Quantitative three-dimensional reconstruction of aneurysmal left ventricles. In vitro and in vivo validation.

ABSTRACT

BACKGROUND: Current two-dimensional (2D) echocardiographic measures of left ventricular (LV) volume are most limited by aneurysmal distortion, which restricts application of simple geometric models that assume symmetrical shape. 2D methods also fail to provide separate volumes of the aneurysm and nonaneurysmal residual LV cavity, which could help assess the stroke volume wasted by dyskinesis and the potential residual LV body to guide surgical approaches and predict their outcome. Three-dimensional (3D) echocardiographic reconstruction has potential advantages for assessing aneurysmal left ventricles because it is not dependent on geometric assumptions, does not require standardized views that may exclude portions of the aneurysm, and can potentially measure separate aneurysm and nonaneurysm cavity volumes of any shape. The purpose of this study was first, to validate the accuracy of 3D echocardiographic reconstruction for quantifying total LV and separate LV body and aneurysm volumes in vitro so as to provide direct standards for the separate volumes; and second, to determine the feasibility and accuracy of 3D echocardiographic reconstruction for quantifying the total volume and function of aneurysmal left ventricles in an animal model, providing a reference standard for instantaneous LV volume. METHODS AND RESULTS: A recently developed 3D system that automatically combines 2D images and their locations was applied (1) to reconstruct 10 aneurysmal ventricular phantoms and 12 gel-filled autopsied human hearts with aneurysms, comparing cavity volumes (total and aneurysm) to those measured by fluid displacement; and (2) to reconstruct the left ventricle during 19 hemodynamic stages in four dogs with surgically created LV aneurysms, comparing total volumes with actual instantaneous values measured by an intracavitary balloon attached to an external column for validation and also calculating the stroke volume wasted by aneurysmal dyskinesis. 3D reconstruction reproduced the distorted aneurysmal LV shapes. In vitro, calculated volumes (aneurysm, nonaneurysm, and total) agreed well with actual values, with correlation coefficients of .99 and SEEs of 3.2 to 6.1 cm3 for phantoms and 3.4 to 4.2 cm3 for autopsied hearts (mean error, < 4% for both). In vivo, LV end-diastolic, end-systolic, and stroke volumes as well as ejection fraction calculated by 3D echocardiography correlated well with actual values (r = .99, .99, .95, and .99, respectively) and agreed closely with them (SEE = 4.3 cm3, 3.5 cm3, 1.7 cm3, and 2%, respectively). The stroke volumes wasted by the aneurysm were -20.1 +/- 19.3% of LV body (nonaneurysm) stroke volume. CONCLUSIONS: Despite distorted ventricular shapes, a recently developed 3D echocardiographic system and surfacing algorithm can accurately reconstruct aneurysmal left ventricles and quantify total LV volume (validated in vivo and in vitro) as well as the separate volumes of the aneurysm and residual LV body (validated in vitro). This should improve our ability to evaluate such ventricles and guide surgical approaches.