Head-to-head comparison of left ventricular strain assessed by CMR post-processing tools and fast Strain-ENCoded imaging

Abstract Aims Cardiovascular magnetic resonance (CMR) strain imaging allows early detection of subclinical myocardial dysfunction and provides incremental diagnostic and prognostic information. Strain can be derived from dedicated sequences such as fast Strain-ENCoded imaging (fSENC) or from post-processing of cine images using feature tracking (FT) and tissue tracking (TT). However, it remains unclear whether strain values from different approaches are directly comparable, making the definition of universal reference values difficult. This study compared left ventricular (LV) strain assessed by FT, TT, and fSENC. Methods We studied 240 individuals (183 patients recovered from COVID-19 and 57 age- and sex-matched healthy controls), who underwent standardized CMR including cine imaging and fSENC. LV global longitudinal (GLS), circumferential (GCS), and radial strain (GRS) were analyzed using FT and TT; fSENC provided GLS and GCS. Results Global strain values differed significantly between all methods (p0.001). Agreement between FT and TT was high for GLS (bias -0.8%, r=0.77) and moderate for GCS (bias -1.2%, r=0.63), but poor for GRS (bias -6.0%, r=0.37). Compared with fSENC, GLS showed moderate agreement for FT (bias 2.3%, r=0.57) and TT (bias 3.0%, r=0.59), while agreement for GCS was weaker. All approaches demonstrated excellent reproducibility. Post-COVID-19 patients showed a consistent but mild reduction in GLS compared with controls across all techniques (all p0.05). Conclusion CMR strain imaging provides fast, reliable, and reproducible measurements. However, strain values are not directly interchangeable even between similar post-processing methods or when compared to dedicated sequences, highlighting the need for standardization and method-specific reference values.