Inter-study repeatability of left ventricular strain measurement using feature tracking on MRI cine images

Abstract

Feature tracking (FT) is a novel approach for myocardial strain evaluation from standard cine MRI. Conversely to tagging, an original feature of FT is its ability to analyze all heart chambers, independently of their wall thickness. However, few technical challenges, including its inter-study repeatability need to be investigated and this is the aim of our study. We studied 10 healthy individuals (59±9 years), who underwent 2 MRI exams, separated by two weeks. For each exam, cine short-axis slices corresponding to left ventricular (L V) basal, mid-L V and apical locations were analyzed, by the same operator, using a custom FT software to extract global peaks of 1-radial thickening (Er), 2-circumferential strain (Ec), and 3-endocardial radial motion (Erm). Briefly, the FT, based on spatial correlation, was applied to myocardial contours initially drawn on a single time phase. Bland-Altman analysis revealed low bias for comparison between the 2 exams (mean bias [±1.96·standard deviation] were: 0.02 [-0. 11,0. 14]% for Er, 0.00 [-0.04,0.05]% for Ec, 0.00 [-0.05,0.06]% for Erm). As expected, strain magnitudes decreased with age. The application of an automated FT on data systematically acquired during MRI provided consistent and reproducible LV myocardial strain measurements.