基于超声图像的颈总动脉内中膜厚度测量算法研究

发布时间：2018-03-11 03:35

本文选题：医学图像处理　切入点：B超图像　出处：《天津大学》2014年硕士论文　论文类型：学位论文

【摘要】：人体颈总动脉内中膜厚度(Intima-Media Thickness,IMT)可以作为心脑血管疾病严重程度的一个评估指标,对于心脑血管疾病的预防及辅助治疗具有重要意义。在临床医学上,利用B超技术对颈总动脉进行成像,并利用计算机辅助方法进行IMT测量。为了满足大批量数据和高准确度的测量要求,本文的目标是提出一种自动、快速、准确的IMT测量方案。在本文提出的方案中,IMT的测量过程具体分为六个步骤:(1)利用模板匹配算法在复杂背景中自动提取感兴趣区域(Region of Interest,ROI),(2)在ROI内进行双边滤波以去除噪声干扰,(3)在ROI中基于灰度阈值估计管腔-内膜边界(Lumen-Intima Interface,LII),(4)根据LII和梯度图估计中膜-外膜边界(Media-Advantia Interface,MAI),(5)利用改进的动态规划算法将两条估计的边界分别调整至准确的位置,(6)计算并评估IMT。方案设计完成后,利用含有100幅颈总动脉B超图像的图像库进行算法测试。对于每幅图像,分别进行两种方式的IMT测量。第一种方式是利用手动勾画测量IMT,将结果作为真实值;第二种方式是利用本文设计方案自动测量IMT,将结果作为实验值。在正确测量的93幅图像中,IMT的实验值与真实值之间的相关系数为0.9629,二者偏差的均值?标准差为0.0231?0.0348 mm。MATLAB程序在Windows XP Intel Core 2 Duo CPU E7500@2.93GHz环境下运行,平均每幅图像的处理时间为0.32秒。与国内外同领域的研究相比,本文方案实现了自动、快速、准确的IMT测量目标。
[Abstract]:IMT can be used as an index to evaluate the severity of cardiovascular and cerebrovascular diseases. It is of great significance in the prevention and treatment of cardiovascular and cerebrovascular diseases. The common carotid artery was imaged by B-ultrasound and IMT was measured by computer aided method. In order to meet the requirements of mass data and high accuracy, the aim of this paper is to propose an automatic and fast method. Accurate IMT measurement scheme. In the proposed scheme, the measurement process is divided into six steps: 1) using template matching algorithm to automatically extract region of interest in complex background. (2) Two-sided filtering is carried out in ROI. Estimation of lumen-Intima Interface based on Gray threshold in ROI) estimation of media-advantia boundary based on LII and gradient map) using improved dynamic programming algorithm to adjust the two estimated boundaries to accuracy. To calculate and evaluate IMT. upon completion of the design, The algorithm is tested by using the image library containing 100 common carotid artery B-mode ultrasound images. For each image, two kinds of IMT measurements are carried out. The first way is to use manual sketching to measure the IMTs and take the results as the true values. The second way is to measure the IMT automatically by using the design scheme in this paper, and take the result as the experimental value. The correlation coefficient between the experimental value of IMT and the real value is 0.9629 in 93 correctly measured images. The standard deviation is 0.0231? The 0.0348 mm.MATLAB program runs under the environment of Windows XP Intel Core 2 Duo CPU E7500 @ 2.93 GHz, and the average processing time of each image is 0.32 seconds. Compared with the research in the same field at home and abroad, this scheme achieves the automatic, fast and accurate IMT measurement target.
【学位授予单位】：天津大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：R543.4;TB559

【相似文献】