Fluid-Structure-Interaction Modelling of the Heart Hemodynamics using Statistical Shape Models

心脏血流动力学分析在心脏病的诊断和治疗中发挥着越来越重要的作用。作为心血管疾病血液动力学生物标志物的复杂参数的常规评估目前是困难的。然而，数值模拟允许计算患者特定心脏功能的空间和时间分辨信息。为了分析心脏功能，人们提出了各种建模方法，从简单的零维集总单元模型(LEM)到非常复杂的耦合电、结构和流体力学的4D模型。该项目的最终目标是允许在临床条件下进行特定于患者的模拟，从而在临床例行公事中支持医生。所提出的数值方法将结合计算流体力学(CFD)和统计形状模型(SSM)。SSM允许将患者特定的几何形状描述为加权形状模式的叠加。因此，可以基于平均形状创建左心室的标准化模拟，然后通过调整网格节点对其进行个性化。不同的成像方式，即CT、MRI和超声心动图，被用来对模拟进行参数化。使用规定的运动来模拟心室收缩，这意味着收缩是从图像数据中获取的。虽然开放的主动脉和二尖瓣叶的形状也是由形状模型描述的，但它们的运动是通过机械有限元模拟来计算的，因为对叶的时间分辨分割通常是不可能的，或者与高不确定性相关。此外，描述前负荷和后负荷的边界条件，特别是在肺静脉和主动脉，由个性化的LEM提供。所提出的CFD-SSM数值方法是灵活的，并允许识别最优模型的复杂性：足够简单以适用于临床环境，并且足够复杂以提供精确诊断和治疗决策所需的信息水平。这种方法应该能够为临床医生提供健壮和快速的心脏血流动力学模拟。它有可能实现患者特定模型的完全自动化。自动化将提高可用性，并将所有与用户相关的错误降至最低。该项目最初的临床重点是心脏瓣膜疾病。该方法的可用性将通过使用磁共振成像的体外实验来展示，以分析模型中的血流动力学。该项目的成功实现需要临床和工程科学家之间的密切合作，这一点在柏林夏洛特大学心血管计算机辅助医学研究所很好地建立了。

Analysis of cardiac hemodynamics plays an increasing role in diagnosis and treatment of heart diseases. A routine assessment of complex parameters proposed as hemodynamic biomarkers for cardiovascular diseases is currently difficult. However, numerical modeling allows calculation of spatially and temporally resolved information of patient-specific heart function. Various modelling approaches are proposed for analysis of heart function, ranging from simple 0D Lumped Element Models (LEM) to very complex 4D models coupling electro-, structural, and fluid mechanics. The final aim of this project is to allow patient-specific simulations under clinical conditions and thus support physicians during clinical routine. The proposed numerical approach will combine computational fluid dynamics (CFD) and statistical shape models (SSM). SSM allow the description of patient-specific geometries as a superposition of weighted shape modes. It is therefore possible to create a standardized simulation of the left ventricle based on an average shape, which is then personalized by adjusting mesh nodes. Different imaging modalities, namely CT, MRI and echocardiography, are used to parameterize the simulation. The ventricular contraction is modelled using a prescribed motion, meaning the contraction is acquired from image data. Although the shape of the opened aortic and mitral valve leaflets is also described by a shape model, their motion is calculated by means of mechanical finite element simulations, since a time-resolved segmentation of leaflets is often not possible or associated with high uncertainty. Furthermore, the boundary conditions describing pre- and afterload, specifically at the pulmonary veins and aorta, are provided by a personalized LEM. The proposed CFD-SSM numerical approach is flexible and allows identification of an optimal model complexity: simple enough to be applicable in a clinical setting and complex enough to provide required level of information for precise diagnosis and treatment decision. The approach should enable robust and fast simulation of cardiac hemodynamics for clinicians. It has the potential to allow complete automation of patient-specific modelling. Automation will improve usability and minimize all user-related errors. The initial clinical focus of the project are heart valve diseases. The usability of the approach will be shown using in-vitro experiments using MRI for an analysis of hemodynamics in phantom models. The successful realization of the project requires close cooperation between clinical and engineering scientists, which is well established at the Institute of Cardiovascular Computer-assisted Medicine at Charité – Universitätsmedizin Berlin.