基于高斯混合噪声和随机模型的冠脉造影数据仿真  

Simulation of Coronary Angiographic Data Based on Hybrid Gaussian Noise and Stochastic Model

作　　者：杨俊[1] 尚鹏[2] 周寿军[2] 郑曲波[1] 吴桂良[1] 

机构地区：[1]中国人民解放军第四五八医院,广州510602 [2]中国科学院深圳先进技术研究院,深圳518055

出　　处：《系统仿真学报》2013年第11期2655-2661,共7页Journal of System Simulation

基　　金：国家自然科学基金委面上项目(61179020);青年基金项目(60902103;31000450)

摘　　要：提出了一种冠状动脉CT血管造影图像数据的仿真方法。该方法能在三维空间模拟血管目标和背景,可以为血管造影图像的分析计算提供一种仿真数据。方法的主要内容是:管状目标的常规形态和特异形态的模拟;多参数高斯混合噪声背景的产生;基于Markov随机场的背景纹理生成;基于随机路径模型的心脏左冠状动脉树模拟;目标/背景数据合成与比例控制。由此可以生成多变的血管结构和复杂的背景。实验中利用上述方法有效模拟了多种典型血管形态及其狭窄结构,并有效生成了左冠状动脉树及其血管造影仿真数据。A method to simulate the data of coronary CT angiographic image was proposed. The method could model vesscular object and its background within three-dimensional space so as to provide a digital phantom for analysis and computation of angiographic image. The main contents of this method include： Simulations of the normal and specific tubular forms; production of noise background with multi-parameters gaussian mixture model; production of texture background based on Markov Random Field; Simulation of left coronay artery tree based on stochastic path generator; proportion control on the synthetical data of vessel and background. Under the aforementioned technologies, varied vascular structures and complex background could be simulated. In experiments, various natural vessel structure as well as injury one （e.g. stenosis） was simulated, so as to effectively produce left coronary artery tree and its angiographic phantom.

关 键 词：血管仿真 高斯混合模型 MARKOV随机场 CT冠状动脉造影 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]