基于ICA成分相位信息的复数fMRI数据分析

发布时间：2018-04-29 08:24

本文选题：复数fMRI数据 + ICA　；参考：《大连理工大学》2014年硕士论文

【摘要】：功能磁共振成像(function Magnetic Resonance Imaging, fMRI)是一种重要的脑功能成像技术。通过对采集到的fMRI数据进行独立分量分析(independent component analysis,ICA),可以获取脑认知所需的空间成分及其时间成分。完整的fMRI数据是复数,但由于fMRI相位数据噪声严重且特性未知,人们通常只分析其幅值数据(即实数fMRI)。然而,越来越多的证据表明,相位数据含有独特的脑功能信息,对其进行有效的利用,有助于揭示更为完整的脑功能信息。目前,复数fMRI数据的ICA分析采用了预处理消噪法,但存在信息损失问题；ICA后处理方法也不够完善。为此,本文提出一种能够充分利用ICA估计成分相位信息的后处理分析框架,具体内容如下： (1)针对复数ICA固有的相位模糊问题,提出了基于时间成分的相位模糊矫正方法。通过时间成分和空间成分的非环形度对比,以及对激活区体素相位原理和fMRI时空关系的分析,提出了时间成分实部能量最大化的相位模糊矫正准则,并利用标准时间序列或空间参考信息去除了矫正过程中的符号错误,成功解决了相位模糊问题。实验结果表明,本文基于时间成分的相位模糊矫正方案具有鲁棒性好、准确率高的优点； (2)针对相位噪声对ICA空间成分激活区的严重干扰问题,提出了相位定位思想和相位掩蔽方法,以及将相位范围和幅值强度相结合的空间可视化方案。相位定位利用ICA空间成分的相位信息区分有用体素和干扰体素,相位掩蔽法利用相位定位mask去除ICA空间成分中的干扰体素,提取激活区域。最后,结合体素的幅值强度信息,精确提取激活区域。实验结果表明,这些方法具有去噪性能好、激活区提取准确度高等优点； (3)利用本文提出的复数fMRI数据分析框架,对运动刺激下的主要脑功能网络进行了提取以及交互作用分析,验证了本文框架的有效性和普适性,并为复数fMRI数据的功能连接工作提供了有效的支持。
[Abstract]:Functional magnetic resonance imaging (function Magnetic Resonance Imaging (fMRI)) is an important brain functional imaging technique. By means of independent component analysis (independent component analysis, ICA) of the collected fMRI data, the spatial components and time components needed for brain cognition can be obtained.
The complete fMRI data is plural, but because of the serious noise and unknown characteristics of the fMRI phase data, people usually only analyze its amplitude data (i.e. real fMRI). However, more and more evidence shows that the phase data contains unique brain function information, and the effective use of it can help to reveal more complete brain function information. The ICA analysis of fMRI data adopts the preprocessing denoising method, but there is a problem of information loss, and the post-processing method of ICA is not perfect. Therefore, this paper proposes a post processing analysis framework which can make full use of ICA to estimate the phase information of the component. The specific contents are as follows:
(1) in view of the inherent phase ambiguity of complex ICA, a method of phase fuzzy correction based on time component is proposed. By comparing the non ring degree of the time component and the space component, and the analysis of the phase principle of the voxel in the active region and the analysis of the time and space relation of the fMRI, the phase fuzzy correction criterion of the time component real energy maximization is put forward, and the benefit of the phase fuzzy correction is put forward. The standard time sequence or spatial reference information is used to remove the symbol errors in the correction process, and the phase ambiguity problem is successfully solved. The experimental results show that the phase fuzzy correction scheme based on the time component has the advantages of good robustness and high accuracy.
(2) in view of the serious interference of phase noise to the active area of ICA space components, the phase location idea and phase masking method are proposed, and the space visualization scheme combining phase range and amplitude intensity is proposed. Phase location uses phase information of ICA space component to distinguish between voxel and interferin, phase masking method uses phase Mask removes the disturbing voxel in the ICA space components and extracts the active region. Finally, the activation region is extracted accurately with the amplitude intensity information of the voxel. The experimental results show that these methods have the advantages of good denoising performance and high accuracy in the activation area.
(3) using the complex fMRI data analysis framework proposed in this paper, the main brain function network under the motion stimulus is extracted and the interaction analysis is used to verify the validity and universality of the framework, and provide an effective support for the functional connection of the complex fMRI data.

【学位授予单位】：大连理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：R445.2;TN911.7

【参考文献】