冠状动脉CTA图像的序列分割算法

发布时间：2018-07-21 18:37

【摘要】：随着经济不断发展,人们的生活方式发生了巨大变化。心脑血管疾病由于高患病率和高死亡率对人类健康构成严重威胁,每年因心脑血管疾病导致的死亡率居高不下。所以对于心血管疾病早期的诊断和预防显得尤其重要。作为冠心病诊断的有效无创手段,多层螺旋CTA近些年发展迅速,而基于CTA图像的血管分割可以准确提取冠脉轮廓,是冠状动脉狭窄诊断的重要临床辅助工具,并可以提供钙化程度、斑块负担以及狭窄程度的定量分析,因此,成为医学图像处理领域研究的热点。而针对冠脉CTA序列图像的自动或半自动分割算法就有着重要的临床意义和实际价值。本文针对冠脉血管的序列分割,研究工作主要分为一下三部分:第一,主动脉血管的序列分割。传统的基于质心的方式虽然能够实现序列分割,却无法解决分裂的问题;本文提出的基于ISODATA和区域生长的序列分割新算法,通过对主动脉的目标进行分割,然后对结果利用ISODATA算法进行聚类,将聚类后得到的目标区域的聚类中心作为下一幅CT图像新的种子点再进行区域生长。该算法很好地解决了目标分裂的问题。第二,冠状动脉血管的序列分割。冠状动脉血管在CTA图像中目标区域较小,结构复杂,使自动分割有一定难度。于是,本文提出了基于特征匹配的追踪分割算法。通过对所有数据阈值化,然后利用区域特征对目标进行匹配,最后实现冠脉的序列分割。该算法能较好地适应小区域血管的识别与追踪。第三,冠状动脉序列分割算法的改进。针对第二部分冠脉序列分割算法中出现的遗漏冠脉细小血管的不足,提出了基于卡尔曼滤波的改进算法。改进主要包括两部分:一是取代全局阈值的粗分割,在上一帧数据先验知识的指导下计算最优局部阈值。二是考虑到目标血管运动幅度过大导致不在ROI区域内的问题,引入了卡尔曼滤波,以预测位置点作为ROI区域中心。改进算法明显可以识别和追踪到更多的冠脉细小血管,提高了追踪精确度。
[Abstract]:With the development of economy, people's way of life has changed greatly. Cardio-cerebrovascular diseases (CVDs) pose a serious threat to human health due to high morbidity and mortality, and the mortality caused by cardiovascular and cerebrovascular diseases (CVDs) remains high every year. Therefore, early diagnosis and prevention of cardiovascular disease is particularly important. As an effective and noninvasive method for the diagnosis of coronary heart disease, multilayer spiral CTA has developed rapidly in recent years. The segmentation of coronary artery based on CTA image can accurately extract the coronary artery profile, and it is an important clinical assistant tool for the diagnosis of coronary artery stenosis. It can provide quantitative analysis of calcification degree, plaque burden and stenosis degree, so it has become a hot spot in the field of medical image processing. The automatic or semi-automatic segmentation algorithm for coronary CTA images has important clinical significance and practical value. In this paper, the sequence segmentation of coronary artery is divided into three parts: first, the sequence segmentation of aortic vessels. Although the traditional centroid method can achieve sequence segmentation, it can not solve the problem of fragmentation. A new algorithm based on ISODATA and region growth is proposed in this paper. Then the result is clustered by ISODATA algorithm, and the cluster center of the target region is used as the new seed point of the next CT image to grow the region. The algorithm solves the problem of target splitting well. Second, the sequence segmentation of coronary artery. The target area of coronary artery in CTA image is small and the structure is complex, which makes automatic segmentation difficult. Therefore, a tracking segmentation algorithm based on feature matching is proposed in this paper. By thresholding all the data and matching the target with regional features, the coronary artery segmentation is realized. The algorithm is suitable for small area vascular recognition and tracking. Third, the improvement of coronary artery sequence segmentation algorithm. In the second part, an improved algorithm based on Kalman filter is proposed to solve the problem of missing small coronary vessels in the segmentation algorithm of coronary artery sequence. The improvement mainly includes two parts: one is to replace the rough segmentation of global threshold and to calculate the optimal local threshold under the guidance of prior knowledge of the previous frame data. Secondly, considering the problem that the target vessel movement amplitude is too large, the Kalman filter is introduced to predict the location point as the center of ROI region. The improved algorithm can identify and track more small coronary vessels and improve the tracking accuracy.
【学位授予单位】：河北大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R816.2;TP391.41

【参考文献】