First demonstration of fully self-gated respiratory and cardiac motion-resolved 5D flow at 0.55T (RF_TH_232)

Recent advancements in free-running radial whole-heart phase-contrast MRI [1,2] enable comprehensive cardiovascular hemodynamic assessment without ECG-gating or respiratory navigation, using self-gating. These sequences simplify scan planning and compensate for respiratory motion via respiratory-resolved 5D flow or respiratory-corrected 4D flow reconstructions, validated at 1.5T [1,3]. With increasing interest in low-field MRI, our study extends the 5D flow MRI framework to 0.55T. This technique, and 4D flow generally, has not been demonstrated at this field strength due to limited availability of these systems in technically oriented institutions and the lower gradient performance. We present the first implementation and comparative analysis of 5D flow MRI at 0.55T and 1.5T, hypothesizing that 0.55T can yield comparable hemodynamic quantification in the aorta and pulmonary artery.

Methods:

Six healthy subjects (ages 24 – 41, 3 females) were scanned back-to-back on 0.55T research and 1.5T clinical scanners (Siemens Healthcare) using the free-running radial phase-contrast MRI sequence as previously reported [1]. Acquisition parameters for both sequences can be seen in table 1. The acquisitions were performed during free-breathing and in a random order to avoid fatigue bias. Cardiac and respiratory signals were extracted from k-space using compressed sensing [4,5]. The datasets acquired were analyzed using Circle CVI42 (Calgary, Canada). Three 2D planes (ascending aorta, AAo, descending aorta, DAo, and main pulmonary artery, MPa) were selected from each of the flow datasets and used to measure net-volume, peak-flow, and peak velocity. Variability across the two datasets were measured using paired t-tests.

Results:

Flow measurements at 0.55T and 1.5T were comparable, as shown by flow curves and streamlines (Figures 1 and 2). There is a strong correlation between net volume and moderate correlation for peak flow measurements when comparing 5D flow at 1.5T to 5D flow at 0.55T (r²_net-volume=0.76, r²_peak-flow=0.69) (Figure 3A), and a small bias between the measurements (Figure 3B) Paired t-test results showed only a small difference in AAo peak flow measurements (t-statistic = 2.45, p-value = 0.058), Dao and MPa measurement showed no significant differences (t-statistic = -0.33, p-value = 0.76 and t-statistic = -0.35, p-value = 0.74 respectively). Significant differences were found between the measurements at 1.5T and 0.55T, reported in the AAo measures of peak velocity (189.3±50.0 cm/s vs 109.7±19.5 cm/s, p=0.03) but not for DAo (137.2±32.3 cm/s vs 133.8±34.8 cm/s, p=0.7), and MPa (112.2±34.3 cm/s vs 99.6±29 cm/s, p=0.6).

Conclusion:

Fully self-gated, free-running 5D flow MRI was successfully implemented at 0.55T for the first time, showing comparable results to standard 1.5T imaging. Scan times were longer at 0.55T due to weaker gradients. Lower field strengths, while reducing susceptibility artifacts, lead to lower SNR [6], which can affect peak velocity measurements. This discrepancy warrants further investigation. These findings indicate that further validation and optimization is required for 5D flow MRI at 0.55T in keeping with the goal of creating a more accessible and cost-effective alternative to higher-field MRI systems without compromising diagnostic accuracy. Future work will focus on internal consistency checks of flow volumes, optimizing the pulse sequence for weaker gradients, and improving the reconstruction algorithm to better address the reduced SNR.