Left and Right Atrial Volume by Freehand Three-dimensional Echocardiography: In Vivo Validation Using Magnetic Resonance Imaging

INTRODUCTION: Measurement of left and right atrial size is important for the management of arrhythmias, valvular and congenital heart disease. We have demonstrated that freehand three-dimensional (3D) echocardiography is more accurate and reproducible than two-dimensional (2D) echocardiography for measurement of left ventricular mass and volume. However, no prior study has validated the accuracy of freehand 3D for the determination of left or right atrial volume. METHODS: End-systolic (maximum) left and right atrial volumes were determined in 21 volunteer patients and normal subjects by one, two, and freehand 3D transthoracic echocardiography and compared to volumes obtained by gradient recalled magnetic resonance imaging. Three-dimensional echocardiographic determination of atrial volume was obtained using an acoustic spatial locator, a line-of-intersection display, and a surface reconstruction algorithm. Two-dimensional echocardiographic atrial volumes were obtained from apical biplane images of the left atrium and an apical single plane image of the right atrium using a summation of disks method. One-dimensional (ID) estimates of left atrial volume were determined by cubing the M-mode ID antero-posterior dimension obtained on the parasternal long axis view. RESULTS: An excellent correlation was Obtained between freedhand 3D echocardiography and magnetic resonce imaging (MRI) for the left atrium (r = 0.90, SEE=9.6 ml) and for the right atrium (r = 0.91, SEE = 8.8 ml) with a small bias (left atrium 5.25 ml, right atrium 12.06 ml) and narrow limits of agreement (left atrium 22.14 ml, right atrium 25.54 ml). Two-dimensional echocardiography correlated less well (left atrium r = 0.87, SEE = 10.23 ml, right atrium r = 0.79, SEE = 19.74 ml), and had a higher bias (left atrium 14.46 ml, right atrium 8.99 ml) and larger limits of agreement (left atrium 24.37 ml, right atrium 41.16 ml). One-dimensional estimates of left atrial volume correlated poorly with magnetic resonance determined left atrial volume (r = 0.80, SEE = 6.61 ml) and had unacceptably high bias (45.09 ml) and limits of agreement (35.52 ml). Interobserver variability was lowest for 3D echocardiography (left atrium 7.2 ml, 11%, right atrium 8.7 ml, 16%). CONCLUSIONS: Freehand 3D echocardiography using the line of intersection display for guidance of image positioning and a polyhedral surface reconstruction algorithm is a valid, accurate, reproducible method for determining left and right atrial volume in humans that is comparable to magnetic resonance imaging and is superior to current ID and 2D echocardiographic techniques.