心脏弥散张量磁共振图像弹性配准的研究

发布时间：2018-05-26 16:28

本文选题：弥散张量 + 磁共振　；参考：《浙江大学》2014年硕士论文

【摘要】：生物体的组织结构会对其中的水分子弥散运动造成影响,反过来水分子的弥散运动反映了所处位置的组织特性。弥散张量磁共振成像(DT-MRI)在磁共振成像的基础上利用多于六个的非线性相关方向梯度磁场检测组织内水分子在这些方向上的弥散速率,在此基础上计算出水分子的弥散张量,用以研究这些组织内部的精细结构。当前DT-MRI技术被广泛地应用于研究脑白质神经纤维,而较少涉及心脏领域。心脏疾病的高发使得对心脏结构特征与疾病的发病机理的研究变得越发迫切。与脑图像刚性特征不同,心脏处于永不停歇的运动状态,这使得利用心脏图像进行疾病的诊断严重依赖于医生的经验和专业知识,所得结果也有较大误差。为更好地对比不同心脏弥散张量图像所反映的信息从而为诊断提供强有力的支撑,就需要对心脏弥散张量进行配准,从而使其达到空间对准。心脏的收缩、舒张是一个复杂的不规则运动,这种不规则性使处于不同状态的两个心脏之间存在较大的形变。单纯利用参数模型,如刚性变换、仿射变换、投影变换都不能很好地描述两者之间的这种差异。心脏各部位的形变程度不同使得对心脏图像的配准需要分解为局部区域。为此,本文利用基于B样条的自由变换模型来对局部差异进行描述,得到局部范围内的形变场,通过挨个地进行局部配准达到最终的配准目的。为了在配准精度与速度之间取得平衡,本文使用金字塔模型将整个的配准分割为一个由粗到精的过程,提高了配准的精度和速度。得到配准结果后,使用保持主向量方向PPD策略对张量进行重定向以使张量主向量的方向与所在位置的周围组织方向保持一致。本文使用这种方法对心脏的二维弥散张量图像进行了配准获得了比较好的结果,并为将来进一步的研究打下了基础。
[Abstract]:The tissue structure of the organism will affect the diffusion of water molecules, which in turn reflects the tissue characteristics of the water molecules. Diffusion Zhang Liang Magnetic Resonance Imaging (DT-MRI) on the basis of magnetic resonance imaging, the diffusion rate of water molecules in these directions is measured by using more than six nonlinear correlation gradient magnetic fields, and the diffusion Zhang Liang of water molecules is calculated on this basis. Used to study the fine structure of these tissues. At present, DT-MRI technique is widely used to study white matter nerve fibers, less involved in the field of heart. The high incidence of heart disease makes it more urgent to study the structure and pathogenesis of heart disease. Unlike the rigid features of brain images, the heart is in a state of unceasing motion, which makes the diagnosis of diseases by using cardiac images rely heavily on the experience and professional knowledge of the doctors, and the results also have a large error. In order to better compare the information reflected in different diffused Zhang Liang images and provide strong support for diagnosis, it is necessary to register the cardiac diffused Zhang Liang to achieve spatial alignment. The contraction and relaxation of the heart is a complex irregular movement, which causes a large deformation between two hearts in different states. Simply using parametric models such as rigid transformation affine transformation and projection transformation can not well describe the difference between them. The degree of deformation in different parts of the heart makes the registration of the heart image need to be decomposed into local regions. In this paper, a free transformation model based on B-spline is used to describe the local difference, and the deformation field in the local range is obtained, and the final registration is achieved by local registration one by one. In order to achieve a balance between registration accuracy and speed, the pyramidal model is used to divide the whole registration into a coarse to fine process, which improves the accuracy and speed of registration. After the registration results are obtained, the Zhang Liang is redirected by using the PPD strategy to keep the direction of the principal vector so that the direction of the principal vector of the Zhang Liang is consistent with the direction of the surrounding tissue in the position. In this paper, this method is applied to the registration of two-dimensional diffused Zhang Liang images of the heart, and a good result is obtained, which will lay a foundation for further research in the future.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：R445.2

【参考文献】