Semi-automatic surface detection for quantification of left ventricular volumes and mass by real-time three-dimensional echocardiography

Abstract

Two-dimensional echocardiographic (2DE) assessment of left ventricular (LV) function is based on extrapolation of manually traced endocardial borders using geometric modeling. Real-time three-dimensional (RT3D) echocardiography, which does not rely on geometrical assumptions, could potentially provide more accurate measurements. The aim of this study was to use MRI measurements of end-diastolic (EDV) and end-systolic (ESV) LV volumes, ejection fraction (EF), and LV mass to validate a novel semi-automated endocardial and epicardial surface detection technique from RT3D images, acquired with a full-matrix array transducer in consecutive patients. Thirteen patients underwent RT3D and cardiac MRI. EDV, ESV and EF were derived from the semi-automatically detected LV cavity from RT3D data. All measurements were compared to MRI values using linear regression and Bland–Altman analyses. RT3D was feasible in all patients and resulted in high correlations with MRI ( r, EDV: 0.91; ESV: 0.94; EF: 0.95, LV mass: 0.84). RT3DE also showed a small bias and narrow limits of agreement with MRI (bias±2S.D., EDV: −0.6±27.8 ml; ESV: 0.7±15.1 ml; EF: −0.8±8.9%; LV mass: 1.2±37 g). Semi-automated LV endocardial and epicardial surface detection from RT3D images is feasible in consecutive patients and results in fast and accurate assessment of LV dimensions and mass.