Background

The usage of tissue-mapping techniques in various cardiac conditions is increasingly popular, as it allows a robust tool in differentiating between normal and abnormal myocardial tissue. Several T1-mapping sequences have been proposed for pixel-wise quantification of longitudinal relaxation with variable reproducibility and precision of native T1 values. The ongoing optimisation of T1 mapping sequences by the vendors and others further compound a uniform solution for T1 mapping with established the normal values and pathological ranges, which are sequence-dependent. We undertook a head to head comparison of two MOLLI sequences for native T1 in reproducibility and discrimination between healthy and pathological myocardium.

Methods

A total of 51 subjects (controls, n=11; patients with LV hypertrophy n=1; and LV dilatation (n=32) underwent T1 mapping at 1.5 and 3T using FFM-MOLLI and VendorProduct MOLLI). Native T1 values were measured in a mid-ventricular short axis slice by drawing region of interest (ROI) conservatively within the septal myocardium (Figure 1). The reproducibility of measurements (the intra- and interobserver and interstudy reproducibility) of myocardial T1 mapping measurements has been assessed using Bland-Altman plots and Pearsons correlations. Discrimination between health and disease was performed using the ROC curve analysis.

Results

Patients had significantly higher native T1 values compared to controls for both sequences (p<0.05, Figure 2). Effect size by Cohen-D was 2.1 for FFM MOLLI and 1.7 for Vendor MOLLI, respectively. Intra- and inter-observer agreements for native T1 values across the whole cohort were very high (FFM MOLLI r=0.991; r=0.972; Vendor MOLLI: r=0.96 and r=0.89, for all). Similarly, the intra- and inter-observer coefficients of variation (CoV) for T1 (0.8%; 1.5% vs. 1.1% 2.7%) were low.There was an overall smaller bias in measurements for intra observer reproducibility compared to inter-observer measurements (Figure 3). Comparison of reproducibility between controls and patients with LVH and DCM, respectively, as well as between 3 and 1.5T yielded no observable discrepancies in reproducibility (Figures 2 and 3). AUCs across the cohort (controls vs. patients) were: 0.94 (0.86-0.99) vs. 0.79 (0.65 -0.95),

respectively.

Conclusion

Different MOLLI sequences differ for normal values, effect sizes and in discrimination between health and disease. Myocardial T1 mapping with is a promising, reproducible method to characterize the contribution of myocardial disease, at both clinically used field strengths. Further work needs to be conducted to minimize the possibility of systematic bias by establishing standards in measurement, data acquisition and image analysis.