Handbook of Functional MRI Data Analysis

Functional magnetic resonance imaging (fMRI) has become the most popular method for imaging brain function. Handbook for Functional MRI Data Analysis provides a comprehensive and practical introduction to the methods used for fMRI data analysis. Using minimal jargon, this book explains the concepts behind processing fMRI data, focusing on the techniques that are most commonly used in the field. This book provides background about the methods employed by common data analysis packages including FSL, SPM, and AFNI. Some of the newest cutting-edge techniques, including pattern classification analysis, connectivity modeling, and resting state network analysis, are also discussed. Readers of this book, whether newcomers to the field or experienced researchers, will obtain a deep and effective knowledge of how to employ fMRI analysis to ask scientific questions and become more sophisticated users of fMRI analysis software. Title 4 Copyright 5 Contents 6 Preface 10 1 Introduction 12 1.1 A brief overview of fMRI 12 1.2 The emergence of cognitive neuroscience 14 1.3 A brief history of fMRI analysis 15 1.4 Major components of fMRI analysis 18 1.5 Software packages for fMRI analysis 18 1.6 Choosing a software package 21 1.7 Overview of processing streams 21 1.8 Prerequisites for fMRI analysis 21 2 Image processing basics 24 2.1 What is an image? 24 2.2 Coordinate systems 26 2.3 Spatial transformations 28 2.4 Filtering and Fourier analysis 42 3 Preprocessing fMRI data 45 3.1 Introduction 45 3.2 An overview of fMRI preprocessing 45 3.3 Quality control techniques 45 3.4 Distortion correction 49 3.5 Slice timing correction 52 3.6 Motion correction 54 3.7 Spatial smoothing 61 4 Spatial normalization 64 4.1 Introduction 64 4.2 Anatomical variability 64 4.3 Coordinate spaces for neuroimaging 65 4.4 Atlases and templates 66 4.4.1 The Talairach atlas 66 4.4.2 The MNI templates 66 4.5 Preprocessing of anatomical images 67 4.5.1 Bias field correction 67 4.5.2 Brain extraction 67 4.5.3 Tissue segmentation 68 4.6 Processing streams for fMRI normalization 69 4.7 Spatial normalization methods 71 4.7.1 Landmark-based methods 71 4.7.2 Volume-based registration 71 4.7.3 Computational anatomy 72 4.8 Surface-based methods 73 4.9 Choosing a spatial normalization method 74 4.10 Quality control for spatial normalization 76 4.11 Troubleshooting normalization problems 77 4.12 Normalizing data from special populations 77 5 Statistical modeling: Single subject analysis 81 5.1 The BOLD signal 81 5.2 The BOLD noise 97 5.2.1 Characterizing the noise 97 5.2.2 High-pass filtering 99 5.2.3 Prewhitening 101 5.2.4 Precoloring 103 5.3 Study design and modeling strategies 103 6 Statistical modeling: Group analysis 111 6.1 The mixed effects model 111 6.1.1 Motivation 111 6.1.2 Mixed effects modeling approach used in fMRI 113 6.1.3 Fixed effects models 115 6.2 Mean centering continuous covariates 116 6.2.1 Single group 117 6.2.2 Multiple groups 117 7 Statistical inference on images 121 7.1 Basics of statistical inference 121 7.2 Features of interest in images 123 7.3 The multiple testing problem and solutions 127 7.3.1 Familywise error rate 128 7.3.1.1 Bonferroni correction 128 7.3.1.2 Random field theory 128 7.3.1.3 Parametric simulations 130 7.3.1.4 Nonparametric approaches 130 7.3.2 False discovery rate 132 7.3.3 Inference example 134 7.4 Combining inferences: masking and conjunctions 134 7.5 Use of region of interest masks 137 7.6 Computing statistical power 137 8 Modeling brain connectivity 141 8.1 Introduction 141 8.2 Functional connectivity 142 8.2.1 Seed voxel correlation: Between-subjects 142 8.2.2 Seed voxel correlation: Within-subjects 143 8.2.2.1 Avoiding activation-induced correlations 144 8.2.3 Beta-series correlation 144 8.2.4 Psychophysiological interaction 145 8.2.4.1 Creating the PPI regressor 146 8.2.4.2 Potential problems with PPI 147 8.2.5 Multivariate decomposition 147 8.2.5.1 Principal components analysis 148 8.2.5.2 Independent components analysis 149 8.2.5.3 Performing ICA/PCA on group data 153 8.2.6 Partial least squares 154 8.3 Effective connectivity 155 8.4 Network analysis and graph theory 166 8.4.1 Small world networks 166 8.4.2 Modeling networks with resting-state fMRI data 167 8.4.3 Preprocessing for connectivity analysis 168 9 Multivoxel pattern analysis and machine learning 171 9.1 Introduction to pattern classification 171 9.1.1 An overview of the machine learning approach 171 9.1.1.1 Features, observations, and the “curse of dimensionality” 172 9.1.1.2 Overfitting 172 9.2 Applying classifiers to fMRI data 174 9.3 Data extraction 174 9.4 Feature selection 175 9.5 Training and testing the classifier 176 9.5.1 Feature selection/elimination 176 9.5.2 Classifiers for fMRI data 178 9.5.2.1 Linear vs. nonlinear classifiers 178 9.5.2.2 Computational limitations 181 9.5.2.3 Tendency to overfit 181 9.5.3 Which classifier is best? 181 9.5.4 Assessing classifier accuracy 182 9.6 Characterizing the classifier 182 10 Visualizing, localizing, and reporting fMRI data 184 10.1 Visualizing activation data 184 10.2 Localizing activation 187 10.2.1 The Talairach atlas 188 10.2.2 Anatomical atlases 189 10.2.3 Probabilistic atlases 190 10.2.4 Automated anatomical labeling 190 10.3 Localizing and reporting activation 190 10.4 Region of interest analysis 194 10.4.1 ROIs for statistical control 194 10.4.2 Defining ROIs 194 10.4.3 Quantifying signals within an ROI 196 10.4.3.1 Voxel-counting 196 10.4.3.2 Extracting signals for ROI analysis 196 10.4.3.3 Computing percent signal change 197 10.4.3.4 Summarizing data within an ROI 200 Appendix A: Review of the General Linear Model 202 A.1 Estimating GLM parameters 202 A.2 Hypothesis testing 205 A.3 Correlation and heterogeneous variances 206 A.4 Why "general'' linear model? 208 Appendix B: Data organization and management 212 B.1 Computing for fMRI analysis 212 B.2 Data organization 213 B.3 Project management 215 B.4 Scripting for data analysis 216 Appendix C: Image formats 219 C.1 Data storage 219 C.2 File formats 220 Bibliography 222 Index 236 9780521517669 Cambridge University Press Title......Page 4 Copyright......Page 5 Contents......Page 6 Preface......Page 10 1.1 A brief overview of fMRI......Page 12 1.2 The emergence of cognitive neuroscience......Page 14 1.3 A brief history of fMRI analysis......Page 15 1.5 Software packages for fMRI analysis......Page 18 1.8 Prerequisites for fMRI analysis......Page 21 2.1 What is an image?......Page 24 2.2 Coordinate systems......Page 26 2.3 Spatial transformations......Page 28 2.4 Filtering and Fourier analysis......Page 42 3.3 Quality control techniques......Page 45 3.4 Distortion correction......Page 49 3.5 Slice timing correction......Page 52 3.6 Motion correction......Page 54 3.7 Spatial smoothing......Page 61 4.2 Anatomical variability......Page 64 4.3 Coordinate spaces for neuroimaging......Page 65 4.4.2 The MNI templates......Page 66 4.5.2 Brain extraction......Page 67 4.5.3 Tissue segmentation......Page 68 4.6 Processing streams for fMRI normalization......Page 69 4.7.2 Volume-based registration......Page 71 4.7.3 Computational anatomy......Page 72 4.8 Surface-based methods......Page 73 4.9 Choosing a spatial normalization method......Page 74 4.10 Quality control for spatial normalization......Page 76 4.12 Normalizing data from special populations......Page 77 5.1 The BOLD signal......Page 81 5.2.1 Characterizing the noise......Page 97 5.2.2 High-pass filtering......Page 99 5.2.3 Prewhitening......Page 101 5.3 Study design and modeling strategies......Page 103 6.1.1 Motivation......Page 111 6.1.2 Mixed effects modeling approach used in fMRI......Page 113 6.1.3 Fixed effects models......Page 115 6.2 Mean centering continuous covariates......Page 116 6.2.2 Multiple groups......Page 117 7.1 Basics of statistical inference......Page 121 7.2 Features of interest in images......Page 123 7.3 The multiple testing problem and solutions......Page 127 7.3.1.2 Random field theory......Page 128 7.3.1.4 Nonparametric approaches......Page 130 7.3.2 False discovery rate......Page 132 7.4 Combining inferences: masking and conjunctions......Page 134 7.6 Computing statistical power......Page 137 8.1 Introduction......Page 141 8.2.1 Seed voxel correlation: Between-subjects......Page 142 8.2.2 Seed voxel correlation: Within-subjects......Page 143 8.2.3 Beta-series correlation......Page 144 8.2.4 Psychophysiological interaction......Page 145 8.2.4.1 Creating the PPI regressor......Page 146 8.2.5 Multivariate decomposition......Page 147 8.2.5.1 Principal components analysis......Page 148 8.2.5.2 Independent components analysis......Page 149 8.2.5.3 Performing ICA/PCA on group data......Page 153 8.2.6 Partial least squares......Page 154 8.3 Effective connectivity......Page 155 8.4.1 Small world networks......Page 166 8.4.2 Modeling networks with resting-state fMRI data......Page 167 8.4.3 Preprocessing for connectivity analysis......Page 168 9.1.1 An overview of the machine learning approach......Page 171 9.1.1.2 Overfitting......Page 172 9.3 Data extraction......Page 174 9.4 Feature selection......Page 175 9.5.1 Feature selection/elimination......Page 176 9.5.2.1 Linear vs. nonlinear classifiers......Page 178 9.5.3 Which classifier is best?......Page 181 9.6 Characterizing the classifier......Page 182 10.1 Visualizing activation data......Page 184 10.2 Localizing activation......Page 187 10.2.1 The Talairach atlas......Page 188 10.2.2 Anatomical atlases......Page 189 10.3 Localizing and reporting activation......Page 190 10.4.2 Defining ROIs......Page 194 10.4.3.2 Extracting signals for ROI analysis......Page 196 10.4.3.3 Computing percent signal change......Page 197 10.4.3.4 Summarizing data within an ROI......Page 200 A.1 Estimating GLM parameters......Page 202 A.2 Hypothesis testing......Page 205 A.3 Correlation and heterogeneous variances......Page 206 A.4 Why "general'' linear model? ......Page 208 B.1 Computing for fMRI analysis......Page 212 B.2 Data organization......Page 213 B.3 Project management......Page 215 B.4 Scripting for data analysis......Page 216 C.1 Data storage......Page 219 C.2 File formats......Page 220 Bibliography......Page 222 Index......Page 236 Functional Magnetic Resonance Imaging (fmri) Has Become The Most Popular Method For Imaging Brain Function. Handbook Of Functional Mri Data Analysis Provides A Comprehensive And Practical Introduction To The Methods Used For Fmri Data Analysis. Using Minimal Jargon, This Book Explains The Concepts Behind Processing Fmri Data, Focusing On The Techniques That Are Most Commonly Used In The Field. This Book Provides Background About The Methods Employed By Common Data Analysis Packages Including Fsl, Spm And Afni. Some Of The Newest Cutting-edge Techniques, Including Pattern Classification Analysis, Connectivity Modeling And Resting State Network Analysis, Are Also Discussed. Readers Of This Book, Whether Newcomers To The Field Or Experienced Researchers, Will Obtain A Deep And Effective Knowledge Of How To Employ Fmri Analysis To Ask Scientific Questions And Become More Sophisticated Users Of Fmri Analysis Software. Russell A. Poldrack, Jeanette A. Mumford, Thomas E. Nichols. Title From Publisher's Bibliographic System (viewed On 01 Jun 2016). Mode Of Access: World Wide Web. "Functional magnetic resonance imaging (fMRI) has become the most popular method for imaging brain function. Handbook of Functional MRI Data Analysis provides a comprehensive and practical introduction to the methods used for fMRI data analysis. Using minimal jargon, this book explains the concepts behind processing fMRI data, focusing on the techniques that are most commonly used in the field. This book provides background about the methods employed by common data analysis packages including FSL, SPM, and AFNI. Some of the newest cutting-edge techniques, including pattern classification analysis, connectivity modeling, and resting state network analysis, are also discussed. Readers of this book, whether newcomers to the field or experienced researchers, will obtain a deep and effective knowledge of how to employ fMRI analysis to ask scientific questions and become more sophisticated users of fMRI analysis software"--Provided by publisher. Machine generated contents note: 1. Introduction; 2. Image processing; 3. Preprocessing; 4. Normalization; 5. Statistical modeling; 6. Statistical modeling: group analysis; 7. Statistical inference; 8. Connectivity; 9. Visualization; 10. Machine learning; Appendix A. GLM intro/review; Appendix B. Data organization and management; Appendix C. Image formats.