冠状动脉斑块材料性质在体确定，预测及其力学分析

Coronary artery disease (CAD) is the number one killer globally. Atherosclerosis is the underlying mechanism for CAD. Atherosclerosis happens as plaque comprised of lipid-rich necrotic core, calcium, fibrous tissue and other substances builds up in the inner lining of the coronary. Over time the buildup of plaque and its rupture will eventually lead to the narrowing, stiffening of the artery, which prevents proper blood flow to the heart and causes serious CAD. Numerous clinical studies have shown plaque morphology alone is not enough for accurate diagnosis and treatment to this disease. Even though patient-specific plaque material properties are extremely difficult to quantify in vivo due to lack of data and methods, they are the vital indicators for atherosclerosis severity and also critical for accurately simulating the plaque biomechanics. Patient-specific follow-up in vivo Cine-IVUS and VH-IVUS were acquired, and together with 3D thin-slice models for plaque material properties quantification in vivo. Three-dimensional IVUS-based fluid-structure interaction model with in vivo plaque material properties will be constructed for accurately simulating the biomechanical conditions. Key risk factors related to plaque progression and rupture will be extracted from plaque material properties, plaque morphology and biomechanics. In this project, patient-specific plaque material properties change from baseline to follow-up will be predicted using the baseline plaque material properties, morphology and biomechanical conditions. Generalized linear mixed models will be used to find the best indicator(s) for predicting material properties change from baseline to follow-up. Prediction accuracy are measured by specificity and sensitivity. Plaque material properties and biomechanics will also be used to classify the stable/vulnerable/ruptured plaque, furthermore be used in assessing plaque vulnerability. This project will help us better understand the mechanisms for plaque progression and rupture, and furthermore provide efficient guidance in prevention, early diagnosis, clinical treatment and management for CAD.

冠心病是当今的致死率和致残率最高的疾病之一。冠状动脉血管粥样硬化过程是造成冠心病的潜在机理。动脉粥样硬化斑块在血管内膜处增长和破裂，会使血管材料变硬，管腔变窄，导致心脏组织得不到足够的血液供应，产生功能异常。大量的研究表明斑块形态信息并不足以作为斑块发展和破裂的诊断指标。尽管与斑块硬化发展和斑块内力学模拟密切相关，斑块组织材料性质的在体确定一直是该领域的一个瓶颈。本项目按照随诊复查制度，获取患者在两个时间点（初诊时刻和复查时刻）的血管内超声等数据。我们结合血管内超声图像和有限元数值方法来确定斑块组织的在体材料性质，并将其应用到流固耦合数值模型当中，模拟斑块力学情况。本项目将通过引进统计方法来研究斑块材料性质，形态和力学风险因子对斑块硬化发展预测和易损性量化评价的有效性，从而进一步对冠心病的预防，早期诊断和临床治疗提供重要指导。

本项目围绕冠状动脉粥样硬化这一疾病，通过患者医学影像分析和建立力学模型获得了冠脉的在体材料性质。在此基础上建立了基于患者血管内超声(IVUS)影像的三维流固耦合模型，采用统计模型结合斑块形态和力学去预测斑块进展。具体的内容如下：.(1) 我们从 13 名患者获取 20 个动脉粥样硬化斑块的体内 Cine IVUS 和VH-IVUS 数据建立单固体薄片模型，通过迭代算法患者斑块特定的在体材料参数。研究结果表明，斑块材料特性在患者间和患者内存在很大差异。所有斑块的杨氏模量平均值分别为轴向599.5 kPa和圆周方向1042.8 kPa。通过对在体材料和离体材料进行了对比分析，观察到在体材料对应变分布的影响很大。我们的结果可以为心血管疾病研究提供更准确的应变/应力计算。 .(2) 本研究采用基准时刻提取的斑块形态学和生物力学因素对斑块进展进行预测分析。我们采用斑块从基准时刻到随访时刻的斑块形态易损指标(MPVI)的变化量作为斑块进展的衡量标准。选取的三种预测模型分别广义线性混合模型(GLMM),支持向量机(SVM)和随机森林(RF)方法。使用RF的最佳预测因子组合的预测精度为0.9147，而SVM的预测精度为0.9078，GLMM的预测精度为0.8556。RF将GLMM的预测精度提高了5.91%。预测模型的结果也证明了形态学和生物力学因素结合在一起可以提供比单独的任何单一风险因素更高的预测准确性。此外，我们还提出一种新的多因素决策方法，该方法简单、可扩展性好的特点在该工作中得到验证。 (3) 我们提出了一种基于冠脉相干断层成像(OCT)的新的流固耦合模型(FSI)模型。该模型考虑了冠脉内血管-血液-血栓三者之间的相互作用的力学关系，用于冠状动脉血栓对冠脉生物力学的影响。初步结果表明认为冠状动脉血栓会影响冠状动脉血流动力学和固体力学，并进一步会影响抗栓治疗的预后。该建模方法可以进一步用于研究血栓相关临床问题的生物力学机制，如血栓形成、血栓溶解、血栓脱落和血栓变形与医疗器械相互作用。