Quantification of Left Ventricular Volumes and Function in Patients with Cardiomyopathies by Real-time Three-dimensional Echocardiography: A Head-to-Head Comparison Between Two Different Semiautomated Endocardial Border Detection Algorithms

Abstract

We evaluated two different commercially available real-time 3-dimensional echocardiographic semiautomated border detection algorithms for left ventricular (LV) volume analysis in patients with cardiomyopathy and distorted LV geometry. A total of 53 patients in sinus rhythm with various types of cardiomyopathy (mean age 56 +/- 11 years, 28 men) and adequate 2-dimensional image quality were included. The real-time 3-dimensional echocardiographic multiplane interpolation (MI) and full volume reconstruction (FVR) methods were used for LV volume analysis. Magnetic resonance imaging was used as the reference method. A strong correlation (R(2) > 0.95) was found for all LV volume and ejection fraction measurements by either real-time 3-dimensional echocardiographic method. Analysis time was shorter with the FVR method (6 +/- 2 vs 15 +/- 4 minutes, P < .01) as compared with the MI method. Bland-Altman analysis showed greater underestimation of end-diastolic and end-systolic volumes by MI compared with FVR. For the MI method a bias of -24.0 mL (-15.0% of the mean) for end-diastolic volume and -11.3 mL (-18.0% of the mean) for end-systolic volume was found. For FVR analysis these values were -9.9 mL (-6.0% of the mean) and -5.0 mL (-9.0% of the mean), respectively. Ejection fraction was similar for the MI and FVR method with a mean difference compared with magnetic resonance imaging of 0.6 (1.0%) and 0.8 (1.3%), respectively. In patients with cardiomyopathy, distorted LV geometry, and good 2-dimensional image quality, the FVR method is faster and more accurate than the MI method in assessment of LV volumes.