Miyazaki Shohei, Itatani Keiichi, Furusawa Toyoki, Nishino Teruyasu, Sugiyama Masataka, Takehara Yasuo, Yasukochi Satoshi
Cardio Flow Design, Inc., Chiyoda, Tokyo, Japan.
Department of Cardiovascular Surgery, Kyoto Prefectural University of Medicine, 1-5 Hangi-cho, Shimogamo, Sakyo-ku, Kyoto, 606-8522, Japan.
Heart Vessels. 2017 Aug;32(8):1032-1044. doi: 10.1007/s00380-017-0979-2. Epub 2017 Apr 25.
Computational fluid dynamics (CFD) are the gold standard in studying blood flow dynamics. However, CFD results are dependent on the boundary conditions and the computation model. The purpose of this study was to validate CFD methods using comparison with actual measurements of the blood flow vector obtained with four-dimensional (4D) flow magnetic resonance imaging (MRI). 4D Flow MRI was performed on a healthy adult and a child with double-aortic arch. The aortic lumen was segmented to visualize the blood flow. The CFD analyses were performed for the same geometries based on three turbulent models: laminar, large eddy simulation (LES), and the renormalization group k-ε model (RNG k-ε). The flow-velocity vector components, namely the wall shear stress (WSS) and flow energy loss (EL), of the MRI and CFD results were compared. The flow rate of the MRI results was underestimated in small vessels, including the neck vessels. Spiral flow in the ascending aorta caused by the left ventricular twist was observed by MRI. Secondary flow distal to the aortic arch was well realized in both CFD and MRI. The average correlation coefficients of the velocity vector components of MRI and CFD for the child were the highest for the RNG k-ε model (0.530 in ascending aorta, 0.768 in the aortic arch, 0.584 in the descending aorta). The WSS and EL values of MRI were less than half of those of CFD, but the WSS distribution patterns were quite similar. The WSS and EL estimates were higher in RNG k-ε and LES than in the laminar model because of eddy viscosity. The CFD computation realized accurate flow distal to the aortic arch, and the WSS distribution was well simulated compared to actual measurement using 4D Flow MRI. However, the helical flow was not simulated in the ascending aorta. The accuracy was enhanced by using the turbulence model, and the RNG k-ε model showed the highest correlation with 4D Flow MRI.
计算流体动力学（CFD）是研究血流动力学的金标准。然而，CFD结果取决于边界条件和计算模型。本研究的目的是通过与使用四维（4D）血流磁共振成像（MRI）获得的血流矢量实际测量值进行比较，来验证CFD方法。对一名健康成人和一名患有双主动脉弓的儿童进行了4D血流MRI检查。对主动脉腔进行分割以可视化血流。基于三种湍流模型，即层流、大涡模拟（LES）和重整化群k-ε模型（RNG k-ε），对相同的几何结构进行CFD分析。比较了MRI和CFD结果的流速矢量分量，即壁面剪应力（WSS）和流动能量损失（EL）。MRI结果的流速在包括颈部血管在内的小血管中被低估。通过MRI观察到左心室扭转导致升主动脉中的螺旋流。在CFD和MRI中均很好地实现了主动脉弓远端的二次流。儿童的MRI和CFD速度矢量分量的平均相关系数在RNG k-ε模型中最高（升主动脉中为0.530，主动脉弓中为0.768，降主动脉中为0.584）。MRI的WSS和EL值不到CFD的一半，但WSS分布模式非常相似。由于涡粘性，RNG k-ε和LES中的WSS和EL估计值高于层流模型。CFD计算实现了主动脉弓远端的精确血流，并且与使用4D血流MRI的实际测量相比，WSS分布得到了很好的模拟。然而，升主动脉中的螺旋流未被模拟。使用湍流模型提高了准确性，并且RNG k-ε模型与4D血流MRI显示出最高的相关性。
