Scan-rescan reproducibility of ventricular and atrial MRI feature tracking strain

A feature tracking (FT) was designed to simultaneously extract myocardial strains in main cardiac chambers from cine MRI images. Its inter-observer and scan-rescan reproducibility was assessed and sample sizes required to detect predefined longitudinal changes in strain values were provided.

FT was applied on left (LV) and right (RV) ventricles as well as left atrium (LA) of 21 individuals (66 ± 10 years) who underwent 2 MRIs 2 weeks apart. Global peaks for radial, circumferential, longitudinal strains, radial motion fraction (Mr), fractional area change (FAC) and tricuspid annular plane excursion (TAPSE) were estimated. Inter-operator and inter-exam reproducibility were evaluated using coefficients of variations (CV) and intra-class correlation coefficients (ICC).

Reproducibility of all measurements were good to excellent for inter-operator (LV:CV<6.5%, ICC>0.91; RV:CV<12%, ICC>0.86; LA:CV<14%, ICC>0.85) and inter-study (LV:CV<15%, ICC>0.65; RV:CV<20%, ICC>0.71; LA:CV<20.5%, ICC>0.83) evaluations. Reasonable sample sizes are required to detect a longitudinal difference of 10-15% in strain values (LV:5 to 33 individuals, RV:14 to 62 individuals, LA:4 to 65 individuals).

FT-based functional evaluation of main heart chamber deformation from cine MRI is repeatable and thus suitable for follow-up. Strain measurements may help for the joint clinical evaluation of LV, RV or LA implication in various cardiomyopathies.