Left ventricular volumes assessed by different new three-dimensional echocardiographic methods and ordinary biplane technique

Research paper by Olaf Rodevand, Reidar Bjornerheim, Svend Aakhus, John Kjekshus

Indexed on: 01 Feb '98Published on: 01 Feb '98Published in: The International Journal of Cardiovascular Imaging


Three-dimensional (3D) echocardiography may overcome the problems with inadequate accuracy and reproducibility of 2D volume measurements of the left ventricle. Aims: To establish the in vitro accuracy and reproducibility of two new methods for 3D echocardiographic volume determination as compared to biplane measurements. Methods: Validation of volume measurements by a multiplane 3D method was performed on asymmetric latex phantoms (n=8, true volumes 45-304 ml) using rotational acquisition of 90 image planes. Porcine agarose-filled asymmetrical left ventricles (n=7, true volumes 34 – 280 ml) were measured by the same multiplane 3D method based on images acquired by probe rotation axis perpendicular (A) and parallel (B) to the ventricular long axis. Ventricular volumes were also obtained by a simplified 3D system using only the three standard apical views (C) and by the ordinary biplane Simpson’s method (D). Results: On latex phantoms systematic deviation from true volumes by multiplane 3D was less than 2%, and 95% variability of individual measurements from this mean was ± 4,9%. For accuracy on left ventricles, systematic bias was small with all the methods (<5%), but 95% variability of individual measurements was ±9,0%, 15.4%, 18.8% and 41.3% of true volumes for methods A-D respectively. Corresponding results in the same range were obtained for inter- and intraobserver variability. Conclusion: Individual in vitro volume estimates of left ventricles are of similar quality using apical multiplane or apical triplane 3D echocardiography. Both methods were superior to the ordinary apical biplane method, but inferior to multiplane 3D method with the probe directed perpendicular to the ventricular long axis.

More like this: