Possible benefits of ultra high field MRI in cardiac diffusion tensor imaging

Introduction: In vivo diffusion tensor cardiac magnetic resonance (DT-CMR) has recently been used to characterize myocardial microstructure dynamics during left ventricular wall thickening, showing distinct, abnormal sheetlet function in HCM and DCM patients. However, the necessary trade-off between the number of averages, diffusion weighted directions, and image resolution in clinically used parameter settings have been shown to lead to poor precision of sheetlet and helix angles (HA) of±31∘ and±14∘ respectively. MRI at higher field strength such as 7T may provide higher signal-to-noise ratio (SNR), allowing for improvement of spatial and angular resolution, image quality as well as consistency of diffusion metrics in clinically applied parameter settings. Purpose: To assess feasibility of DT-CMR at 7T and compare SNR and diffusion metrics in spin echo based diffusion tensor acquisitions of the unfixed, ex vivo porcine heart at 7T and 3T. Methods: Measurements were performed on 7T and 3T whole body MRI systems using 1Tx/32Rx head coils. Porcine hearts were imaged in physiological saline solution, unfixed, at room temperature within 10 hours after cardiac arrest. MRI with spatial resolution of 1.3x1.3x1.3 mm3 was performed obtaining 25 and 8 cDTI datasets at 3T and 7T as a reference. Scans at 7T were repeated using parallel imaging acceleration (GRAPPA) factors R = 2, R = 3, R = 4. Apparent diffusion coefficient (ADC), fractional anisotropy (FA), HA and secondary eigenvector angle (E2A) were compared within segmentation according to the American Heart Association (AHA). Results: DTI data was successfully acquired for all parameter settings used. SNR values in b0 images of scans at 3T and 7T was 34±3 and 45±4 for identical scans. On average DTI analysis was performed for 6±2 slices of apical cap, 8±1 apical, 8±1 mid-cavity, and 8±1 basal slices. Maximal and minimal deviations of ADC [10-3mm2/s], FA, and |E2A| [∘] within the AHA segments were used to define a bias range for 7T acquisitions relative to the 3T reference. The smallest bias was found for scans using GRAPPA factor R = 2 (Figure 1). Derived metrics from scans using GRAPPA factor R = 3 showed significant differences in FA in mid-cavity and basal parts. No significant differences were found for scans using GRAPPA factor R = 2. Conclusion: Feasibility of DTI acquisitions of ex vivo porcine hearts was demonstrated for spin echo based acquisitions with spatial resolution of 1.3x1.3x1.3 mm3 via comparison with a reference data set measured at 3 T. Derived metrics were statistically similar to the 3T reference data using GRAPPA factors R = 2 and R = 3. The relative SNR gain in 7T acquisitions may benefit clinically used parameter settings for DT-CMR allowing optimization of the trade-off between averages, diffusion directions, and image resolution, especially, when using stimulated echo approaches, which can utilize the increased T1 at higher field strengths.