Regularization of flow streamlines in multislice phase-contrast MR imaging
Regularization of flow streamlines in multislice phase-contrast MR imaging
复制标题
DOI：
10.1109/tmi.2003.814786
复制
发表时间：
2003-06-01
影响因子：
10.6
通讯作者：
Amini, AA
中科院分区：
文献类型：
--
作者：
Fatouraee, N;Amini, AA
磁共振血管造影（MRA）已成为临床评价血管疾病的重要工具。利用相位对比（PC）磁共振（MR）成像进行血流测量为评价血管内血流速度信息提供了一种强有力的方法。然而，来自复杂流动模式的图像伪影（包括慢流、再循环区和脉动流）会对结果的准确性产生不利影响。在本文中，我们介绍了一种新的数值计算公式，用于提高PC速度场和相应的流线的精度，基于流体动力学的物理约束，在正则化框架内。该方法利用流函数，自动施加不可压缩流的连续性约束，从PC图像中重建出流线。我们将该算法应用于轴对称腹主动脉瘤的流动体模中获得的复杂MR成像流速。该算法显着改善流线的结果，特别是在回流区，其中文物更加明显。文中还提出了一种原始变量形式的速度重建方法，并与流函数法进行了比较。为了验证从PC MR图像获得的流动特性，我们使用FLUENT计算流体动力学软件包，模拟与我们的体模相同几何结构内的流动模式。有一个很好的协议之间的数值模拟和恢复PC流线结果。经过处理的流线，在流函数和原始变量的方法，更现实，并提供更精确的流动模式比未经处理的PC数据。此外，在正常志愿者的主动脉中证明了该方法的可行性。
Magnetic resonance angiography (MRA) has become an important tool for the clinical evaluation of vascular disease. Flow measurement with phase-contrast (PC) magnetic resonance (MR) imaging provides a powerful method for evaluation of blood velocity information inside vessels. However, image artifacts from complex flow patterns including slow now, recirculation zone, and pulsatile flow can adversely affect accuracy of results. In this paper, we introduce a new numerical formulation for improving the accuracy of PC velocity fields and corresponding streamlines, based on a physical constraint from fluid dynamics, within a regularization framework. The formulation which makes use of a stream function, automatically enforces continuity constraint of incompressible now and reconstructs the flow streamlines from PC images. We applied the algorithm to complex MR imaging flow velocities obtained in a flow phantom of an axisymmetric abdominal aortic aneurysm. The algorithm significantly improved streamline results especially inside the recirculation zone, where artifacts are more pronounced. A velocity reconstruction method in primitive variable form is also presented and results are compared with the stream function method. In order to validate flow characteristics derived from PC MR images, we used the FLUENT computational fluid dynamics software package, to simulate flow patterns within the same geometry as our phantom. There was a good agreement between the numerical simulations and recovered PC streamline results. Processed streamlines, in both stream function and primitive variable methods, were more realistic and provided more precise flow patterns than unprocessed PC data. Additionally, the feasibility of the method was demonstrated in the aorta of a normal volunteer.