Freehand three-dimensional echocardiography for determination of left ventricular volume and mass in patients with abnormal ventricles: comparison with magnetic resonance imaging

Abstract

Objective: The objective of this study was to validate the freehand three-dimensional echocardiographic method in patients with abnormal ventricular geometry compared with two-dimensional echocardiography using magnetic resonance imaging as a standard.

Background: Two-dimensional echocardiographic methods for estimating left ventricular volume and mass in clinical use today are limited by inaccuracies and variations caused by use of geometric assumptions and errors in image plane positioning. Freehand three-dimensional echocardiography with operator guidance by a "line of intersection" display eliminates these assumptions and errors. This method of volume and mass computation has been validated as highly accurate and reproducible in healthy subjects.

Methods: Left ventricular end-systolic and end-diastolic volumes and myocardial mass were determined by freehand three-dimensional echocardiography, by conventional two-dimensional echocardiography using the apical biplane summation of discs method (volume) and the truncated ellipsoid method (mass), by M-mode echocardiography using the Penn method (mass), and by magnetic resonance imaging in 30 patients selected only for the presence of an abnormal ventricle. Results were compared by means of linear regression and the Bland-Altman method of analysis.

Results: There was excellent correlation, low bias, and low variability between three-dimensional echocardiography and magnetic resonance imaging for end-diastolic volume (r = 0.90, standard error of the estimate = 31.8 ml, bias = -28.4 ml), end-systolic volume (r = 0.93, standard error of the estimate = 24.1 ml, bias = -13.1 ml), and mass (r = 0.90, standard error of the estimate = 27.3 gm, bias = -22.6 ml). Two-dimensional echocardiography was less accurate and more variable as follows: end-diastolic volume (r = 0.70, standard error of the estimate = 39.8 ml, bias = -33.5 ml), end-systolic volume (r = 0.78, standard error of the estimate = 31.2 ml, bias = -26.7 ml), and mass (r = 0.80, standard error of the estimate = 37.3 gm, bias = 28.9 ml). M-mode echocardiography mass determination (Penn method) was least accurate and most variable (r = 0.075, standard error of the estimate = 78.3 gm, bias = 78.3 gm).

Conclusions: Freehand three-dimensional echocardiography is a method of high accuracy and low variability for computing left ventricular volume and mass in clinical patients with abnormal ventricles. It is superior to conventional one- and two-dimensional echocardiography. The improvement achieved is attributed to elimination of geometric assumptions and image plane positioning errors and additional sampling of the ventricle.