• No results found

Test-Retest Reliability and Its Computation of Resting State Functional Magnetic Resonance Imaging

N/A
N/A
Protected

Academic year: 2021

Share "Test-Retest Reliability and Its Computation of Resting State Functional Magnetic Resonance Imaging"

Copied!
17
0
0

Loading.... (view fulltext now)

Full text

(1)

学校编码:10384 分类号 密级

学号:31520080150544 UDC

博 士 学 位 论 文

人脑静息态功能磁共振成像的重测及其计算分析

Test-Retest Reliability and Its Computation of Resting

State Functional Magnetic Resonance Imaging

陈兵

指导教师姓名: 翁 旭 初 教 授

专 业 名 称 : 人 工 智 能 基 础

论文提交日期: 2 0 1 5 年 4 月

论文答辩日期: 2 0 1 5 年 5 月

学位授予日期:

年 月

答辩委员会主席:

评 阅 人:

2015 年 4 月

厦门大学博硕士论文摘要库

(2)

申请厦门大学博士学位论文

人脑静息态功能磁共振成像的重测

及其计算分析

作者:陈兵

指导教师:翁旭初教授

专业:人工智能基础

信息科学与技术学院

2015 年 4 月

厦门大学博硕士论文摘要库

(3)

厦门大学学位论文原创性声明

本人呈交的学位论文是本人在导师指导下,独立完成的研究

成果。本人在论文写作中参考其他个人或集体已经发表的研究成

果,均在文中以适当方式明确标明,并符合法律规范和《厦门大

学研究生学术活动规范(试行)》

另外,该学位论文为(

)课

题(组)的研究成果,获得(

)课题(组)经费

或实验室的资助,在(

)实验室完成。(请在以

上括号内填写课题或课题组负责人或实验室名称,未有此项声明

内容的,可以不作特别声明。)

声明人(签名):

月 日

厦门大学博硕士论文摘要库

(4)

厦门大学学位论文著作权使用声明

本人同意厦门大学根据《中华人民共和国学位条例暂行实施办

法》等规定保留和使用此学位论文,并向主管部门或其指定机构送交

论文(包括纸质版和电子版),允许论文进入厦门大学图书馆及其数

据库被查阅、借阅。本人同意厦门大学将学位论文加入全国博士、硕

士学位论文共建单位数据库进行检索,将学位论文的标题和摘要汇编

出版,采用影印、缩印或者其他方式合理复制。

本学位论文属于:

) 1 、 经 厦 门 大 学 保 密 委 审 查 核 定 的 保 密 论 文 , 于

年 月

日解密,解密后适用上述授权。

)2、不保密,适用上述授权。

(请在以上相应括号内打“√”或填上相应内容。保密学位论文

应是已经校保密委审定过的,方可打“√”

,未经审批均为公开论文。

此声明栏不填写的,默认为公开论文,均适用上述授权。

声明人(签名):

年 月 日

厦门大学博硕士论文摘要库

(5)

摘要 I

摘要

功能连接组(Functional connectomes)即人脑内部功能连接的全体。目前功 能连接组指标已成为人脑功能研究中的重要参考依据。在人脑发育和疾病研究中 出现了越来越多的方法用于关注脑功能的变化。由于具有不受任务设计限制的优 势,因此静息态功能磁共振(rfMRI)成为人脑功能连接组学研究的主要方法之一。 静息态功能磁共振图像的计算指标主要包括功能分化和功能整合两部分。鉴于开 发具备高重复性的脑功能连接组学计算方法对研发基于脑成像的人脑功能影像 学标记物重要性,静息态脑功能指标的变异和重测研究就成为了探索人脑活动变 化的重要参考工具。 本文以静息态人脑功能连接组为研究对象,以重测信度和指标动态变化为主 线。围绕静息态功能磁共振数据的重测周期、数据采集、质量控制、计算指标加 以发展和创新,采集更丰富的纵向重测数据,用以更准确的反应人脑功能连接组 指标随时间的变异和重测信度。本文主要研究内容由以下三部分组成。 第一,针对目前静息态指标的重测研究存在着持续周期过短、重测次数不足、 重测样本量不够、重测时时间不一致等问题。本文从实验设计的角度,提出了在 约 1 个月的时间内对 30 名实验参与者进行了大致间隔 3 天的 10 次重测,在每次 重测时,尽量保证参与者的扫描时间一致,降低不同重测时间段造成的静息态指 标差异。

第二,论文采用功能连接组计算系统(Connectome Computation System, CCS) 的数据质量控制模块,对静息态功能磁共振重测数据处理中的每一步骤的中间结 果进行监控和校正,较好的保证了结构像和功能像的图像质量,确保了数据的可 靠性。在静息态功能磁共振指标计算方式上采用 2D 结构,充分考虑了大脑皮层 沟回的形状结构,这一办法较以往的 3D 方式计算更准确的反应了大脑的功能组 织特点。 第三,在静息态功能连接组指标的变化上,本研究依次从全脑水平到网络水 平、认知成分、顶点分别测量了 1 个月内实验参与者的被试内变异、被试间变异、 重测信度,较以往的静息态重测研究计算了更全面的功能连接组指标的变化情

厦门大学博硕士论文摘要库

(6)

摘要 II 况。 本研究在皮层结构上从功能分化和功能整合的角度分别计算了静息态功能 连接组指标的变异性和重测信度,为全面了解人脑静息态功能连接组的指标变化 情况提供了数据支持和指标参考,弥补了静息态重测中缺乏短时间内大样本多次 重测的数据空白。 论文系统地研究了不同网络和认知成分在人脑静息态功能网络拓扑属性的 变异程度和重测信度。结果表明:人脑功能网络的重测可靠性在功能分化指标上 具有中等程度的可靠性,但在功能整合指标方面则表现较低。各网络和认知成分 的可靠性在 6 个度量指标(ALFF, fALFF, ReHo, ReHo2, DCw, ECw)之间表现出 较大的变异。在视觉、体感运动、额顶控制、默认网络部分功能分离指标的重测 信度大致为 0.5。而在注意网络的重测信度则表现较低,类内相关系数小于 0.4。 各网络指标的被试间变异小于被试内变异,在注意和体感运动网络的被试内和被 试间变异均较小。通过对 12 个认知成分的指标重测观察,我们了解到对于听觉 和默认这两种成分,具有中等程度的重测信度,约为 0.5。最后我们对 6 个计算 指标采用了层次聚类方法进行分析,发现 ALFF 和 fALFF 为一类,ReHo 和 ReHo2 为一类。以上两类组成的新类与 DCw 和 ECw 组成的功能整合指标的类别差异 较大。这些研究发现为研究者选取稳定可靠的脑网络和认知成分的功能连接组指 标提供了重要的参考。

关键字:功能连接组;静息态;功能分化;功能整合;重测

(7)

Abstract

III

Abstract

Functional connectomes is a comprehensive map of functional interaction or functional connectivity of the brain. Among the emerging methods, resting state functional magnetic resonance imaging (rfMRI) has evolved as a major approach to address functional connectomes, in which data acquisition is independent of the experiment tasks.. The metrics of rfMRI, such as functional connectomes, measure the segmentation and integration of brain functions. As functional connectomes are believed to be a potential biomarker for normal brain and its disorders, their variability and test-retest reliability are vital to explore the dynamics of the human brain.

This thesis covers the test-retest reliability and variability of functional connectomes. By refining test-retest interval, data acquisition procedure, quality control technology and measurements, the author acquired a large set of longitudinal data. The research encounter made the functional connectomes metrics more valid to assess temporal stability and variance of rfMRI. The project extended previous studies from three perspectives.

(1) Research Design: The majority of previous studies of test-retest reliability have a number of weaknesses, such as short testing intervals, few retesting times, small sample size, and inconsistent testing time slots, while the present research took account of and minimized all of these potential confounds. In particular, thirty subjects underwent ten fMRI scans in one month by three days interval. The time slots of each subject were fixed.

(2) Data Quality: Control The Data Quality Control Module of Connectome Computation System (CCS) was used to monitor the intermediate process which assured the quality of structural and functional images. To better characterize the anatomical and functional organizations, the author calculated rfMRI metrics in 2D way which was more precise in depicting gyri and sulci than 3D approach.

(8)

Abstract

IV

(3) Functional Connectomes Metrics: Based on the month wide intra-/inter- individual dataset, the author estimated the variability and reliability of functional connectomes metrics across whole brain, sub-networks, cognitive functions and vertices, allowing for drawing a complete picture of functional connectomes dynamics than previous rfMRI studies. The main findings of this study as follows. The functional connectomes metrics of functional segmentation were medium reliable. However, the reliabilities of functional integration metrics were low. The reliabilities of six metrics (ALFF, fALFF, ReHo, ReHo2, DCw, ECw) varied in different sub-networks and cognitive functions. They were around 0.5 in the visual, somato-motor, frontal-parietal control, and default model network, but lower than 0.4 in the attention network. The metrics varied less between subjects than within subjects. The variabilities of both intra-/inter-individuals were low in attention and somato-motor networks. Further, the reliability test of cognitive functions showed that the metrics of auditory and default model network were around 0.5, which were medium stable. Finally, the author conducted hierarchical clustering for the metrics and identified two clusters. ALFF and fALFF were in the same group, while ReHo and ReHo2 were clustered together. These clusters were significantly different from functional integration cluster of DCw and ECw. These findings provided important references for selecting functional connectomes metrics of brain networks and cognitive functions.

Keywords: Functional Connectomes; Resting-state; Functional Segregation; Functional Integration; Test-Retest

厦门大学博硕士论文摘要库

(9)

目录 V

目录

第一章 绪论 ... 1

1.1 研究现状 ... 1 1.1.1 静息态功能磁共振简介 ... 1 1.1.2 静息态功能磁共振重测信度和变异研究意义 ... 3 1.1.3 静息态功能磁共振重测信度和变异研究现状 ... 4 1.2 本文研究问题和意义 ... 11 1.2.1 研究问题 ... 11 1.2.2 研究意义 ... 12 1.3 本文主要工作 ... 13 1.4 本文组织结构 ... 13

第二章 静息态功能磁共振重测研究简介 ... 15

2.1 静息态功能磁共振方法学研究 ... 15 2.1.1 静息态功能分离 ... 16 2.1.2 静息态功能整合 ... 17 2.2 静息态功能整合中常见的分析方法 ... 18 2.2.1 独立成分分析方法 ... 18 2.2.2 聚类方法 ... 19 2.2.3 模式识别方法 ... 20 2.3 重测信度研究 ... 20 2.3.1 重叠率度量重测信度 ... 21 2.3.2 一致性相关系数度量重测信度 ... 22 2.3.3 变异系数度量重测信度 ... 22 2.3.4 静息态功能磁共振的重测信度研究 ... 23 2.3.5 静息态功能磁共振的变异研究 ... 24 2.4 人脑皮层网络和认知成分的模块划分 ... 25

厦门大学博硕士论文摘要库

(10)

目录 VI 2.4.1 人脑皮层网络划分 ... 25 2.4.2 人脑认知成分的模块划分 ... 26 2.5 静息态数据共享和计算平台 ... 28 2.6 本章小结 ... 33

第三章 rfMRI 重测数据采集和质量评估 ... 34

3.1 实验数据采集 ... 34 3.1.1 实验参与者的选择 ... 34 3.1.2 核磁共振设备参数设置和采集序列 ... 34 3.1.3 扫描相关数据的记录 ... 35 3.2 数据质量评估 ... 36 3.2.1 结构像数据质量评估 ... 36 3.2.2 功能像数据质量评估 ... 39 3.3 本章小结 ... 41

第四章 rfMRI 重测数据处理过程 ... 42

4.1 数据处理步骤 ... 42 4.2 数据预处理平台 ... 43 4.2.1 硬件环境 ... 43 4.2.2 软件环境 ... 43 4.2.3 数据准备 ... 43 4.3 结构像预处理 ... 44 4.3.1 结构像降噪 ... 44 4.3.2 去除非脑组织 ... 45 4.3.3 大脑皮层重建与脑组织分割 ... 48 4.3.4 结构像配准 ... 49 4.4 功能像预处理 ... 49 4.4.1 功能像信号基本处理 ... 49 4.4.2 功能像的配准与分割 ... 50 4.4.3 功能像去噪 ... 50

厦门大学博硕士论文摘要库

(11)

目录 VII 4.4.4 滤波、去时间趋势与平滑 ... 51 4.5 静息态指标变异和重测计算 ... 51 4.6 本章小结 ... 55

第五章 静息态下人脑活动区域指标的变异和重测研究 ... 56

5.1 区域指标数据后处理 ... 56 5.1.1 低频振幅方法 ... 56 5.1.2 比值低频振幅方法 ... 57 5.2 区域指标数据后处理结果 ... 58 5.2.1 低频振幅结果 ... 58 5.2.2 比值低频振幅结果 ... 61 5.3 本章小结 ... 63

第六章 静息态下人脑活动短距离功能连接指标的变异和重测 ... 64

6.1 短距离功能连接数据后处理 ... 64 6.2 短距离功能连接数据后处理结果 ... 66 6.2.1 局部一致性 ReHo 结果 ... 66 6.2.2 局部一致性 ReHo2 结果 ... 68 6.3 本章小结 ... 71

第七章 静息态下人脑活动长距离功能连接指标的变异和重测 ... 72

7.1 长距离功能连接数据后处理 ... 72 7.1.1 基于种子点功能连接的分析 ... 72 7.1.2 图论方法 ... 73 7.1.3 加权的度中心度 ... 75 7.1.4 加权的特征向量中心度 ... 75 7.2 长距离功能连接数据后处理结果 ... 76 7.2.1 加权度中心度结果 ... 76 7.2.2 加权特征向量中心度结果 ... 78 7.3 本章小结 ... 81

厦门大学博硕士论文摘要库

(12)

目录 VIII

第八章 rfMRI 各水平的变异、重测和聚类研究 ... 82

8.1 全脑活动指标的变异和重测研究 ... 82 8.2 顶点水平指标的变异和重测研究 ... 83 8.3 七个网络条件下全脑指标的变异和重测研究 ... 85 8.4 十二个认知成分上全脑指标的变异和重测研究 ... 87 8.5 全脑指标变异和重测的聚类研究 ... 90 8.6 本章小结 ... 93

第九章 总结与展望 ... 94

9.1 主要研究工作 ... 94 9.2 主要创新点 ... 95 9.3 工作展望 ... 96

参考文献 ... 98

攻读学位期间发表的科研成果及参与项目 ... 110

致谢 ... 112

附录中英文缩略词表 ... 113

厦门大学博硕士论文摘要库

(13)

Contents IX

Contents

Chapter 1 Introduction ... 1

1.1 Literature Review ... 1 1.1.1 Introduction to rfMRI ... 1

1.1.2 Significance of Test-Retest Reliability and Variability of rfMRI ... 3

1.1.3 Literature Review of Test-Retest Reliability and Variability of rfMRI 4 1.2 Research Goals and Significance ... 11

1.2.1 Research Goals. ... 11

1.2.2 Research Significance ... 12

1.3 Research Overview ... 13

1.4 The Roadmap of the Dissertation ... 13

Chapter 2 rfMRI and Its Test-Retest Reliability ... 15

2.1 Methodology of rfMRI ... 15

2.1.1 Functional Segmentation of rfMRI ... 16

2.1.2 Functional Integration of rfMRI ... 17

2.2 Research Methods of Functional Integration ... 18

2.2.1 Independent Component Analysis ... 18

2.2.2 Clustering ... 19

2.2.3 Pattern Recognition ... 20

2.3 Test-retest Reliability ... 20

2.3.1 Reliability of Overlapping ... 21

2.3.2 Reliability of Concordance Correlation Coefficient ... 22

2.3.3 Reliability of Coefficient of Variation ... 22

2.3.4 Test-retest Reliability of rfMRI ... 23

2.3.5 Metrics Variability of rfMRI ... 24

2.4 Brain Network and Cognitive Modularities ... 25

(14)

Contents

X

2.4.1 Brain Network ... 25

2.4.2 Cognitive Modularities ... 26

2.5 Data Sharing and Computing Platform of rfMRI ... 28

2.6 Summary ... 33

Chapter 3 Data Collection and Quality Evaluation of rfMRI

Test-Retest Reliability ... 34

3.1 Data Acquisition ... 34

3.1.1 Subject Choose ... 34

3.1.2 fMRI Scanning Parameters & Sequences ... 34

3.1.3 Record about Scanning ... 35

3.2 Data Quality Assessment ... 36

3.2.1 QA of Structural Images ... 36

3.2.2 QA of Functional Images ... 39

3.3 Summary ... 41

Chapter 4Test-Retest Reliability of Human Brain Activity in rfMRI

and its Computation ... 42

4.1 Processing Steps ... 42

4.2 Preprocessing Platform ... 43

4.2.1 Hardware Environment ... 43

4.2.2 Software Environment ... 43

4.2.3 Data Preparation ... 43

4.3 Preprocessing of Structural Images ... 44

4.3.1 Noise Reduction ... 44

4.3.2 Image Removal of Non-brain Tissue ... 45

4.3.3 Reconstruction of Cortex & Image Segmentation of Brain Tissue ... 48

4.3.4 Co-registration ... 49

4.4 Preprocessing of Functional Images ... 49

4.4.1 Primary Processing ... 49

(15)

Contents

XI

4.4.2 Co-registration & Segmentation ... 50

4.4.3 Noise Reduction ... 50

4.4.4 Filtering, De-trending & Smoothing ... 51

4.5 Calculation of rfMRI Variability & Reliability ... 51

4.6 Summary ... 55

Chapter 5 Variability & Reliability of Regional Metrics ... 56

5.1 Post-processing of Regional Metrics ... 56

5.1.1 Calculation of Amplitude of Low Frequency Fluctuations (ALFF) ... 56

5.1.1 Calculation of Fractional ALFF (fALFF) ... 57

5.2 Findings of Regional Metrics ... 58

5.2.1 Findings of ALFF ... 58

5.2.2 Findings of fALFF ... 61

5.3 Summary ... 63

Chapter 6 Variability & Reliability of Short Distance Functional

Connectivity ... 64

6.1 Post-processing of Short Distance Functional Connectivity (FC) ... 64

6.2 Findings of short distance FC ... 66

6.2.1 Regional Homogeneity ... 66

6.2.2 Regional Homogeneity 2 ... 68

6.3 Summary ... 71

Chapter 7 Variability & Reliability of Long Distance Functional

Connectivity ... 72

7.1 Post-processing of long distance (FC) ... 72

7.1.1 Calculation of Seed-based Functional Connectivity ... 72

7.1.2 Calculation of Graph Theory . ... 73

7.1.3 Calculation of Weighted Centrality ... 75

7.1.4 Calculation of Centrality of eigenvector ... 75

7.2 Findings of Long Distance FC ... 76

(16)

Contents

XII

7.1.3 Finding of Weighted Centrality . ... 76

7.2.2 Finding of Centrality of Eigenvector ... 78

7.3 Summary ... 81

Chapter 8 Variability, Reliability & Clustering of rfMRI ... 82

8.1 Whole Brain Metrics ... 82

8.2 Vertex Metrics ... 83

8.3 Whole Brain Metrics of Seven Sub-networks... 85

8.4 Whole Brain Metrics of Twelve Cognitive Functions ... 87

8.5 Clustering of Variability & Reliability of Whole Brain Metrics ... 90

8.6 Summary ... 93

Chapter 9 Conclusions and Future Direction ... 94

9.1 Main Findings ... 94

9.2 New Findings of the Thesis ... 95

9.3 Future Direction ... 96

References ... 98

Author’s Publications and Participated Projects ... 110

Acknowledgements ... 112

Appendix Abbreviation of Terminologies ... 113

(17)

Degree papers are in the “Xiamen University Electronic Theses and Dissertations Database”.

Fulltexts are available in the following ways:

1. If your library is a CALIS member libraries, please log on

http://etd.calis.edu.cn/ and submit requests online, or consult the interlibrary loan department in your library.

2. For users of non-CALIS member libraries, please mail to etd@xmu.edu.cn

for delivery details.

References

Related documents

Although the Reserve Bank did not dramatically raise real interest rates or attempt to defend the real exchange rate during these inflationary episodes, it does not mean that

If at any time prior to the Department’s issuance of IRS Forms 8609 to the Development Owner the Department determines that (i) the amount of the TCAP Loan that has been funded to

4 was obtained by implanting the water body information extracted by use of the object oriented technology and building land information extracted by use of the mono-temporal

If the data happens to reside on storage that is not local to the user making the request, then dCache creates an internal copy and allows access from that local

In terms of interpersonal conflict and values, I believe that, whilst officers and therapists do share the value of rehabilitation of prisoners, the joumey is different for

In SAR, the virtual scene was projected directly onto the physical objects whereas a white wooden surface located at the same position was used in the SCREEN

In order to test the response of migrating cancer cells to anti-cancer drugs such as Paclitaxel (Taxol) in different microenvironments, both MDA-MB-231 and MCF-7 cells were

Previous studies investigating voluntary segment disclosures have focused on a range of variables originating from agency, information asymmetry, political costs and proprietary