-dimensional...

To compare two-dimensional (2D) and three-dimensional (3D) methods to estimate left ventricular ejection fraction (LVEF) with respect to feasibility, time consumption, and retest reproducibility.

A total of 100 patients planned to undergo coronary artery bypass grafting and/or aortic valve replacement were included consecutively. 2D and 3D echocardiography was performed on all patients. Acquisition and analysis time as well as intra- and inter-examiner variability were assessed in 50 consecutive patients with 3 repeated echocardiographic examinations and analyses. LVEF was estimated by five different methods: uniplane, biplane, and single-beat triplane (SB3P), as well as semi-automated biplane (AutoEF) and 3D volumetric tracings (4D Auto LVQ). All methods were compared to Simpson's biplane method and feasibility was determined.

Feasibility of Simpson's uniplane method, Simpson's biplane method, AutoEF, SB3P, and 4D Auto LVQ was 97%, 92%, 86%, 70%, and 89%, respectively. All methods evaluated were 18%-33% faster (P < 0.001) than Simpson's biplane method (115 seconds, standard deviation 15 seconds). Compared to Simpson's biplane method mean LVEF was slightly underestimated by 4D Auto LVQ (-2 ± 8%, P = 0.02), but not significantly different when assessed by the other methods. AutoEF and 4D Auto LVQ showed the lowest test variability (intra-examiner coefficient of variation (CV) 10%-11%; inter-examiner CV 10%-12% vs intra-examiner CV 12%-18%; inter-examiner CV 12%-20%).

Estimation of LVEF by modern semi-automated 2D and 3D echocardiographic modalities is feasible, time-efficient, and reproducible.