Introduction

In recent years quantitative myocardial tissue characterization has focused on assessment of non-ischemic cardiomyopathies and diffuse myocardial changes - areas where conventional techniques may fail ^1,2,3. Drawbacks of conventional cardiac relaxometry include: separate data acquisitions required or quantitative T1/T2 mapping, acquired series of images requiring sophisticated motion-correction algorithms and results may be biased by fitting errors or inherent T1-T2 dependencies (e.g. T2 bias). Cardiac magnetic resonance fingerprinting (cMRF) has been proposed to acquire multiple tissue properties in a time-efficient single acquisition ⁴. There is limited experience in comparison of multiple tissue parameters derived by cMRF with commonly used cardiac relaxometry approaches. Therefore, the objective of this it to compare cMRF derived quantitative myocardial T1 and T2 data with conventional cardiac relaxometry in a cohort of hypertrophic cardiomyopathy (HCM) patients.

Methods

23 patients referred for cardiovascular MR with known/suspected HCM were prospectively enrolled. All examinations were performed at 3T (MAGNETOM Skyra , Siemens Healthineers). For myocardial T1 tissue evaluation a pre-(5(3)3) and post-contrast (4(1)3(1)2) modified Look-Locker inversion recovery (MOLLI) technique was applied in a mid-ventricular short axis orientation; a T2-prep fast low-angle shot (FLASH) technique was employed for myocardial T2 tissue evaluation in a matching slice location. ECG triggered magnetic resonance fingerprinting (cMRF) with fixed square FoV (300 or 400mm) with an acquisition time of 15HB was performed pre- and post-contrast injection in the matching MOLLI/T2-prep slice location. All quantitative post-contrast imaging was performed >10min after injection of 0.15mmol/kg Gadobutrol (Gadovist, Bayer). Inline motion correction and fitting was performed on the scanner for MOLLI T1 and T2-prep FLASH T2 maps. All acquired cMRF raw data was exported and respective T1, T2 and M0 maps reconstructed off-line (MatLab, The Math Works).All quantitative T1 and T2 maps were visually assessed for general image quality/artifacts using a 5-point Likert scale (1=non-diagnostic to 5=excellent). Quantitative assessment of all maps was performed using a commercially available algorithm (cvi42, Circle CVI) by carefully drawing endo- and epicardial contours as well as a blood-pool region-of-interest (ROI). To further avoid contamination a 15% inner/outer offset was applied for all myocardial ROI’s. Calculation of the extracellular volume (ECV) fraction was performed using standard algorithms with the use of the patient’s hematocrit acquired using a non-invasive finger probe prior to MRI. Statistical comparison included a Wilcoxon rank-sum test and Spearman’s correlation. Data are presented as median and interquartile range (IQR).

Results

Image quality assessment demonstrated superior map quality of pre-contrast T1 MOLLI over pre-contrast cMRF T1 (4 [4, 5] vs. 4 [3, 4]; P=0.004) and T2-prep FLASH over cMRF T2 (4 [4, 4] vs. 3 [3, 4]; P=0.008). No significant differences were found for MOLLI and cMRF post-contrast T1 maps (4 [4, 4] vs. 4 [3, 4]; P=0.193). There were significant differences between T1 MOLLI and cMRF T1 for both pre- contrast (1252ms [1237ms, 1286ms] vs. 1397ms [1358ms, 1437ms]; P<0.0001), and post-contrast (488ms [449ms, 521ms] vs. 520ms [458ms, 549ms]; P=0.173) maps. Pre-contrast (r=0.6039; P=0.0029) and post-contrast (r=0.8831; P<0.0001) cMRF T1 values correlated significantly with respective MOLLI data (Figure 1). T1 derived ECV demonstrated significant differences between cMRF and MOLLI acquisitions (27.6% [23.6%, 28.9%] vs. 23.4% [21.9%, 25.8%]; P=0.003) with moderate to high correlation between techniques (r=0.7721; P=0.0005) (Figure 2). cMRF T2 map data were significantly lower than T2-prep FLASH results (29.0ms [26.3ms, 31.6ms] vs. 39.7ms [37.5ms, 40.8ms]; P=0.0001) also moderate correlation between techniques (r=0.4953; P=0.07). Overall, cMRF derived T1/ECV data demonstrated a wider range of values compared to MOLLI based T1/ECV data while for quantitative T2 data the ranges between both approaches were similar.

Discussion

The results of this study demonstrate that cMRF allows for single-slice multiparametric quantitative data (T1, T2, M0) within a single heartbeat (Figures 3,4). cMRF data generally correlates well with conventional cardiac relaxometry, however demonstrate significant differences in absolute values. With respect to pre-contrast T1 values this most likely relates to the known limited accuracy of MOLLI. A potentially current limitation of cMRF includes the slightly inferior image quality which may impede the accurate delineation of myocardial borders and as such affect potential contamination of data. The overall wider range of T1 derived data for cMRF may be beneficial for differentiation of various pathologies, however, this also requires adequate data precision and reproducibility.

Conclusion

Cardiac MRF generally allows a more simplified approach to cardiac relaxometry and avoids the need for multiple separate single-contrast acquisitions. Its diagnostic capability in the differentiation and assessment of different myocardial pathologies would require further evaluation and the generation of normal healthy volunteer data for assessment of discriminatory power and establishment of normal reference ranges.

Acknowledgements

No acknowledgement found.