血管内超声斑点的概率模型建立及应用

发布时间：2018-03-10 05:31

本文选题：斑点　切入点：血管内超声　出处：《南方医科大学学报》2017年11期 　论文类型：期刊论文

【摘要】：超声斑点是由人体组织中散射体的反射信号相干作用所形成的,其概率分布与生物组织的结构信息密切相关,即不同的组织结构所产生斑点的概率分布形式不同。根据血管内超声(IVUS)斑点的概率分布特性,本文提出用斑点的伽马混合模型和高斯混合模型分别描述血管内超声组织中的钙化斑块、软斑块和正常血管区域。通过KS检验,KL散度和相关系数等指标分析,发现钙化斑块和正常血管区域的斑点概率分布符合高斯混合模型,而软斑块更接近伽马混合模型。在此研究基础上,本文提出一种结合邻域信息的概率混合模型,用于IVUS图像斑块分割,与现有的概率混合模型比较,分割精度大大地提高,且受噪声影响减少。
[Abstract]:Ultrasonic speckle is formed by coherent action of reflected signal of scatterer in human body. The probability distribution of ultrasonic speckle is closely related to the structure information of biological tissue. That is, the probability distribution of spots produced by different tissue structures is different. According to the probability distribution characteristics of intravascular ultrasound spots, In this paper, the speckle gamma mixed model and Gao Si mixed model are used to describe calcified plaques, soft plaques and normal vascular regions in intravascular ultrasound respectively. The KL divergence and correlation coefficient are analyzed by KS test. It is found that the probability distribution of calcified plaque and normal vascular region accords with Gao Si's mixed model, while the soft plaque is closer to the gamma mixed model. On the basis of this study, a probabilistic mixed model combining neighborhood information is proposed in this paper. Compared with the existing probabilistic mixed model, the segmentation accuracy of IVUS image patch segmentation is greatly improved, and the noise is reduced.
【作者单位】：南方医科大学生物医学工程学院广东省医学图像处理重点实验室;
【基金】：国家自然科学基金(61771233,61271155)~~
【分类号】：R445.1;TP391.41

【相似文献】