Introduction

The clinical gold standard for measuring the severity of aortic valve stenosis (AVS) is TTE, which offers a fast and inexpensive exam that can quantify key biomarkers. However, TTE is sensitive to direction and misalignment of the beam with the aortic jet can cause underestimation of the peak velocity. In addition, certain patient anatomies prohibit proper beam alignment and necessitate transesophageal echocardiography, which is an invasive procedure. TTE has been shown to be suboptimal in 30% of patients^1. Phase-contrast MRI (PC-MRI) based 4D flow measurements do not suffer from the misalignment issues of TTE. However, conventional 4D flow requires long acquisition time, exceeding 10 or even 20 minutes, limiting its clinical applications. The recently proposed ReVEAL4D² demonstrated the feasibility of acquiring 4D flow data within a single breath-hold for peak velocity estimation at the aortic valve. In this work, we compare the efficacy of single breath-hold 4D flow imaging prototype against TTE and navigator respiratory-gated free-breathing 4D flow imaging to assess the severity of AVS in nineteen patients.

Methods

For this study, 19 patients with moderate to severe AVS, as measured by TTE, were recruited to undergo a 4D flow MRI scan within two months of the TTE exam. TTE measured the maximum velocity above the aortic valve (V_max). All MRI exams were performed on 1.5T clinical scanner (MAGNETOM Avanto, Siemens Healthcare, Germany). The MRI protocol included two 4D flow scans, each covering a thin (48 mm) slab encompassing the aortic valve. First, a respiratory navigator-gated acquisition accelerated with rate 3 GRAPPA was performed with an average scan time of 14 min . Second, a single breath-hold (24 heartbeats) 4D flow scan at R=21-27 was performed. The single breath-hold scan was reconstructed using ReVEAL4D² offline with custom Matlab (Mathworks, Natick Massachusetts) code. V_max was calculated from the PC-MRI scans by contouring the four imaging planes above the aortic valve. To extract V_max, the maximum velocity within the region of interest over imaging plane and all cardiac phases was found for each patient. For single breath-hold scans, relevant scan parameters were as follows: TR 4.4 ms, TE 2.18 ms, TRES 36ms, flip angle 10°, FOV 430x346-280x217 mm, matrix size [144,108,12]-[144,84,12] with 50% oversampling, in plane resolution <= 3x3 mm2, and slice thickness 6 mm. For respiratory navigator-gated GRAPPA the relevant scan parameters are as follows: TR 4.4 ms, TE 2.18 ms, TRES 36ms, flip angle 10°, FOV 217x310-343x490 mm, matrix size [160,120,12], in plane resolution <=3mm, and slice thickness 6 mm. Bland-Altman, Pearson correlation, and linear regression where performed to compare the V_max of ReVEAL4D to TTE, ReVEAL4D to GRAPPA, and GRAPPA to TTE.

Results

Statistics comparing ReVEAL4D and TTE are shown in Figure 1. V_max estimated using ReVEAL4D is in good agreement (r = 0.643) with TTE. In Figure 2, the statistics comparing V_max estimated from ReVEAL4D and GRAPPA are shown. ReVEAL4D shows stronger agreement with GRAPPA (r=0.682) than with TTE. Comparison between TTE and GRAPPA is shown in Figure 3. GRAPPA and TTE show the strongest agreement (r=0.843). However, GRAPPA consistently overestimates the peak velocity and takes 20-30 times longer to acquire. Representative magnitude and velocity maps from GRAPPA and ReVEAL4D reconstructions for a single slice are shown in Figure 4.

Discussion and Conclusion

We have demonstrated the feasibility of quantifying V_max in a single breath hold using 4D flow imaging. Future work will explore additional AVS biomarkers calculated from single breath-hold acquisitions, such as effective valve area and peak pressure gradient across the aortic valve.

Acknowledgements

This work was funded in part by NIH projects R21EB021655 and R01HL135489. The Tesla K40 used for this research was donated by the NVIDIA Corporation. RA and OPS receive research support from Siemens Healthineers. OPS receives funding support from The Robert F. Wolfe and Edgar T. Wolfe Foundation.