基于HARDI模型的脑纤维跟踪可视化研究

发布时间：2018-03-17 15:32

本文选题：脑纤维跟踪　切入点：像素条　出处：《浙江工业大学》2015年硕士论文　论文类型：学位论文

【摘要】：扩散张量成像(Diffusion Tensor Imaging,DTI)难以描述复杂的纤维微结构,而高角度分辨率扩散成像技术(High Angular Resolution Diffusion Imaging,HARDI)是一种刻画人脑复杂神经纤维走向的有效方法。并且,描述脑神经纤维交叉、汇聚、扩散的方向分布函数过于复杂,使可视化以及分析脑神经纤维产生困难。在展示脑神经纤维的三维空间内,仅通过键盘、鼠标等交互工具用户对脑神经纤维操作存在一定困难。其次,基于概率性跟踪的脑神经纤维结果存在冗余纤维过多的情况,一些错误的跟踪结果会遮挡或影响脑神经纤维的三维展示效果。本文提出一种新的可视化分析脑神经纤维的手段,用户通过使用一种新的可视化聚类方法来探索脑神经纤维束的可能走向。总结人脑神经纤维跟踪和可视化的研究背景以及国内外发展现状。对核磁共振成像技术、脑神经纤维跟踪技术以及脑神经可视化应用的研究进展进行论述,分析现有研究成果的发展过程及其存在的不足后,本文的具体工作和成果如下:1.针对DTI技术只能描绘体素内脑神经纤维单一走向的问题,本文使用HARDI模型刻画脑神经纤维的交叉、汇聚、分散等微结构,展现体素内脑神经纤维细致的走向和方向分布情况。其次,提出一种新的基于先前跟踪结果的先验概率计算方法,以解决概率性脑神经纤维跟踪过程中先前跟踪结果无法很好融入当前方向分布函数的问题。2.在概率性脑神经纤维跟踪结果的基础上,使用可视化工具pixel bar(像素条)来表示脑神经纤维的多维属性,并提供便捷直观的交互手段,为可视分析脑神经纤维提供依据。针对三维脑神经纤维空间内交互的困难,本文将脑神经纤维映射为pixel bar,提出脑神经纤维与pixel bar的映射方法,包括:方向-颜色映射算法,不透明度映射方法等。其次使用基于密度的DBSCAN聚类算法将pixel bar聚类,并根据不透明度进行簇内排序。用户通过选择一定不透明度范围内的pixel bar来获得相应置信区间内的脑神经纤维结果。该方法解决概率性跟踪结果冗余脑神经纤维过多,错误的跟踪结果遮挡或影响脑神经纤维空间展示效果的问题。3.本文采用Qt搭建基础平台,在OpengGL和OpenGLSL开发库的基础上,面向对象编程实现脑神经纤维跟踪可视化系统。主要实现概率性脑神经纤维跟踪过程中种子区域的选择,脑神经纤维三维展示时光照效果,线条反走样效果,显示帧率等功能。从专家用户的评价来看HARDI数据的概率性脑神经纤维跟踪结果能够真实还原脑神经纤维交叉、汇聚、扩散等复杂微结构。使用pixel bar可以有效灵活地在可视化分析中对脑神经纤维进行交互操作。
[Abstract]:It is difficult to describe complex fiber microstructures by diffusion Zhang Liang imaging Diffusion Tensor Imaging (DTI), and High Angular Resolution Diffusion Imaging (HARDI) is an effective method to describe the direction of complex nerve fibers in human brain. The distribution function of the direction of diffusion is too complicated to visualize and analyze the nerve fibers. Users of interactive tools such as the mouse have some difficulties in manipulating the nerve fibers. Secondly, the results of the probabilistic tracking of the brain nerve fibers result in excessive redundant fibers. Some wrong tracking results may block or affect the three-dimensional display effect of cerebral nerve fibers. This paper presents a new visual analysis method of cerebral nerve fibers. Users use a new visual clustering method to explore the possible direction of brain nerve fiber bundles, summarize the research background of human brain nerve fiber tracking and visualization, and the current situation of development at home and abroad. The research progress of brain nerve fiber tracking technology and the application of brain nerve visualization are discussed, and the development process of the existing research results and their shortcomings are analyzed. The specific work and results of this paper are as follows: 1. Aiming at the problem that DTI technology can only describe the single direction of brain nerve fibers in voxel, this paper uses HARDI model to describe the intersecting, converging, dispersing and other microstructures of cerebral nerve fibers. The distribution of cerebral nerve fibers in voxel is shown in detail. Secondly, a new priori probability calculation method based on previous tracking results is proposed. In order to solve the problem that the previous tracking results can not be well integrated into the current direction distribution function in the process of probabilistic nerve fiber tracking. 2. On the basis of the probabilistic nerve fiber tracking results, Using the visualization tool pixel bar (pixel bar) to express the multidimensional properties of the cerebral nerve fibers, and to provide a convenient and intuitive means of interaction, to provide the basis for visual analysis of the brain nerve fibers. In this paper, we map the nerve fibers to pixel bar.The mapping methods between the nerve fibers and pixel bar are presented, including direction-color mapping algorithm, opacity mapping method and so on. Secondly, the density-based DBSCAN clustering algorithm is used to cluster the pixel bar. According to the opacity, the user selects the pixel bar in a certain opacity range to obtain the corresponding results of the brain nerve fibers in the confidence interval. This method solves the problem of redundant nerve fibers in the probabilistic tracking results. Error tracking results block or affect the effect of spatial display of cerebral nerve fibers. 3. This paper uses QT to build a basic platform, based on OpengGL and OpenGLSL development libraries, Object-Oriented programming is used to realize the visualization system of nerve fiber tracking, which mainly realizes the selection of seed area in the process of probabilistic nerve fiber tracking, the illumination effect of three-dimensional display of brain nerve fiber, the effect of anti-aliasing of line, From the evaluation of expert users, we can see that the probabilistic nerve fiber tracking results of HARDI data can truly reduce the crossover and convergence of brain nerve fibers. The use of pixel bar can effectively and flexibly interoperate the nerve fibers in visual analysis.
【学位授予单位】：浙江工业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP391.41;R445.2

【相似文献】