Feasibility, Reproducibility, and Prognostic Value of Fully Automated Measurement of Right Ventricular Longitudinal Strain.

BACKGROUND:Right ventricular free wall longitudinal strain (RVFWLS) carries important diagnostic and prognostic significance in a variety of clinical settings, but its measurement is time consuming and operator dependent, limiting its value in routine clinical practice. Automated RVFWLS measurements can overcome these limitations. The purpose of this study was to determine the feasibility, reproducibility, and prognostic implications of automated RVFWLS compared with manual assessment of RVFWLS. METHODS:A total of 206 patients with a wide range of right ventricular (RV) function were retrospectively selected for this study. Fully automated (Auto-Strain), semiautomated (automated with editing), and manual (standard manual assessment) RVFWLS were measured using two-dimensional speckle-tracking echocardiography in the same RV-focused apical four-chamber view. RV ejection fraction (RVEF) was measured using three-dimensional echocardiography. Abnormal RV systolic function was defined as RVEF < 45%. Agreement for RVFWLS measurements among the three different methods was based on a combination of Pearson correlation, Bland-Altman analyses, and coefficients of variation (CVs). RESULTS:Automated RVFWLS measurements were feasible in 204 subjects (99%). Automated (8 ± 1 sec/patient) and semiautomated (50 ± 10 sec/patient) assessments had shorter analysis times compared with manual measurement (103 ± 25 sec/patient, P < .001 for both). Semiautomated RVFWLS had a stronger correlation with manual RVFWLS than automated RVFWLS (r = 0.850 vs r = 0.708, P < .001). Semiautomated RVFWLS had smaller biases, narrower limits of agreement (LOA), and lower CVs against manual assessment in comparison with the fully automated method in the whole study population (bias of 0.62 and 1.15, LOA of 6.54 and 10.50, and CVs of 9.3% and 16.9%, respectively), in the subgroup with RVEF < 45% (bias of 0.81 and 1.43, LOA of 6.32 and 10.42, and CVs of 10.2% and 18.6%, respectively), and in those with RVEF ≥ 45% (bias of 0.34 and 0.73, LOA of 6.86 and 10.63, and CVs of 8.0% and 14.5%, respectively). Seventy-six patients experienced adverse cardiovascular events during a median follow-up period of 26 months. Fully automated, semiautomated, and manual RVFWLS were associated with poor outcomes. The prognostic implications of fully automated RVFWLS were inferior to those of semiautomated and manual RVFWLS. Three RVFWLS measurements were highly reproducible. CONCLUSIONS:Both fully automated and semiautomated RVFWLS provide rapid and reproducible assessments of RV function and carry important prognostic implication. Moreover, the semiautomated approach performs better than the fully automated method. Therefore, semiautomated RVFWLS can provide a better balance between feasibility and clinical significance and displays potential for clinical application.