基于IVOCT影像的斑块组成分析及植入支架评价研究

Coronary artery disease has very high mortality and morbidity rate. The current pathology study provides evidence that the rupture of “vulnerable plaque” is the main cause of acute coronary events. However, the traditional low-resolution imaging techniques do not have sufficient resolution to observe individual plaque component. So its effect on plaque vulnerability and the pathogenesis of coronary artery disease are unclear. Moreover, during the intervention treatment, there may be stent malapposition and then causing stent thrombosis, due to the invisibility of implanted stent. Intravascular optical coherence tomography (IVOCT), as a new imaging technique with high spatial resolution at the micron scale, is expected to become the standard for plaques composition analysis and implanted stents assessment. This research is to develop a complete automated analysis system for IVOCT images. First, distinguished features extraction and effective classifier construction methods are proposed to analysis the proportion and morphology of plaques composition. This can provide more accurate and effective information for the research of pathogenesis of coronary artery disease. Second, to assess the implanted stents, an accurate stent detection method is proposed by combining low-level features and 3D spatial features. This can provide effective evaluation basis for assessing stents position during the surgical therapy. Our objective is to understand the pathogenesis of coronary artery disease and improve the accuracy of stent implantation, and to explore new methods and technologies.

冠心病是高发病率、高死亡率的疾病，目前病理学发现引发急性冠心病事件的主要元凶是“易损斑块”的破裂。然而由于传统影像方式分辨率低，无法看清斑块组成成分，其对斑块稳定性的影响及冠心病的发病机理尚未明确；而且介入治疗(PCI)中，由于无法看清植入支架的贴壁情况，容易发生支架贴壁不良而导致术后支架血栓。血管内光学相干断层扫描（IVOCT）是一种新型的空间分辨率达微米级的成像方式，有望成为分析斑块组成及评价植入支架的标准。本项目拟构建一套完整的IVOCT图像自动分析系统，首先提出高可分特征的提取方法以及高判别性分类器，对斑块组成成分的比例、形态等进行分析，为研究冠心病的发病机理提供精确有效的信息；其次提出结合底层特征和3D空间特征的高精度支架检测方法，对植入支架进行评价，为临床手术中评价支架贴壁情况提供有效的评估依据。本项目为明确冠心病的发病机理及提高手术中植入支架的准确性，探索新方法、新技术。

研究IVOCT影像分析相关的高精度、快速、自动化的算法，主要是关于支架植入术前与术后的自动分析算法，为临床研究提供客观、定量的分析依据。主要成果包括： 1）提出一种新的导丝自动分割算法与条状高斯滤波核预处理方法分别解决了导丝阴影和血液伪影的挑战，并通过引入水平集模型到IVOCT影像的血管内壁分割中，解决了因不稳定斑块与分叉血管所造成的血管内壁轮廓形变大的挑战，本算法的内壁分割的DSC平均值为0.98±0.01，与现有方法相比，本算法的精度更高、更稳定。2）提出一种关于主血管分割与分叉口检测的算法，创造性地提出以轮廓的法向量与点到中心向量的夹角值为特征检测分叉口，并用差分滤波与两个向量的方向性进一步强化了该特征，与专家手工标注对比，主血管分割的DSC平均值为0.96，分叉口检测误差为0.22mm，证明了该算法的有效性。3）提出了一种先检测后分割的可降解支架贴壁情况自动分析的新框架，并提出基于改进的多层Adaboost分类器的支架检测算法与基于动态规划的支架分割算法；与专家手工标注对比，支架检测的F值平均为0.90，支架分割的DSC平均值为0.81，一组回拉数据平均分析时间约14.8秒，证明了该算法的有效性与时效性。4）提出基于支架的四个角点检测及其具有显著的盒形特征先验知识的支架分割方法，与专家手工标注对比，支架分割的DSC平均值为0.82，实现的支架自动检测的进一步优化。

内容获取失败，请点击重试