血管内超声图像的仿真  

Simulation of intravascular ultrasound images

作　　者：张麒[1] 汪源源[1] 王威琪[1] 马剑英[2] 钱菊英[2] 葛均波[2] 

机构地区：[1]复旦大学电子工程系,上海200433 [2]复旦大学附属中山医院心内科,上海200032

出　　处：《声学学报》2008年第6期512-519,共8页Acta Acustica

基　　金：国家重点基础研究规划基金项目(2006CB705707);国家自然科学基金项目(30570488，30671998);上海市重点学科建设项目(B112);上海曙光追踪计划(06GG03)。

摘　　要：血管内超声(IVUS)图像的仿真有助于检验诸如图像分割等图像处理算法的性能。提出一种IVUS图像仿真的方法。该方法在极坐标图像生成模型中引入环晕、导丝伪影,并分别对粥样硬化斑块的纤维、脂质、钙化三个区域进行模拟,实现静态图像的仿真;运用血管随心脏搏动的变化规律,实现序列图像的仿真。通过对15例真实IVUS图像、每例各50次的实验表明,相对于传统的极坐标图像生成模型,该方法仿真的图像与真实图像的相关系数提高了56.9%,互信息提高了24.3%,其仿真效果更加接近真实图像。It is helpful to use simulated intravascular ultrasound （IVUS） images during the validation of image processing algorithms such as the image segmentation. A method is proposed to simulate IVUS images. To accomplish the simulation of the static IVUS image, it introduces the ringdown and guidewire artifacts to the polar image-formation model, and simulates the fibrous, lipid and calcific content in the atherosclerotic plaque respectively. Utilizing characteristics of the pulsatile artery, the simulation of sequential IVUS images can also be achieved. The method is evaluated via 15 real IVUS images （with 50 experiments on each image）, and results demonstrate that it outperforms the traditional polar image-formation model by 56.9% in terms of the correlation coefficient of simulated and real images, and 24.3% in terms of the mutual information of simulated and real images.

关 键 词：超声图像 图像仿真 管内 图像处理算法 生成模型 图像分割 静态图像 变化规律 

分 类 号：TH776[机械工程—仪器科学与技术]