Free-running Cardiac MRF for Simultaneous T1, T2 Mapping and Cine Imaging at 0.55 T (RF_TH_280)

Cine MRI is the gold standard for left ventricular functional assessment and myocardial motion characterization^1,2. Quantitative parametric maps are widely used for disease characterization, such as fibrosis, myocardial infarction or edema^3,4. Traditionally, cine imaging, T1 and T2 mapping are performed in separate acquisitions^5-8, leading to long scan times, patient fatigue and misregistration between the images. Free-running cardiac MRF has been proposed to enable simultaneous T1, T2 mapping and cine at 1.5T and 3T^9,10. Lower-fields systems have been introduced to allow more affordable cardiac MRI and benefit of lower B0 inhomogeneity. In this study, we aim to assess the feasibility of a bSSFP free-running cardiac MRF approach for simultaneous T1/T2 mapping and cine imaging at 0.55T. 



Methods:

The proposed sequence was evaluated over a standardized T1/T2 phantom¹¹ and 2 healthy volunteers (26±1 yo, 2 male). The sequence follows a continuous breath-hold set of radial tiny golden angle (23°) bSSFP readouts. The radiofrequency (RF) pulse pattern consists of a series of sinusoidal lobes (100 RFs each) with flip-angles between 10° and [30°, 50°, 70°]¹². An adiabatic Inversion Recovery (IR) pulse was applied at the start of the sequence and four T2 preparation (T2prep) pulses were applied every 3 lobes¹². The acquired data is retrospectively assigned to different cardiac phases using a simultaneously acquired ECG signal, where each RR interval was divided into P= 8 cardiac bins using a soft-gating approach¹³. Each binned MRF time-series is reconstructed via low-rank inversion (LRI)¹⁴ and HDPROST regularization¹⁵ to obtain the cardiac phase images. One dictionary was simulated with Bloch equations, for all reconstructions, in KomaMRI¹⁶. 

Acquisition parameters include FOV=256x256 mm², voxel size=2x2x10 mm³, TE/TR=3.9/7.8 ms, IR TI=20 ms, T2prep duration=55 ms, radial spokes=1500, for a total scan time of ~13 s. Spin-echo sequences were employed to obtain both T1 and T2 phantom reference maps. All experiments were performed on a clinical 0.55T MRI scanner (MAGNETOM Free.Max, Siemens Healthineers AG, Erlangen), using a 12-channel chest array and a 9-channel embedded posterior receiver coil. ECG-synchronization was performed using an external monitoring system (MAGLIFE serenity, SCHILLER). 

Results:

Phantom results are shown in Fig. 1, T1 and T2 maps (averaged over all cardiac phases) are compared against IR-SE reference maps. Correlation results (R² >0.99 for T1 and R² >0.96 for T2) show good agreement with the reference, except for large (out of cardiac range) T2 values ( >150 ms). T1 and T2 maps across all cardiac phases for 2 healthy volunteers are shown in Fig. 2. Average septal myocardial values across all phases for both healthy volunteers are, for T1 and T2, 757±32 ms and 58±5 ms, respectively. An overestimation can be observed for T1 values and good agreement for T2 values with respect to reference literature values (T1_ref=701±24 ms, T2_ref=58±6 ms)¹⁷. Fig. 3 shows the corresponding free-running cardiac MRF cine images for both healthy volunteers. 

Conclusion:

The proposed free-running cardiac MRF framework allows for simultaneous quantification of co-registered T1 and T2 maps and cardiac cine imaging. Good accuracy and precision were observed in phantom, T1 overestimation and good agreement for T2 was observed in-vivo. Further extensions, that include non-rigid motion correction and increase the cardiac temporal resolution (number of cardiac phases), will be investigated.