基于血管内超声成像的虚拟血管三维重建

发布时间：2018-03-20 23:44

本文选题：血管内超声　切入点：冠状动脉造影　出处：《哈尔滨工业大学》2014年硕士论文　论文类型：学位论文

【摘要】：血管内超声(Intravascular Ultrasound,IVUS)与冠状动脉造影(Coronary angiography,CAG)在冠状动脉疾病的临床诊断上起着重要的作用,且应用最为广泛。血管内超声成像能够显示血管的内部信息,如管腔的横截面信息,斑块的形态、分布及组成等。X射线冠脉造影显示的是血管的空间几何信息,通过观察血管外部形态来确定血管的病变部位。因此通过将两种方法进行融合,能够使它们自身的优势互补,得到的虚拟三维血管模型不仅能够形象而准确的显示血管的空间几何信息,同时还能显示感兴趣血管段的内部信息。早先的研究主要是针对机械旋转型探头获取的IVUS图像序列,图像随着导引丝的回撤会出现扭曲,而现有IVUS图像内外膜分割方法效果不是很理想,导致在和CAG图像融合时复杂且误差大,重建出来的三维血管只能看到血管空间几何信息和病变的大概部位,并不能观测斑块的种类以及斑块的形态。在此基础上本文针对电子相控阵系统获取的IVUS图像序列、超声信号分析得到斑块图像开展了与CAG图像融合的研究,完成斑块图像和三维血管的融合,实现虚拟血管的三维重建,并对IVUS图像的内外膜分割方法进行了改进。本文的研究内容主要包括:IVUS图像的内外膜分割、CAG图像血管中心线的三维重建、冠状动脉血管模型的三维重建、斑块分类图像与血管超声图像的融合、斑块图像与血管三维模型的融合即虚拟血管的三维重建。本文利用图像锐化和中值滤波去除超声图像上的噪声和刻度,并在此基础上实现基于灰度和纹理的血管内外膜自动分割。由于使用的IVUS图像不存在导引丝回撤时扭曲的现象,所以仅需提取血管的中心线,完成血管中心线的三维重建,作为血管内超声图像获取时的回撤路径。并将IVUS图像序列在最佳角度下,准确定位在三维血管中心线上,同时对重构出的血管表面利用非均匀有理B样条的方法进行平滑处理,得到血管的三维模型。最后利用RF信号频谱分析得到的斑块分类图像与超声图像融合实现图像信息匹配,并使斑块图像与重建的三维模型融合,准确的重构出了虚拟感兴趣冠状动脉血管的三维模型,该模型准确的再现了血管的内部和外部形态,在临床上对辅助医生诊断冠状动脉疾病有重要的意义。
[Abstract]:Intravascular ultrasound (IVU) and coronary angiography (CAG) play an important role in the clinical diagnosis of coronary artery disease. The morphology, distribution and composition of plaques. X-ray coronary angiography shows the spatial geometric information of blood vessels, and determines the location of the lesions by observing the external morphology of the vessels. Therefore, the two methods are fused. The virtual 3D vascular model can not only visualize and accurately display the spatial geometric information of blood vessels, but also make them complement each other. Earlier studies focused on IVUS image sequences obtained by mechanically rotating probes, which were distorted as the guiding wire retreated. However, the existing internal and external membrane segmentation methods of IVUS images are not very satisfactory, which leads to the complexity and error of fusion with CAG images. The reconstructed 3D vessels can only see the geometric information of the vascular space and the general location of the lesions. The types and morphology of plaques can not be observed. On the basis of this, the image sequences of IVUS obtained by electronic phased array system are analyzed by ultrasonic signal and fused with CAG images. To complete the fusion of plaque image and three-dimensional blood vessels, realize the three-dimensional reconstruction of virtual blood vessels. The internal and external membrane segmentation method of IVUS image is improved. The main contents of this paper include the 3D reconstruction of vascular center line of IVUS image and the 3D reconstruction of coronary artery model. The fusion of plaque classification image and vascular ultrasound image, the fusion of plaque image and vascular three-dimensional model is the 3D reconstruction of virtual blood vessel. In this paper, the noise and scale of ultrasonic image are removed by image sharpening and median filtering. On this basis, the automatic segmentation of vascular inner and outer membrane based on grayscale and texture is realized. Because there is no distortion of guiding wire in the IVUS image, it is only necessary to extract the center line of blood vessel and complete the 3D reconstruction of the center line of blood vessel. As the retreat path of intravascular ultrasound image acquisition, the IVUS image sequence is accurately located on the three-dimensional vascular center line at the best angle, and the reconstructed blood vessel surface is smoothed by non-uniform rational B-spline method. Finally, using RF signal spectrum analysis, the plaque classification image and ultrasonic image are fused to achieve the image information matching, and the plaque image and the reconstructed 3D model are fused. The three-dimensional model of virtual interested coronary artery was reconstructed accurately. The model accurately reproduced the internal and external morphology of the vessel. It is of great significance to assist doctors in the diagnosis of coronary artery disease.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：R445.1;TP391.41

【参考文献】