Inter-vendor variability of left ventricular volumes and strains determined by three-dimensional speckle tracking...

Inter-vendor discordance in three-dimensional speckle-tracking echocardiography (3DS) remains uncharacterized. We aimed to examine inter-vendor discordance of left ventricular (LV) volumes, and functional parameters and their reproducibilities between two commercially available 3DS systems.

Echocardiographic examinations with 3DS were performed in 26 healthy subjects (age 34 ± 13 years, 85% men) using a Vivid E9 system (V1) with 4V probe (GE Health Care) and Artida (V2) with PST-25SX probe (Toshiba Medical Systems) on the same day. LV variables and global LV longitudinal, circumferential, radial, and area strains were measured by vendor-specific softwares, 4D strain EchoPAC BT11 (for V1) and 3D WMT (for V2), respectively. Reproducibility of data was assessed by an intra-class correlation coefficient (ICC).

The mean time required for 3DS analysis was 5.4 ± 1.5 min for V1, being 21% less than that for V2 (6.8 ± 1.9 min, P < 0.01). Reproducibilities of all LV strains were comparable between V1 (ICC 0.50-0.82) and V2 (ICC 0.51-0.76), except for intra-observer and inter-observer reproducibilities of radial strain being lower in V2 (ICC for V1 0.82 and 0.82 and ICC for V2 0.44 and 0.40, respectively). LV strains in all directions and area were significantly different between V1 and V2, though LV volumes and ejection fraction were comparable.

Global longitudinal, circumferential, and area LV strains are reproducible in both 3DS vendors. However, values of three-dimensional LV strains by 3DS are highly vendor-dependent.