What is the primary source of discordance in strain measurement between vendors: imaging or analysis?

Abstract

Strain imaging quantifies myocardial deformation in both overt and subclinical myocardial diseases. Strain values have been shown to vary between vendors. The aim of this study was to investigate whether this variation reflects differences of image characteristics or analysis software. We enrolled 45 patients (age 54 ± 14 years, 26 males) who underwent two echocardiograms on the same day with two different commercially available ultrasound systems. Three apical views were analyzed by two vendor-specific and two digital imaging and communications in medicine (DICOM)-based software packages. The results were compared in two ways; comparison of global longitudinal strain (GLS) by nine different combinations of ultrasound systems and strain software, and comparison of strains by wall, view and global levels. Comparison of GLS between two vendor-specific software packages yielded poor correlation (ρ = 0.35). The use of the same software on the images from different vendors was concordant (ρ = 0.74 and 0.64; bias = 0.0 and -0.5). There were no significant differences in correlation coefficients among the comparisons with DICOM-based software. There was no significant difference in correlation coefficients among three views (4CV vs. 3CV, p = 0.44; 4CV vs. 2CV, p = 0.47). Comparison of walls showed the septum to have the best correlation (ρ = 0.73), and the posterior wall (ρ = 0.31; p = 0.005 vs. septum) had the worst. Use of the same software to measure strain in images from different vendors minimizes the variation of GLS. Post-processing is the most important determinant in inter-vendor variation, with differences in acquisition having a small effect. These findings should be examined and confirmed with other combinations of ultrasound machines and deformation software.