Abstract 14232: Performance Of Cardiac Magnetic Resonance Fingerprinting Mapping - Evaluation By Comparison With Standard Techniques In A Large Patient Population

Abstract

Background: Parametric mapping is now widely used as a part of cardiac MRI protocols. Cardiac Magnetic Resonance Fingerprinting (cMRF) techniques for parametric mapping allow for simultaneous evaluation of T1 and T2 within a single acquisition. In this study, we evaluate the performance of cMRF through comparison with clinically utilized pulse sequences for T1 and T2 in a large heterogeneous population. Hypothesis: cMRF T1 and T2 measurements agree with clinically available mapping sequences in a heterogeneous patient cohort. Methods: We obtained myocardial T1 and T2 values across the whole mid short axis slice of the left ventricle in 175 patients using cMRF (T1/T2), Modified Look-Locker Inversion Recovery (MOLLI, T1), and T 2 -prepared balanced steady-state free precession (bSSFP). Patients were classified into the groups “Ischemic cardiomyopathy (ICM)”, “Nonischemic Cardiomyopathy (NICM)”, “Normal”, and “Other”. The group classified as “Other” was not analyzed separately. Results: cMRF had similar T1 compared with MOLLI (1036 vs 1046 ms, p=0.06, r=0.76) and slightly lower T2 compared with bSSFP (46.1 vs 49.6 ms, p<0.01, r=0.5) in the entire cohort. Mean T1 was higher in ICM compared to Normal when measured by both MRF (1074 vs 997, p<0.01) and MOLLI (1073 vs 1015, p<0.01). The same was true when comparing NICM and Normal for both MRF and MOLLI (p<0.01). T1 was more concordant in NICM (r=0.82) vs ICM (r=0.46) and Normal (r=0.54) while T2 had similar concordance across the spectrum of pathologies. Bland-Altman plots showed good agreement between cMRF and MOLLI (Bias: -9.86, 95% CI: -98.58, 78.86)/bSSFP (Bias: -3.55, 95% CI: -11.68, 4.57). The concordance was stable across a range of heart rates. Conclusions: cMRF is an emerging technique for simultaneous T1/T2 mapping in the myocardium. In a large heterogeneous population, cMRF T1 and T2 values showed good agreement with clinically-utilized pulse sequences. Further work is needed to identify sources of discrepancies.