EEG - fMRI: Physiological Basis, Technique, and Applications

Preface to the Second Edition
Reference
Contents
Part I: Background
1: Principles of Multimodal Functional Imaging and Data Integration
1.1 Introduction
1.2 Modes of Data Integration
1.3 Multimodal Data Acquisition Strategies: Degree of Synchrony
1.4 Multimodal Data Integration Strategies
1.4.1 Spatial Coregistration
1.4.2 Asymmetric Integration
1.4.3 Symmetrical Data Fusion
1.5 Summary
References
2: EEG: Origin and Measurement
2.1 Introduction to the Electrophysiology of the Brain
2.2 Origin of EEG and MEG: Cellular Sources
2.3 Main Types of Rhythmical EEG/MEG Activities: Phenomenology and Functional Significance
2.3.1 Sleep EEG Phenomena
2.3.2 Theta Rhythms
2.3.3 Alpha Rhythms of Neocortex and Thalamus
2.3.4 Beta and Gamma Activity of the Neocortex
2.3.5 DC and Ultraslow Potentials
2.4 Origin of the EEG/MEG II: Generators, Volume Conduction and Source Estimation
2.5 Localisation Methods Applied to Spontaneous Oscillatory Activities: Alpha, Mu and Sleep Spindles
2.6 Conclusions
References
3: The Basics of Functional Magnetic Resonance Imaging
3.1 The Basics of MR Imaging
3.1.1 Spins in an External Magnetic Field
3.1.2 The Magnetic Resonance Effect
3.1.3 Spatial Encoding in MR Imaging
3.1.3.1 Frequency Encoding
3.1.3.2 Phase Encoding
3.1.3.3 Slice Selection
3.1.4 Relaxation Times T1 and T2
3.1.5 The Relaxation Time T2* and Gradient Echoes
3.1.6 k-Space
3.1.7 Echo Planar Imaging (EPI)
3.1.8 Spin Echoes
3.1.9 The Specific Absorption Rate (SAR)
3.2 The Cerebral Blood Flow (CBF)
3.2.1 Definition, Order of Magnitude and Measurement
3.2.2 Arterial Spin Labelling Measurements
3.2.3 Labelling Methods
3.2.4 Quantification Problems in ASL
3.3 The Cerebral Blood Volume (CBV)
3.3.1 Definition, Order of Magnitude and Measurement
3.3.2 Contrast Agent-Based Methods
3.3.2.1 Dynamic Imaging
3.3.2.2 Steady-State Imaging
3.3.3 Contrast Agent-Free Method: Vascular Space Occupancy Measurement
3.4 The BOLD Effect and Functional MRI
References
4: Locally Measured Neuronal Correlates of Functional MRI Signals
4.1 Blood Oxygenation Level-Dependent (BOLD) Signals
4.2 Extracellular Neurophysiological Signals
4.3 Relationship Between Neuronal Activity and fMRI Signals
4.4 Correlations Between Neurophysiological Signals and fMRI Responses
4.5 What Is the Neural Origin of fMRI Responses?
4.6 Neuronal Correlates of Negative BOLD Responses
4.7 Neuronal Correlates of Spontaneous Fluctuations in fMRI Signals
4.8 Dissociations Between BOLD Responses and Neurophysiological Activity
References
5: What Can fMRI Add to the ERP Story?
5.1 Introduction
5.2 ERP Generator Localization
5.3 The Inverse Problem of EEG
5.4 Does fMRI Help to Solve the Inverse Problem?
5.5 Further Aspects
5.6 Serial Processing vs. Parallel and Reciprocal Network Activity
5.7 Subcortical Processing
5.8 Conclusions
References
6: The Added Value of EEG-fMRI in Imaging Neuroscience
6.1 Introduction
6.2 The EEG-fMRI Integrated Source Space
6.3 Data Integration Strategies for EEG-fMRI Studies
6.4 Illustration of the Integration of fMRI and EEG in the Temporal Domain
6.5 Illustration of the Integration of fMRI and EEG in the Spatial Domain
6.6 Direct Integration of fMRI and Intra-cranial EEG in the Spatial Domain
6.7 Discussion
References
Part II: Technical and Methodological Aspects of Combined EEG-fMRI Experiments
7: EEG Instrumentation and Safety in the MRI Environment
7.1 Introduction
7.2 EEG Instrumentation
7.2.1 Electrodes
7.2.1.1 Electrode Materials
7.2.1.2 Electrode Lead Arrangement
7.2.1.3 Electrode Lead Movement
7.2.2 EEG Recording System
7.2.2.1 Filters
7.2.2.2 Sampling Rate
7.2.2.3 Signal Range
7.2.2.4 Signal Resolution
7.2.3 RF Emissions
7.2.4 Miscellaneous Factors
7.2.5 Summary
7.3 Safety
7.3.1 Safety Limits
7.3.2 Static Field
7.3.3 Gradient Fields
7.3.4 Eddy Currents
7.3.5 RF Fields
7.3.6 Implanted Electrodes
7.3.7 Summary
References
8: EEG Quality: The Pulse Artifact
8.1 Introduction
8.2 Biophysical Mechanisms
8.2.1 Hypothesized Sources
8.2.2 Experimental Evidence
8.3 Data Acquisition Considerations
8.4 Artifact Reduction Methods
8.4.1 Temporal Waveform-Based Methods
8.4.2 Spatiotemporal Pattern-Based Methods
8.4.3 Sensor-Based Methods
8.4.4 Artifact Reduction Evaluation
8.5 Conclusion
References
9: EEG Quality: The Image Acquisition Artefact
9.1 Origin of the Image Acquisition Artefact
9.2 Characteristics of the Image Acquisition Artefact
9.2.1 Characterisation of the Cooling Pump Artefact
9.2.1.1 Cooling Pump Artefact Prevention
9.2.1.2 Cooling Pump Artefact Removal
9.3 Avoiding Image Acquisition Artefacts: Interleaved EEG–fMRI Protocols
9.4 Reduction of Image Acquisition Artefacts
9.4.1 Reduction at the Source
9.4.1.1 Stepping-Stone Sampling
9.4.2 Synchronisation of EEG and fMRI Data Acquisitions
9.5 Correction of the Image Acquisition Artefact Using EEG Post-Processing
9.5.1 Artefact Template Subtraction
9.5.2 Computing and Correcting Timing Errors
9.5.3 Temporal Principal Component Analysis
9.5.4 Independent Component Analysis
9.5.5 Filtering in the Frequency Domain
9.5.6 Between Prevention and Correction: Prospective Motion Correction and EEG Artefacts
9.6 Evaluation of Correction Methods
References
10: Image Quality Issues
10.1 fMRI Pulse Sequences
10.2 GE-EPI
10.2.1 Image Blurring
10.2.2 Geometric Distortion
10.2.3 Signal Dropout
10.2.4 Image Ghosting
10.2.5 RF Interference
10.3 Other Sources of Image Artefact in fMRI
10.3.1 Bulk Head Motion
10.3.2 Physiological Noise
10.4 The Impact of EEG Recording on MR Image Quality
10.4.1 Main Static Magnetic Field (B0) Effects
10.4.2 Transverse Rotational Magnetic Field (B1) Effects
10.4.3 Impact on SNR
10.5 fMRI Quality Assurance (QA)
10.5.1 Quantification of SNR and Temporal SNR
10.5.2 The Weisskoff Test
10.5.3 Coherent Noise Testing
10.6 Summary and Conclusions
References
11: EEG-fMRI at Ultrahigh Magnetic Fields: B0 ≥ 3 Tesla
11.1 Introduction
11.2 Safety Considerations
11.2.1 Physical Principles and Relevant Safety Guidelines
11.2.2 Safety Studies at High Fields
11.2.3 Safe Imaging with High-Density EEG Nets
11.3 EEG Recording and Quality
11.3.1 Pulse-Related Artefact
11.3.2 Other Noise Sources at High Field
11.3.3 EEG Noise Removal Strategies at High Field
11.4 Image Quality
11.5 Example of an Application of EEG-fMRI at 7 T: Auditory Steady-State Response (ASSR)
11.6 Conclusions
References
12: Experimental Design and Data Analysis Strategies
12.1 Introduction
12.2 Data Acquisition and Experimental Design
12.2.1 Interleaved EEG and fMRI Acquisitions: Triggered and Sparse Scanning
12.2.2 Simultaneous EEG and fMRI Acquisitions: Continuous Scanning
12.2.3 Experimental Protocol
12.2.3.1 Resting-State EEG–fMRI: Spontaneous Brain Activity
12.2.3.2 Stimulus-Driven Paradigms
12.3 Analysis of Simultaneously Acquired EEG–fMRI Data
12.3.1 Model-Based Analysis of fMRI Time-Series Data
12.3.1.1 Preprocessing
12.3.1.2 The General Linear Model (GLM) and Statistical Inference
12.3.2 EEG-Derived GLM: Use of Event on sets and Illustration in Epilepsy
12.3.3 EEG-Derived GLM: Parametric Design and Single Trial
12.3.4 EEG-Derived GLM: EEG Spectrum
12.3.5 Multivariate Analysis
12.3.6 EEG-informed fMRI Functional Connectivity
12.3.7 Use of intracerebral EEG in the context of concurrent fMRI recordings
12.4 EEG and fMRI Localization: Modes of Integration
12.4.1 Comparison of Independently Derived Results
12.4.2 fMRI as a Spatial Constraint for EEG Source Reconstruction
12.4.3 Towards Symmetrical Models of EEG and fMRI Fusion
12.5 Unresolved Problems and Caveats
12.5.1 Relationship Between Neuronal Activity, EEG and fMRI Signals
12.5.2 Specific Issues Related to Spontaneous Brain Activity
12.5.2.1 HRF
12.5.2.2 Experimental Efficiency of Paradigm-less fMRI
12.5.3 The Impact of Data Acquisition and Processing Artefacts on fMRI Data Analysis
12.5.3.1 Artefacts in the Signals
12.5.3.2 Artefacts Introduced by EEG Preprocessing
12.6 Summary and Outlook
References
13: Real-Time fMRI Neurofeedback with Simultaneous EEG
13.1 Introduction
13.2 Regulation of Amygdala BOLD Activity and Frontal EEG Asymmetry
13.3 Regulation of Thalamic BOLD Activity and Alpha EEG Rhythm
13.4 Simultaneous Real-Time fMRI and EEG Neurofeedback
13.5 Real-Time Independent Component Analysis for EEG-fMRI
13.6 Conclusions
References
14: Non-invasive Brain Stimulation with Multimodal Acquisitions
14.1 Brain Imaging: Possibilities and Limitations
14.2 Invasive and Non-invasive Brain Stimulation
14.2.1 The Physics and Physiology of Single-Pulse Transcranial Magnetic Stimulation (TMS)
14.2.2 From Single-Pulse to Repetitive TMS: Stimulation Protocols
14.2.3 Clinical Applications of TMS
14.3 The Multimodal Approach: Combinations of Brain Stimulation and Brain Imaging
14.3.1 Brain Imaging Before Brain Stimulation
14.3.2 Brain Imaging After Brain Stimulation
14.3.3 Simultaneous Brain Stimulation and Brain Imaging
14.3.3.1 Technical Challenges and Practical Implementation
14.3.3.2 TMS Affects Networks, Not Just a Local Region
14.3.3.3 TMS Network Effects Depend on Brain State
14.3.3.4 TMS Network Effects Are Functionally Relevant
14.4 New Developments
14.4.1 Closed-Loop Neuroscience
14.4.2 Simultaneous TMS-fMRI-EEG
14.5 Conclusions
References
Part III: Applications of EEG-fMRI
15: Brain Rhythms
15.1 Multimodal Studies of Brain Rhythms
15.1.1 Considerations for the Study of Rest
15.1.2 From Unimodal to Multimodal Approaches to the Resting State
15.1.3 Unimodal Approaches to Resting State
15.1.3.1 Resting State in fMRI
15.1.3.2 Spontaneous Neural Activity in Electrophysiology
15.1.4 Multimodal Studies of Rest
15.1.4.1 Direct and Indirect Measurement of Neural Activity by (f)MRI
15.1.4.2 Functional Imaging Studies of “Brain Oscillations”
15.1.4.3 Endogenous Brain Oscillations in Healthy Subjects
15.1.4.4 Similar Electrical Oscillations, Different fMRI Networks
15.1.4.5 Similar fMRI Networks, Different Electrical Oscillations
15.1.4.6 Brain Rhythms and Connectivity
15.1.4.7 Brain Oscillations and Networks During Sleep
15.1.4.8 Endogenous Brain Oscillations in Patients with Epilepsy
15.2 Multimodal Approaches in a Task Setting
15.2.1 General Considerations
15.2.2 Multimodal Measurements in a Task Context: Examples
15.2.3 Linking Neuronal Oscillations to Haemodynamic Changes
15.3 Conclusions
References
16: Sleep
16.1 FMRI in Sleep Research
16.1.1 Sleep
16.1.2 Imaging Sleep
16.1.3 EEG and fMRI in Sleep Research
16.2 fMRI During Sleep: Technical Challenges
16.2.1 Sleep Recording
16.2.1.1 Multimodality of Sleep Recording
16.2.1.2 Referentiation of Recordings
16.2.1.3 Extended Recording Time: Electrophysiological Recordings
16.2.2 MR Imaging
16.2.2.1 Extended Recording Time: fMRI Recordings
16.2.2.2 Movement
16.2.3 Effect on Participant
16.2.3.1 Participant Not Falling or Staying Asleep
16.2.3.2 Extended Recording Time: Subjective Discomfort
16.2.4 Effect on Study Protocols
16.2.4.1 Drop-Out Rate
16.2.4.2 Specific Suppression of Sleep Stages
16.2.4.3 Selection Bias
16.2.4.4 No Control Over Sleep State
16.2.4.5 No Whole-Night Recordings
16.2.4.6 Fluctuation of Microstates
16.2.5 Possible Solutions
16.2.5.1 Habituation
16.2.5.2 Sleep Deprivation
16.2.5.3 MR Recording Techniques
16.3 FMRI in Sleep: Results
16.3.1 Sensory Processing During Sleep
16.3.1.1 NREM Sleep
Acoustic Stimulation
Visual Stimulation
Olfactory Stimulation
16.3.1.2 REM Sleep
16.3.2 EEG-Informed fMRI
16.3.2.1 Falling Asleep
16.3.2.2 Graphoelements: Spindles, K-Complexes, and Slow Oscillations
16.3.2.3 REM Sleep
16.3.2.4 Lucid REM Sleep
16.3.3 Network Analysis
16.3.4 Animal Data
16.4 Summary and Outlook
References
17: EEG–fMRI in Adults with Focal Epilepsy
17.1 Introduction
17.2 Interictal EEG–fMRI
17.2.1 What Is an Interictal Spike?
17.2.2 Interictal Epileptiform Activity in Presurgical Assessment
17.2.3 Methodology
17.2.3.1 Data Acquisition
17.2.3.2 Data Analysis
17.2.4 Relevance of the Observed BOLD Changes
17.2.5 Clinical Utility
17.2.6 The Influence of Lesions
17.2.7 Simultaneous intracranial EEG-fMRI
17.3 Ictal EEG–fMRI
17.3.1 Limitations of Ictal EEG–fMRI
17.3.1.1 Unpredictable Nature of Seizures
17.3.1.2 Seizure-Related Motion
17.3.2 Detection of Ictal Activity
17.3.3 Statistical Analysis of Ictal Haemodynamic Changes
17.3.4 Application of Ictal EEG–fMRI
17.3.4.1 Localisation Potential of Ictal EEG–fMRI
17.3.4.2 Mechanism of Epilepsy
17.4 Conclusions
References
18: EEG-fMRI in Generalised Epilepsy: Adults
18.1 Idiopathic Generalised Epilepsy
18.1.1 Definition and Classification of Generalised Epilepsy Syndromes
18.1.2 Diagnosing IGE
18.1.3 IGE Comorbidity
18.2 Cortical and Subcortical Generators of Generalised Spike and Wave Activity
18.3 What EEG fMRI Has to Tell Us About Generators of Generalised Spike and Wave in Adults
18.3.1 EEG-fMRI Provides a Topographic Map of Structures Involved in the Generation of GSW
18.3.1.1 The Thalamus and Cortex in GSW
18.3.1.2 The Thalamus
Non-Thalamic Subcortical Contributions
18.3.1.3 Cortical BOLD: The Importance of the Default Mode Network
Cortical Change in Adults Beyond the DMN
18.3.1.4 Cortical Change in Lennox-Gastaut Syndrome
18.3.1.5 Connectivity Analysis Insights into the Mechanisms of GSW
18.4 Conclusions
References
19: EEG-fMRI in Children with Epilepsy
19.1 EEG-fMRI in Children with Epilepsy
19.2 Methodological Issues Specific to Paediatric EEG-fMRI Studies
19.2.1 Patient Selection and Scanning
19.2.2 Modelling IED-Related BOLD Changes in Children: Variability and Developmental Changes
19.3 Results of EEG-fMRI Studies in Paediatric Epilepsy
19.3.1 Self-Limited Focal Epilepsies
19.3.2 Symptomatic and Cryptogenic Focal Epilepsies
19.3.3 Idiopathic Generalized Epilepsies
19.3.4 Epileptic Encephalopathies
19.4 Summary and Future Perspectives
References
20: EEG-fMRI in Psychiatry
20.1 Introduction
20.2 Anxiety Disorder
20.3 Attention Deficit Hyperactivity Disorder (ADHD)
20.4 Depression, Posttraumatic Stress Disorder (PTSD) and Neurofeedback
20.5 Schizophrenia
20.6 Obsessive-Compulsive Disorder (OCD)
20.7 Dementia
References
21: Combining Electroencephalography and Functional Magnetic Resonance Imaging in Pain Research
21.1 Introduction
21.2 Combining EEG and fMRI in Pain Research: General Issues
21.3 Combining EEG and fMRI in Pain Research: Practical Issues
21.3.1 Selectivity of the Nociceptive Input in EEG-fMRI Studies
21.3.2 Delivery of Nociceptive Stimuli in EEG/fMRI Environment
21.3.3 Experimental Design
21.4 Studies Combining EEG and fMRI in Pain Research
21.5 Future Directions: EEG-Driven Analysis of fMRI-BOLD Responses to Nociceptive Stimulation
21.5.1 Single-Trial Estimation of the Magnitude of Stimulus-Evoked EEG Responses
21.5.2 Correlation Between EEG and fMRI Responses at Single-Trial Level
References
22: Simultaneous Electroencephalography and Functional Magnetic Resonance Imaging of the Human Auditory System
22.1 Introduction
22.2 Specifics of Auditory Recordings
22.2.1 Interference of the Static Magnetic Field
22.2.2 Interference of Transient Magnetic Fields
22.2.3 BOLD Response to Scanner Noise
22.2.4 Sparse Sampling
22.2.5 Silent fMRI Acquisition
22.2.6 Adjusting Auditory Stimulus Frequencies
22.2.7 Active Noise Cancellation
22.3 Simultaneous EEG and fMRI in Auditory Experiments
22.4 Evaluation Methods of Concurrent Auditory EEG-fMRI
22.5 Conclusion
References
23: Visual System
23.1 Simultaneous EEG-fMRI of the Visual System: Signal Quality
23.2 fMRI-Informed EEG of the Visual System
23.2.1 Localising VEPs
23.2.2 Visual Attention and Other Cognitive Processes
23.3 EEG-Informed fMRI of the Visual System
23.3.1 Spontaneous EEG Oscillations
23.3.2 Task-Related EEG Activity
23.4 Uninformed EEG–fMRI and Other Approaches
23.4.1 Event-Related Oscillations (EROs)
23.4.2 Visual Attention and Other Cognitive Processes
23.5 Investigating Neurovascular Coupling in the Visual System by EEG–fMRI
23.6 Outlook
References
24: Cognition
24.1 Advantages and Disadvantages of Simultaneous EEG–fMRI Recordings of Cognitive Functions
24.2 Attention
24.2.1 Oddball Paradigm
24.2.2 Mismatch Negativity
24.2.3 Preparatory Attention
24.3 Executive functions
24.3.1 Cognitive Flexibility
24.3.2 Performance Monitoring
24.3.3 Decision-Making
24.3.4 Behavioral Inhibition
24.3.5 Working Memory
24.4 Memory
24.5 Limitations and Outlook
References
25: Neuronal Models for EEG–fMRI Integration
25.1 From Correlating Measurements to Models of Neuronal Population Activity
25.2 Direct Correlations Between Field Potentials and BOLD
25.3 Neuronal Population Activity and the BOLD Signal
25.4 Neuronal Population Activity and Field Potential Measurements
25.5 Theoretical Predictions for BOLD and Field Potential Measurements
25.6 Theories About Field Potential Components
25.6.1 Broadband Power Changes
25.6.2 Peaks in the Field Potential Power Spectrum in the Range from 30 to 80 Hz
25.6.3 Low Frequency Alpha Oscillations
25.6.4 Measured Field Potential Data Will Be a Summation Across All Underlying Processes
25.7 Predicting Empirical Data with This Modeling Framework
25.8 Discussion
25.9 Conclusion
References
Part IV: Modelling
26: BOLD-Response and EEG Gamma Oscillations
26.1 Introduction
26.2 Methodical Issues
26.3 Gamma activity and BOLD Response
26.3.1 Covariation of High Frequency Oscillations and BOLD Signal
26.3.2 Gamma Activity and BOLD Response: Variation Across Subjects
26.3.3 Gamma Activity and BOLD Response: Further Reports
26.3.4 Single-Trial Coupling of Auditory Evoked Gamma Band Response and BOLD Signal
26.4 Conclusions
References
27: EEG–fMRI in Animal Models
27.1 Introduction
27.2 Advantages of EEG–fMRI in Animal Models
27.3 Limitations and Technical Challenges of EEG–fMRI in Animal Models
27.4 Anesthesia
27.5 Movement: Curarization and Habituation
27.6 Physiology
27.7 MRI Compatible Electrodes
27.8 fMRI Signal Generation
27.8.1 Measurement of CMRO2 by MR Spectroscopy
27.8.2 Estimation of CMRO2 by Calibrated BOLD
27.8.3 CBV
27.8.4 CBF
27.9 Signal Artifact and Artifact Removal
27.10 Data Analysis
27.11 Sequential EEG–fMRI Studies in Animals
27.12 Applications of Simultaneous EEG–fMRI in Animals
27.13 Epilepsy
27.14 Absence Seizure Models
27.15 Generalized Tonic-Clonic Seizure Models
27.16 Partial Seizure Models
27.17 Sleep
27.18 Sensory–Motor Stimulation Models
27.19 Relating fMRI Signals to Electrophysiological Recordings
27.20 Future Directions
27.21 Conclusions
References
28: EEG–fMRI Information Fusion: Biophysics and Data Analysis
28.1 Introduction
28.2 EEG–fMRI Information Fusion: Limitations
28.2.1 Coupling of Electrophysiological and Hemodynamic Responses
28.2.2 Experimental Limitations
28.3 EEG–fMRI Information Fusion: Solutions
28.3.1 Information Fusion: Definition
28.3.2 Asymmetrical vs. Symmetrical Approaches
28.3.3 EEG to fMRI Approaches
28.3.4 fMRI to EEG Approaches
28.3.5 Symmetrical EEG–fMRI Approaches
28.3.5.1 Model-Driven Approaches
28.3.5.2 Data-Driven Approaches
28.4 Conclusion
References
29: Sparse and Data-Driven Methods for Concurrent EEG–fMRI
29.1 Introduction
29.2 Sparse Sampling of the Hemodynamic Response Function
29.3 Leveraging Sparsity for Artifact Removal
29.4 Data Driven Methods for EEG–fMRI Integration
29.4.1 Tensor Factorization
29.4.2 Canonical Polyadic Decomposition (CPD)
29.4.3 Tucker Decomposition
29.4.4 Coupled Matrix–Tensor Factorization (CMTF)
29.5 Canonical Correlation Analysis
29.6 Conclusion
References
30: Integrating EEG–fMRI Through Brain Simulation
30.1 Introduction
30.2 Brain Network Models
30.3 EEG and fMRI Forward Models
30.4 Evoked Potentials
30.5 Resting-State
30.6 EEG–fMRI (Anti)Correlation
30.7 From EEG–fMRI to Neural Activity
30.8 From EEG–fMRI to Neural Mechanisms
30.9 Outlook: Diagnosis and Therapy
References
Index