自适应形变模型及其在血管内超声图像处理中的应用研究

【作者】 刘泽；

【导师】 孙丰荣；

【作者基本信息】 山东大学 ， 光学工程， 2006， 硕士

【摘要】 二十世纪九十年代中后期，Tim McInerney等学者结合形变模型(Deformable Models)和ACID图像分解(Affine Cell Image Decomposition)技术提出了自适应形变模型(Topologically Adaptable Snakes，T-Snake模型)概念。T-Snake模型是一个拓扑可变的形变模型，继承了Snake模型的诸多优点，但T-Snake模型更重要的特性是其每经过一个微小形变，便要进行一次拓扑变换，因而T-Snake模型相对于Snake模型的突出优点在于其受初始轮廓的影响小，而且能够根据图像的边缘信息自适应地改变其拓扑结构，因而对于复杂拓扑结构图像的处理和适应能力较强。近年来，T-Snake模型已逐渐成为相关学科的研究热点，并开始成功应用于边缘提取、图像分割、三维表面重建等领域的科研工作中，而研究的焦点是T-Snake模型内、外力结构的改进及其形变演化过程的数值实现。 血管内超声(Intravascular Ultrasound，IVUS)是对诸如动脉粥样硬化之类的冠状动脉疾病的诊断和评价的常用工具，它具有安全、实时和无损等优点。提取IVUS图像冠状动脉血管壁的边缘(下文将简称IVUS图像的边缘)并作进一步的图像分析处理，可以对血管腔的面积、血管厚度进行精确测量，还可以发现或识别早期冠状动脉粥样硬化斑块，并对斑块的形态结构实时显示，对斑块成分和偏心距等参数实时检测。所以，IVUS图像的边缘提取对冠状动脉疾病的诊断和治疗有着重要的临床意义。 本文首先介绍T-Snake模型的一般理论，然后描述对T-Snake模型所作的改进，并给出基于改进的T-Snake模型的IVUS序列图像内外膜边缘提取算法。同时，我们将改进的二维T-Snake模型进一步推广到三维情况，并给出了利用改进的三维T-Snake模型对IVUS图像进行三维表面重建的算法。实验结果表明，与以往同类研究报道的方法相比，本文边缘提取方法准确性和可靠性较高，对IVUS序列图像处理的可重复性和鲁棒性较好，而IVUS图像三维表面重建的重建速度及重建的准确性和可靠性也较高。另外，实验结果也表明了本文改进的T-Snake模型的有效性和可实现性。更多还原

【Abstract】 In the later 1990s, Tim McInerney and other scholars proposed the method of auto-adapted deformable model (Topological Adaptable Snakes, the T-Snake model), which is unified the technology of deformable models and ACID (Affine Cell Image Decomposition). The T-Snake model is a deformable model with invariable topology, it has inherited many merits of the Snake models, but which is more important is that the T-Snake model will changes its topology after a small deformation, thus the influence of the initial outline is weaker to the T-Snake model than to the snake models, which is more important is that the T-Snake model can auto-adapted change its topology according to edge information of the image, so the T-Snake model is strong regarding complex topology image processing and adaptiveness. In recent years, the T-Snake model has gradually become the research hot spot of the correlation domain, such as image division, three dimension reconstruction, edge detection and so on. And now scholars main research the structure of the internal and external force and the deformation evolutionary process value realization of the T-Snake.Intravascular Ultrasound (IVUS) is one kind of diagnosis and appraisal method of the scleratheroma, it has the merits of safe, real-time, no damages and so on. Detecting border for Intravascular ultrasound images and doing the further analysis and processing can not only carry on accurate measurement to the area of of blood vessel, and can also find the early scleratheroma plaques of coronary artery, it can also show the shape of the plaques and measure such parameters as composition of the plaque and centrifugal distance, wall thickness in real time, more important, images after the 3D reconstruction can recognize the early scleratheroma plaques, discover the morphological structure of the plaques in real time, and detect the ingredient and eccentricity of the plaques. So the 3D reconstruction of Intravascular ultrasound images更多还原