摘要目的:提出一种自门控欠采样方法结合压缩感知技术实现高时间分辨率快速四维磁共振血流速度成像（4D flow MRI）。方法:本研究将自门控技术和压缩感知技术运用在4D flow MRI技术上，开发出一种高时间分辨率的自门控欠采样4D flow MRI快速成像方法，并通过血管狭窄流体模型实验和6位志愿者实验来验证可行性。在水模流体实验中使用Wilcoxon秩和检验分析传统相位对比磁共振成像（PC-MRI）和自门控欠采样高时间分辨率4D flow MRI数据一致性。2023年9月至10月招募志愿者，使用高时间分辨率的自门控欠采样4D flow MRI采集到的原始志愿者数据通过心脏分相和呼吸加权后进行压缩感知欠采样重建，使用配对 t检验验证高时间分辨率和低时间分辨率下的自门控欠采样4D flow MRI主动脉流速峰值数据一致性。 结果:在流体水模实验中，PC-MRI和高时间分辨率的4D flow MRI测量静止水模得到均值和峰值流速都趋近于0。而在流动数据中，PC-MRI和高时间分辨率的4D flow MRI测得流动数据的均值分别为7.802±7.787，8.339±8.039，两者差异无统计学意义（ P>0.05）。在志愿者实验中，高时间分辨率的心脏相位数是低时间分辨率的2倍以上，高时间分辨率和低时间分辨率的峰值数据比较差异无统计学意义（ P>0.05），具有较强的一致性。 结论:高时间分辨率4D flow MRI和传统的相位对比磁共振以及低时间分辨率4D flow MRI具有良好的一致性，增加自门控采集频率以提高4D flow MRI时间分辨率具有可行性。

abstractsObjective:To propose a self-gating undersampling method combined with compressed sensing technology to achieve high temporal resolution fast 4D flow MRI.Methods:In this study, we applied self-gating and compressed sensing techniques to 4D flow MRI, developing a high temporal resolution self-gating undersampling 4D flow MRI rapid imaging method. The feasibility of this method was validated through vascular stenosis fluid model experiments and experiments with 6 volunteers conducted from September to October 2023. The raw volunteer data acquired using high temporal resolution self-gating undersampling 4D flow MRI were undersampled and reconstructed after cardiac gating and respiratory weighting. In the fluid phantom experiments, the consistency between traditional phase-contrast magnetic resonance imaging (PC-MRI) and self-gating undersampling high temporal resolution 4D flow MRI data was analyzed using the Wilcoxon rank-sum test. In the volunteer experiments, the paired t-test was used to verify the consistency of peak aortic flow velocity data between high temporal resolution and low temporal resolution self-gating undersampling 4D flow MRI. Results:In the fluid phantom experiments, data from PC-MRI and high temporal resolution 4D flow MRI had means and peak velocities close to 0, aligning with theoretical expectations. For flowing data, PC-MRI recorded a mean of 7.802±7.787, while high temporal resolution 4D flow MRI recorded a mean of 8.339±8.039. There was no significant difference between the two methods ( P>0.05). In the volunteer experiments, the number of cardiac phases in high temporal resolution imaging was more than double that of low temporal resolution imaging. The comparison of peak flow velocity data from high and low temporal resolution undersampled 4D flow MRI showed no statistically significant difference ( P>0.05), demonstrating strong consistency. Conclusion:High temporal resolution 4D flow MRI and traditional phase contrast magnetic resonance imaging, as well as low temporal resolution 4D flow MRI, have good consistency. Increasing the self-gating acquisition frequency to improve the temporal resolution of 4D flow MRI is feasible.