Bekkers Eric J, Taylor Charles A
Department of Mechanical Engineering, Stanford University, Stanford, CA 94305, USA.
IEEE Trans Med Imaging. 2008 Mar;27(3):331-41. doi: 10.1109/TMI.2007.905081.
Computational fluid dynamics (CFD) modeling of blood flow from image-based patient specific models can provide useful physiologic information for guiding clinical decision making. A novel method for the generation of image-based, 3-D, multiscale vascular surface models for CFD is presented. The method generates multiscale surfaces based on either a linear triangulated or a globally smooth nonuniform rational B-spline (NURB) representation. A robust local curvature analysis is combined with a novel global feature analysis to set mesh element size. The method is particularly useful for CFD modeling of complex vascular geometries that have a wide range of vasculature size scales, in conditions where 1) initial surface mesh density is an important consideration for balancing surface accuracy with manageable size volumetric meshes, 2) adaptive mesh refinement based on flow features makes an underlying explicit smooth surface representation desirable, and 3) semi-automated detection and trimming of a large number of inlet and outlet vessels expedites model construction.
基于图像的患者特异性模型的血流计算流体动力学（CFD）建模可为临床决策提供有用的生理信息。本文提出了一种用于CFD的基于图像的三维多尺度血管表面模型生成的新方法。该方法基于线性三角剖分或全局光滑非均匀有理B样条（NURB）表示生成多尺度表面。稳健的局部曲率分析与新颖的全局特征分析相结合来设置网格单元大小。该方法对于具有广泛血管大小尺度的复杂血管几何形状的CFD建模特别有用，适用于以下情况：1）初始表面网格密度是平衡表面精度与可管理大小的体积网格的重要考虑因素；2）基于流动特征的自适应网格细化使得需要一个潜在的显式光滑表面表示；3）大量入口和出口血管的半自动检测和修剪加快了模型构建。
