Feasibility, Reproducibility, and Clinical Implications of the Novel Fully Automated Assessment for Global Longitudinal Strain

Background: Despite evidence of its usefulness, measurement of global longitudinal strain (GLS) has not been widely accepted as a clinical routine, because it requires proficiency and is time consuming. Automated assessment of GLS may be the solution for this situation. The aim of this study was to investigate the feasibility, reproducibility, and predictive value of automated strain analysis compared with semiautomated and manual assessment of GLS. Methods: In this validation study, different methods for the assessment of GLS were applied to echocardiograms from 561 asymptomatic subjects (mean age, 71 6 5 years) with heart failure risk factors, recruited from the community. All patients had both data on follow-up outcomes (new heart failure and cardiac death) and interpretable echocardiographic images for strain analysis. Measurement of GLS was repeated using the same apical images with three different measurement packages as follows: (1) fully automated GLS (AutoStrain), (2) semiautomated GLS (automated, corrected by a trained investigator), and (3) manual GLS (standard manual assessment by a trained investigator). Results: AutoStrain measurements were technically feasible in 99.5% of patients. Calculation times for automated (0.5 +/- 0.1 min/patient) and semiautomated assessment (2.7 +/- 0.6 min/patient) were significantly shorter than for manual assessment (4.5 +/- 1.6 min/patient; P < .001 for both). Approximately 40% of patients were thought to need manual correction after automatic calculation of GLS. Therefore, there was considerable discordance between automated and semiautomated and manual GLS. Over a median of 12 months of followup, cardiovascular events (new heart failure and cardiac death) occurred in 66 patients (11.8%). Automated GLS showed the potential to correctly detect normal and abnormal systolic function and predict cardiac events; the predictive value was inferior to that of semiautomated GLS. Conclusions: A novel fully automated assessment for GLS may provide a technically feasible, rapidly reproducible, and clinically applicable means of assessing left ventricular function, but a substantial number of automatic traces still need manual correction by experts. At the present stage, the semiautomated approach using this novel automated software seems to provide a better balance between feasibility and clinical relevance.