C21 - Comparison of Three-dimensional Mitral Valve Morphologic Measurements Obtained Using Two Different Ultrasound Machines

Abstract:

Background: Real-time three-dimensional transthoracic echocardiography (RT-3DTTE) has been utilized to assess the geometry of the mitral valve (MV) in dogs. Although cross-platform software for MV 3D modeling exists, it is unknown whether the machine used to acquire images affects MV morphologic variables.

Objective: To compare MV morphologic variables between RT-3DTTE datasets acquired with two different ultrasound machines.Animals: Eleven dogs

Methods: Prospective diagnostic comparison study. Each dog underwent RT-3DTTE acquisition using two different ultrasound machines: Philips Epiq CVx and Canon Aplio i900. RT-3DTTE datasets were analyzed with dedicated software for MV morphologic analysis by a single observer. Bland-Altman analysis was performed for each variable. The mean of the absolute differences was compared to previously reported inter-observer repeatability coefficients (RC).

Results: A significant bias between Philips and Canon measurements was identified for the following MV morphologic variables: normalized anterolateral-posteromedial annulus diameter (-(-0.11 [-0.21 – -0.02]; P=.02), sphericity index (0.06 [0.005 – 0.11]; P=.03), non-planar angle (12.43 [3.20 – 21.65]; P=.01), tenting area (-0.15 [-0.27 – -0.03]; P=.02), tenting height (-0.98 [-1.79 – -0.17]; P=.02), and normalized commissural diameter (-0.12 [-0.22 – -0.02]; P=.02). The measurement difference was greater than previously reported inter-observer RCs for the non-planar angle, annulus height, tenting height, normalized posterior leaflet area, and annulus height to commissural width ratio, suggesting that the difference is beyond what expected just by inter-observer variability. Conclusions and Clinical Importance: MV morphologic variable measurements can differ between ultrasound machines. These differences may be relevant when utilizing RT-3DTTE for patient assessment influencing surgical planning.