小波变换在静息态功能数据分析中的应用

发布时间：2018-03-31 14:18

本文选题：功能磁共振成像　切入点：静息态　出处：《西安电子科技大学》2012年硕士论文

【摘要】：功能磁共振成像(functional Magnetic Resonance Imaging, fMRI)技术因其拥有较高的空间分辨率、时间分辨率、无辐射损伤以及可在活体上重复进行检测等优势，已成为当今用于脑功能研究的主要技术。近年来的研究表明人脑在静息态时仍然存在着明显的活动，能否快速、准确的获得这些活动的过程、规律，静息态fMRI数据分析方法起着至关重要的作用。以往大多数基于静息态脑功能连接度的研究都要假设时不变‖，例如相关分析和一些数据驱动的方法。然而基于任务的研究和动物电气生理学研究都表明功能连接存在动态、瞬时的变化。基于此，本文引进了一种时频特性方法——小波变换来探索静息态功能连接的这种动态变化行为。本文的主要研究区域有后扣带回(posterior cingulated cortex, PCC)、静息态默认网络中一些重要结点脑区以及激活较为明显的一些负相关区域。本文首先对静息态功能磁共振成像数据处理流程做了详细的介绍，包括数据预处理和一些常用的计算；其次介绍了静息态功能连接性常用分析方法，包括局部一致性分析、种子点相关分析、独立成分分析、时间聚类分析；再次用小波变换方法分析静息态fMRI数据，分别做了连续小波变换、正交小波变换以及相干小波变换。最终三种变换的结果都表明大脑在静息态时的主要活动频率在1/32Hz~1/16Hz,这与以往的研究结果吻合。相干小波变换的结果表明在1/4Hz时PCC与其它10个脑区的变化最为相似。另外，除了以往的静息态默认网络外，PCC与R.DLPFC、R.insula、R.SMG等几个负相关比较大的区域也有明显的相似性。
[Abstract]:Functional Magnetic Resonance imaging (fMRI) has the advantages of high spatial resolution, temporal resolution, no radiation damage and repeated detection in vivo. Has become the main technology for brain function research. Recent studies have shown that there are still obvious activities in the resting state of the human brain, whether the processes and rules of these activities can be obtained quickly and accurately. The static fMRI data analysis method plays an important role. Most previous studies based on resting brain functional connectivity have assumed that time is constant. For example, correlation analysis and some data-driven methods. However, task-based studies and animal electrophysiological studies have shown that functional connections have dynamic, instantaneous changes. In this paper, a time-frequency characteristic method, wavelet transform, is introduced to explore the dynamic behavior of static functional connections. The main research areas in this paper are posterior cingulate cingulated cortexes, PCCs, and some important nodes in resting default networks. Brain area and activation of some of the more obvious negative correlation areas. In this paper, the data processing flow of resting functional magnetic resonance imaging (fMRI) is introduced in detail, including data preprocessing and some commonly used calculations, and then the commonly used analysis methods of resting functional connectivity, including local consistency analysis, are introduced in detail. Seed point correlation analysis, independent component analysis, time clustering analysis, wavelet transform method is used to analyze resting fMRI data. The results of orthogonal wavelet transform and coherent wavelet transform show that the main frequency of brain activity in resting state is 1 / 32 Hz / 1 / 16 Hz, which is consistent with previous results. The results of coherent wavelet transform show that PCC in 1/4Hz. Most similar to the other 10 brain regions. In addition, In addition to the previous resting default networks, there are obvious similarities between PCC and R. DLPFCU R. Peninsula R. SMG and other regions with relatively large negative correlation.
【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：R310

【参考文献】