Towards standardized postprocessing of global longitudinal strain by feature tracking - OptiStrain CMR-FT study

Background: Left ventricular global longitudinal strain (GLS) with cardiovascular magnetic resonance (CMR) is an important prognostic biomarker. Its everyday clinical use is limited due to methodological and postprocessing diversity among the users and vendors. Standardization of postprocessing approaches may reduce the random operator-dependent variability, allowing for comparability of measurements despite the systematic vendor-related differences.

Methods: We investigated the random component of variability in GLS measurements by optimization steps which incrementally improved observer reproducibility and agreement. Cine images in two-, three- and four-chamber-views were serially analysed by two independent observers using two different CMR-FT softwares. The disparity of outcomes after each series was systematically assessed after a number of stepwise adjustments which were shown to significantly reduce the inter-observer and intervendor bias, resulting standardized postprocessing approach. The final analysis was performed in 44 subjects (ischaemic heart disease n = 15, non-ischaemic dilated cardiomyopathy, n = 19, healthy controls, n = 10). All measurements were performed blind to the underlying group allocation and previous measurements. Inter- and intra-observer variability were tested using Bland-Altman analyses, intra-class correlation coefficients (ICCs) and coefficients of variation (CVs).

Results: Compared to controls, mean GLS was significantly lower in patients, as well as between the two subgroups (p < 0.01). These differences were accentuated by standardization procedures, with significant increase in Cohen's D and AUCs. The benefit of standardization was also evident through improved CV and ICC agreements between observers and the two vendors. Initial intra-observer variability CVs for GLS parameters were 7.6 and 4.6%, inter-observer variability CVs were 11 and 4.7%, for the two vendors, respectively. After standardization, intra- and interobserver variability CVs were 3.1 and 4.3%, and 5.2 and 4.4%, respectively.

Conclusion: Standardization of GLS postprocessing helps to reduce the random component of variability, introduced by inconsistencies of and between observers, and also intervendor variability, but not the systematic inter-vendor bias due to differences in image processing algorithms. Standardization of GLS measurements is an essential step in ensuring the reliable quantification of myocardial deformation, and implementation of CMR-FT in clinical routine.