Use of three-dimensional speckle tracking to assess left ventricular myocardial mechanics: inter-vendor consistency and reproducibility of strain measurements

Abstract

Since there is insufficient data available about the inter-vendor consistency of three-dimensional (3D) speckle-tracking (STE) measurements, we undertook this study to (i) assess the inter-vendor consistency of 3D LV global strain values obtained using two different scanners; (ii) identify the sources of inter-vendor inconsistencies, if any; and (iii) compare their respective intrinsic variability. Sixty patients (38 ± 12 years, 64% males) with a wide range of LV end-diastolic volumes (from 74 to 205 ml) and ejection fractions (from 17 to 70%) underwent two 3D LV data set acquisitions using VividE9 and Artida ultrasound systems. Global longitudinal (Lε), radial (Rε), circumferential (Cε) and area (Aε) strain values were obtained offline using the corresponding 3D STE softwares. Despite being significantly different, Lε showed the closest values between the two platforms (bias = 1.5%, limits of agreement(LOA) from -2.9 to -5.9%, P < 0.05). Artida produced significantly higher values of both Cε and Aε than VividE9 (bias = 6.6, LOA: -14.1 to 0.9%, and bias = 6.0, LOA = -28.2-8.6%, respectively, P < 0.001). Conversely, Rε values obtained with Artida were significantly lower than those measured using VividE9 platform (bias = -24.2, LOA: 1.5-49.9, P < 0.001). All strain components showed good reproducibility (intra-class correlation coefficients: 0.82-0.98), except for Rε by Artida, which showed only a moderate reproducibility. Apart from Lε, the inter-vendor agreement of Rε, Cε and Aε measured with Artida and VividE9 was poor. Reference values should be specific for each system and baseline and follow-up data in longitudinal studies should be obtained using the same 3D STE platform.