An information theoretic approach to EEG-fMRI integration of visually evoked responses

Sampling bias corrections for accurate neural measures of redundant, unique, and synergistic information

Shannon Information theory has long been a tool of choice to measure empirically how populations of neurons in the brain encode information about cognitive variables. Recently, Partial Information Decomposition (PID) has emerged as principled way to break down this information into components identifying not only the unique information carried by each neuron, but also whether relationships between neurons generate synergistic or redundant information. While it has been long recognized that Shannon information measures on neural activity suffer from a (mostly upward) limited sampling estimation bias, this issue has largely been ignored in the burgeoning field of PID analysis of neural activity. We used simulations to investigate the limited sampling bias of PID computed from discrete probabilities (suited to describe neural spiking activity). We found that PID suffers from a large bias that is uneven across components, with synergy by far the most biased. Using approximate analytical expansions, we found that the bias of synergy increases quadratically with the number of discrete responses of each neuron, whereas the bias of unique and redundant information increase only linearly or sub-linearly. Based on the understanding of the PID bias properties, we developed simple yet effective procedures that correct for the bias effectively, and that improve greatly the PID estimation with respect to current state-of-the-art procedures. We apply these PID bias correction procedures to datasets of 53117 pairs neurons in auditory cortex, posterior parietal cortex and hippocampus of mice performing cognitive tasks, deriving precise estimates and bounds of how synergy and redundancy vary across these brain regions.

Local similarity of activity patterns during auditory and visual processing

Neuroimaging studies have shown that brain activity is variable and changes according to stimuli and the environmental context, reflecting brain coding or information representations at different processing levels. However, little is known about activity organization that reflects coding strategies. Here, we explored and compared two different coding approaches, spatial via cross-correlation and intensity-based coding using mutual information. Using two fMRI datasets and different seeds, we searched for the spatial and intensity-based similarities with the seeds in brain activity. Our results showed that, apart from the seed regions, significant regions detected by intensity-based similarity analysis differ completely from those found using cross-correlation. These findings may indicate that information shared through spatial coding differs from that transmitted via non-spatial coding processes. Our results suggest that brain coding is organized in several different ways to optimize information processing.

Simultaneous EEG-fMRI: What Have We Learned and What Does the Future Hold?

Simultaneous EEG-fMRI has developed into a mature measurement technique in the past 25 years. During this time considerable technical and analytical advances have been made, enabling valuable scientific contributions to a range of research fields. This review will begin with an introduction to the measurement principles involved in EEG and fMRI and the advantages of combining these methods. The challenges faced when combining the two techniques will then be considered. An overview of the leading application fields where EEG-fMRI has made a significant contribution to the scientific literature and emerging applications in EEG-fMRI research trends is then presented.

Proof-of-concept evidence for trimodal simultaneous investigation of human brain function

The link between spatial (where) and temporal (when) aspects of the neural correlates of most psychological phenomena is not clear. Elucidation of this relation, which is crucial to fully understand human brain function, requires integration across multiple brain imaging modalities and cognitive tasks that reliably modulate the engagement of the brain systems of interest. By overcoming the methodological challenges posed by simultaneous recordings, the present report provides proof‐of‐concept evidence for a novel approach using three complementary imaging modalities: functional magnetic resonance imaging (fMRI), event‐related potentials (ERPs), and event‐related optical signals (EROS). Using the emotional oddball task, a paradigm that taps into both cognitive and affective aspects of processing, we show the feasibility of capturing converging and complementary measures of brain function that are not currently attainable using traditional unimodal or other multimodal approaches. This opens up unprecedented possibilities to clarify spatiotemporal integration of brain function.

Concurrent EEG- and fMRI-derived functional connectomes exhibit linked dynamics

Long-range connectivity has become the most studied feature of human functional Magnetic Resonance Imaging (fMRI), yet the spatial and temporal relationship between its whole-brain dynamics and electrophysiological connectivity remains largely unknown. FMRI-derived functional connectivity exhibits spatial reconfigurations or time-varying dynamics at infraslow (<0.1Hz) speeds. Conversely, electrophysiological connectivity is based on cross-region coupling of fast oscillations (~1-100Hz). It is unclear whether such fast oscillation-based coupling varies at infraslow speeds, temporally coinciding with infraslow dynamics across the fMRI-based connectome. If so, does the association of fMRI-derived and electrophysiological dynamics spatially vary over the connectome across the functionally distinct electrophysiological oscillation bands? In two concurrent electroencephalography (EEG)-fMRI resting-state datasets, oscillation-based coherence in all canonical bands (delta through gamma) indeed reconfigured at infraslow speeds in tandem with fMRI-derived connectivity changes in corresponding region-pairs. Interestingly, irrespective of EEG frequency-band the cross-modal tie of connectivity dynamics comprised a large proportion of connections distributed across the entire connectome. However, there were frequency-specific differences in the relative strength of the cross-modal association. This association was strongest in visual to somatomotor connections for slower EEG-bands, and in connections involving the Default Mode Network for faster EEG-bands. Methodologically, the findings imply that neural connectivity dynamics can be reliably measured by fMRI despite heavy susceptibility to noise, and by EEG despite shortcomings of source reconstruction. Biologically, the findings provide evidence that contrast with known territories of oscillation power, oscillation coupling in all bands slowly reconfigures in a highly distributed manner across the whole-brain connectome.

Adapted cabling of an EEG cap improves simultaneous measurement of EEG and fMRI at 7T

BACKGROUND

The combination of EEG and ultra-high-field (7 T and above) fMRI holds the promise to relate electrophysiology and hemodynamics with greater signal to noise level and at higher spatial resolutions than conventional field strengths. Technical and safety restrictions have so far resulted in compromises in terms of MRI coil selection, resulting in reduced, signal quality, spatial coverage and resolution in EEG-fMRI studies at 7 T.

NEW METHOD

We adapted a 64-channel MRI-compatible EEG cap so that it could be used with a closed 32-channel MRI head coil thus avoiding several of these compromises. We compare functional and anatomical as well as the EEG quality recorded with this adapted setup with those recorded with a setup that uses an open-ended 8-channel head-coil.

RESULTS

Our set-up with the adapted EEG cap inside the closed 32 channel coil resulted in the recording of good quality EEG and (f)MRI data. Both functional and anatomical MRI images show no major effects of the adapted EEG cap on MR signal quality. We demonstrate the ability to compute ERPs and changes in alpha and gamma oscillations from the recorded EEG data.

COMPARISON WITH EXISTING METHODS

Compared to MRI recordings with an 8-channel open-ended head-coil, the loss in signal quality of the MRI images related to the adapted EEG cap is considerably reduced.

CONCLUSIONS

The adaptation of the EEG cap permits the simultaneous recording of good quality whole brain (f)MRI data using a 32 channel receiver coil, while maintaining the quality of the EEG data.

Hemodynamic Correlates of Electrophysiological Activity in the Default Mode Network

Hemodynamic fluctuations in the default mode network (DMN), observed through functional magnetic resonance imaging (fMRI), have been linked to electrophysiological oscillations detected by electroencephalography (EEG). It has been reported that, among the electrophysiological oscillations, those in the alpha frequency range (8–13 Hz) are the most dominant during resting state. We hypothesized that DMN spatial configuration closely depends on the specific neuronal oscillations considered, and that alpha oscillations would mainly correlate with increased blood oxygen-level dependent (BOLD) signal in the DMN. To test this hypothesis, we used high-density EEG (hdEEG) data simultaneously collected with fMRI scanning in 20 healthy volunteers at rest. We first detected the DMN from source reconstructed hdEEG data for multiple frequency bands, and we then mapped the correlation between temporal profile of hdEEG-derived DMN activity and fMRI–BOLD signals on a voxel-by-voxel basis. In line with our hypothesis, we found that the correlation map associated with alpha oscillations, more than with any other frequency bands, displayed a larger overlap with DMN regions. Overall, our study provided further evidence for a primary role of alpha oscillations in supporting DMN functioning. We suggest that simultaneous EEG–fMRI may represent a powerful tool to investigate the neurophysiological basis of human brain networks.

Functional Source Separation for EEG-fMRI Fusion: Application to Steady-State Visual Evoked Potentials

Neurorobotics is one of the most ambitious fields in robotics, driving integration of interdisciplinary data and knowledge. One of the most productive areas of interdisciplinary research in this area has been the implementation of biologically-inspired mechanisms in the development of autonomous systems. Specifically, enabling such systems to display adaptive behavior such as learning from good and bad outcomes, has been achieved by quantifying and understanding the neural mechanisms of the brain networks mediating adaptive behaviors in humans and animals. For example, associative learning from aversive or dangerous outcomes is crucial for an autonomous system, to avoid dangerous situations in the future. A body of neuroscience research has suggested that the neurocomputations in the human brain during associative learning involve re-shaping of sensory responses. The nature of these adaptive changes in sensory processing during learning however are not yet well enough understood to be readily implemented into on-board algorithms for robotics application. Toward this overall goal, we record the simultaneous electroencephalogram (EEG) and functional magnetic resonance imaging (fMRI), characterizing one candidate mechanism, i.e., large-scale brain oscillations. The present report examines the use of Functional Source Separation (FSS) as an optimization step in EEG-fMRI fusion that harnesses timing information to constrain the solutions that satisfy physiological assumptions. We applied this approach to the voxel-wise correlation of steady-state visual evoked potential (ssVEP) amplitude and blood oxygen level-dependent imaging (BOLD), across both time series. The results showed the benefit of FSS for the extraction of robust ssVEP signals during simultaneous EEG-fMRI recordings. Applied to data from a 3-phase aversive conditioning paradigm, the correlation maps across the three phases (habituation, acquisition, extinction) show converging results, notably major overlapping areas in both primary and extended visual cortical regions, including calcarine sulcus, lingual cortex, and cuneus. In addition, during the acquisition phase when aversive learning occurs, we observed additional correlations between ssVEP and BOLD in the anterior cingulate cortex (ACC) as well as the precuneus and superior temporal gyrus.

Neuronal dynamics enable the functional differentiation of resting state networks in the human brain

Intrinsic brain activity is organized in spatial-temporal patterns, called resting-state networks (RSNs), exhibiting specific structural-functional architecture. These networks presumably reflect complex neurophysiological processes and have a central role in distinct perceptual and cognitive functions. In this work, we propose an innovative approach for characterizing RSNs according to their underlying neural oscillations. We investigated specific electrophysiological properties, including spectral features, fractal dimension, and entropy, associated with eight core RSNs derived from high-density electroencephalography (EEG) source-reconstructed signals. Specifically, we found higher synchronization of the gamma-band activity and higher fractal dimension values in perceptual (PNs) compared with higher cognitive (HCNs) networks. The inspection of this underlying rapid activity becomes of utmost importance for assessing possible alterations related to specific brain disorders. The disruption of the coordinated activity of RSNs may result in altered behavioral and perceptual states. Thus, this approach could potentially be used for the early detection and treatment of neurological disorders.

Exploiting neurovascular coupling: a Bayesian sequential Monte Carlo approach applied to simulated EEG fNIRS data

OBJECTIVE

Electrical and hemodynamic brain activity are linked through the neurovascular coupling process and they can be simultaneously measured through integration of electroencephalography (EEG) and functional near-infrared spectroscopy (fNIRS). Thanks to the lack of electro-optical interference, the two procedures can be easily combined and, whereas EEG provides electrophysiological information, fNIRS can provide measurements of two hemodynamic variables, such as oxygenated and deoxygenated hemoglobin. A Bayesian sequential Monte Carlo approach (particle filter, PF) was applied to simulated recordings of electrical and neurovascular mediated hemodynamic activity, and the advantages of a unified framework were shown.

APPROACH

Multiple neural activities and hemodynamic responses were simulated in the primary motor cortex of a subject brain. EEG and fNIRS recordings were obtained by means of forward models of volume conduction and light propagation through the head. A state space model of combined EEG and fNIRS data was built and its dynamic evolution was estimated through a Bayesian sequential Monte Carlo approach (PF).

MAIN RESULTS

We showed the feasibility of the procedure and the improvements in both electrical and hemodynamic brain activity reconstruction when using the PF on combined EEG and fNIRS measurements.

SIGNIFICANCE

The investigated procedure allows one to combine the information provided by the two methodologies, and, by taking advantage of a physical model of the coupling between electrical and hemodynamic response, to obtain a better estimate of brain activity evolution. Despite the high computational demand, application of such an approach to in vivo recordings could fully exploit the advantages of this combined brain imaging technology.

Cross multivariate correlation coefficients as screening tool for analysis of concurrent EEG-fMRI recordings

Over the past decade, the simultaneous recording of electroencephalogram (EEG) and functional magnetic resonance imaging (fMRI) data has garnered growing interest because it may provide an avenue towards combining the strengths of both imaging modalities. Given their pronounced differences in temporal and spatial statistics, the combination of EEG and fMRI data is however methodologically challenging. Here, we propose a novel screening approach that relies on a Cross Multivariate Correlation Coefficient (xMCC) framework. This approach accomplishes three tasks: (1) It provides a measure for testing multivariate correlation and multivariate uncorrelation of the two modalities; (2) it provides criterion for the selection of EEG features; (3) it performs a screening of relevant EEG information by grouping the EEG channels into clusters to improve efficiency and to reduce computational load when searching for the best predictors of the BOLD signal. The present report applies this approach to a data set with concurrent recordings of steady‐state‐visual evoked potentials (ssVEPs) and fMRI, recorded while observers viewed phase‐reversing Gabor patches. We test the hypothesis that fluctuations in visuo‐cortical mass potentials systematically covary with BOLD fluctuations not only in visual cortical, but also in anterior temporal and prefrontal areas. Results supported the hypothesis and showed that the xMCC‐based analysis provides straightforward identification of neurophysiological plausible brain regions with EEG‐fMRI covariance. Furthermore xMCC converged with other extant methods for EEG‐fMRI analysis.

EEG-Informed fMRI: A Review of Data Analysis Methods

The simultaneous acquisition of electroencephalography (EEG) with functional magnetic resonance imaging (fMRI) is a very promising non-invasive technique for the study of human brain function. Despite continuous improvements, it remains a challenging technique, and a standard methodology for data analysis is yet to be established. Here we review the methodologies that are currently available to address the challenges at each step of the data analysis pipeline. We start by surveying methods for pre-processing both EEG and fMRI data. On the EEG side, we focus on the correction for several MR-induced artifacts, particularly the gradient and pulse artifacts, as well as other sources of EEG artifacts. On the fMRI side, we consider image artifacts induced by the presence of EEG hardware inside the MR scanner, and the contamination of the fMRI signal by physiological noise of non-neuronal origin, including a review of several approaches to model and remove it. We then provide an overview of the approaches specifically employed for the integration of EEG and fMRI when using EEG to predict the blood oxygenation level dependent (BOLD) fMRI signal, the so-called EEG-informed fMRI integration strategy, the most commonly used strategy in EEG-fMRI research. Finally, we systematically review methods used for the extraction of EEG features reflecting neuronal phenomena of interest.

Functional Semi-Blind Source Separation Identifies Primary Motor Area Without Active Motor Execution

High time resolution techniques are crucial for investigating the brain in action. Here, we propose a method to identify a section of the upper-limb motor area representation (FS_M1) by means of electroencephalographic (EEG) signals recorded during a completely passive condition (FS_M1bySS). We delivered a galvanic stimulation to the median nerve and we applied to EEG the semi-Blind Source Separation (s-BSS) algorithm named Functional Source Separation (FSS). In order to prove that FS_M1bySS is part of FS_M1, we also collected EEG in a motor condition, i.e. during a voluntary movement task (isometric handgrip) and in a rest condition, i.e. at rest with eyes open and closed. In motor condition, we show that the cortico-muscular coherence (CMC) of FS_M1bySS does not differ from FS_ M1 CMC (0.04 for both sources). Moreover, we show that the FS_M1bySS’s ongoing whole band activity during Motor and both rest conditions displays high mutual information and time correlation with FS_M1 (above 0.900 and 0.800, respectively) whereas much smaller ones with the primary somatosensory cortex [Formula: see text] (about 0.300 and 0.500, [Formula: see text]). FS_M1bySS as a marker of the upper-limb FS_M1 representation obtainable without the execution of an active motor task is a great achievement of the FSS algorithm, relevant in most experimental, neurological and psychiatric protocols.

Variational Bayesian Parameter Estimation Techniques for the General Linear Model

Variational Bayes (VB), variational maximum likelihood (VML), restricted maximum likelihood (ReML), and maximum likelihood (ML) are cornerstone parametric statistical estimation techniques in the analysis of functional neuroimaging data. However, the theoretical underpinnings of these model parameter estimation techniques are rarely covered in introductory statistical texts. Because of the widespread practical use of VB, VML, ReML, and ML in the neuroimaging community, we reasoned that a theoretical treatment of their relationships and their application in a basic modeling scenario may be helpful for both neuroimaging novices and practitioners alike. In this technical study, we thus revisit the conceptual and formal underpinnings of VB, VML, ReML, and ML and provide a detailed account of their mathematical relationships and implementational details. We further apply VB, VML, ReML, and ML to the general linear model (GLM) with non-spherical error covariance as commonly encountered in the first-level analysis of fMRI data. To this end, we explicitly derive the corresponding free energy objective functions and ensuing iterative algorithms. Finally, in the applied part of our study, we evaluate the parameter and model recovery properties of VB, VML, ReML, and ML, first in an exemplary setting and then in the analysis of experimental fMRI data acquired from a single participant under visual stimulation.

Computing integrated information

Integrated information theory (IIT) has established itself as one of the leading theories for the study of consciousness. IIT essentially proposes that quantitative consciousness is identical to maximally integrated conceptual information, quantified by a measure called Φmax, and that phenomenological experience corresponds to the associated set of maximally irreducible cause-effect repertoires of a physical system being in a certain state. With the current work, we provide a general formulation of the framework, which comprehensively and parsimoniously expresses Φmax in the language of probabilistic models. Here, the stochastic process describing a system under scrutiny corresponds to a first-order time-invariant Markov process, and all necessary mathematical operations for the definition of Φmax are fully specified by a system's joint probability distribution over two adjacent points in discrete time. We present a detailed constructive rule for the decomposition of a system into two disjoint subsystems based on flexible marginalization and factorization of this joint distribution. Furthermore, we show that for a given joint distribution, virtualization is identical to a flexible factorization enforcing independence between variable subsets. We then validate our formulation in a previously established discrete example system, in which we also illustrate the previously unexplored theoretical issue of quale underdetermination due to non-unique maximally irreducible cause-effect repertoires. Moreover, we show that the current definition of Φ entails its sensitivity to the shape of the conceptual structure in qualia space, thus tying together IIT's measures of quantitative and qualitative consciousness, which we suggest be better disentangled. We propose several modifications of the framework in order to address some of these issues.

Dynamic spatiotemporal variability of alpha-BOLD relationships during the resting-state and task-evoked responses

Accurate characterization of the spatiotemporal relationship between two of the most prominent neuroimaging measures of neuronal activity, the 8-13Hz, occipito-parietal EEG alpha oscillation and the BOLD fMRI signal, must encompass the intrinsically dynamic nature of both alpha power and brain function. Here, during the eyes-open resting state, we use a 16s sliding-window analysis and demonstrate that the mean spatial network of dynamic alpha-BOLD correlations is highly comparable to the static network calculated over six minutes. However, alpha-BOLD correlations showed substantial spatiotemporal variability within-subjects and passed through many different configurations such that the static network was fully represented in only ~10% of 16s epochs, with visual and parietal regions (coherent on average) often opposingly correlated with each other or with alpha. We find that the common assumption of static-alpha BOLD correlations greatly oversimplifies temporal variation in brain network dynamics. Fluctuations in alpha-BOLD coupling significantly depended upon the instantaneous amplitude of alpha power, and primary and lateral visual areas were most strongly negatively correlated with alpha during different alpha power states, possibly suggesting the action of multiple alpha mechanisms. Dynamic alpha-BOLD correlations could not be explained by eye-blinks/movements, head motion or non-neuronal physiological variability. Individual's mean alpha power and frequency were found to contribute to between-subject variability in alpha-BOLD correlations. Additionally, application to a visual stimulation dataset showed that dynamic alpha-BOLD correlations provided functional information pertaining to the brain's response to stimulation by exhibiting spatiotemporal fluctuations related to variability in the trial-by-trial BOLD response magnitude. Significantly weaker visual alpha-BOLD correlations were found both preceding and following small amplitude BOLD response trials compared to large response trials.

An algorithm for separation of mixed sparse and Gaussian sources

Independent component analysis (ICA) is a ubiquitous method for decomposing complex signal mixtures into a small set of statistically independent source signals. However, in cases in which the signal mixture consists of both nongaussian and Gaussian sources, the Gaussian sources will not be recoverable by ICA and will pollute estimates of the nongaussian sources. Therefore, it is desirable to have methods for mixed ICA/PCA which can separate mixtures of Gaussian and nongaussian sources. For mixtures of purely Gaussian sources, principal component analysis (PCA) can provide a basis for the Gaussian subspace. We introduce a new method for mixed ICA/PCA which we call Mixed ICA/PCA via Reproducibility Stability (MIPReSt). Our method uses a repeated estimations technique to rank sources by reproducibility, combined with decomposition of multiple subsamplings of the original data matrix. These multiple decompositions allow us to assess component stability as the size of the data matrix changes, which can be used to determinine the dimension of the nongaussian subspace in a mixture. We demonstrate the utility of MIPReSt for signal mixtures consisting of simulated sources and real-word (speech) sources, as well as mixture of unknown composition.

Global signal modulation of single-trial fMRI response variability: Effect on positive vs negative BOLD response relationship

In functional magnetic resonance imaging (fMRI), the relationship between positive BOLD responses (PBRs) and negative BOLD responses (NBRs) to stimulation is potentially informative about the balance of excitatory and inhibitory brain responses in sensory cortex. In this study, we performed three separate experiments delivering visual, motor or somatosensory stimulation unilaterally, to one side of the sensory field, to induce PBR and NBR in opposite brain hemispheres. We then assessed the relationship between the evoked amplitudes of contralateral PBR and ipsilateral NBR at the level of both single-trial and average responses. We measure single-trial PBR and NBR peak amplitudes from individual time-courses, and show that they were positively correlated in all experiments. In contrast, in the average response across trials the absolute magnitudes of both PBR and NBR increased with increasing stimulus intensity, resulting in a negative correlation between mean response amplitudes. Subsequent analysis showed that the amplitude of single-trial PBR was positively correlated with the BOLD response across all grey-matter voxels and was not specifically related to the ipsilateral sensory cortical response. We demonstrate that the global component of this single-trial response modulation could be fully explained by voxel-wise vascular reactivity, the BOLD signal standard deviation measured in a separate resting-state scan (resting state fluctuation amplitude, RSFA). However, bilateral positive correlation between PBR and NBR regions remained. We further report that modulations in the global brain fMRI signal cannot fully account for this positive PBR-NBR coupling and conclude that the local sensory network response reflects a combination of superimposed vascular and neuronal signals. More detailed quantification of physiological and noise contributions to the BOLD signal is required to fully understand the trial-by-trial PBR and NBR relationship compared with that of average responses.

Early haemodynamic changes observed in patients with epilepsy, in a visual experiment and in simulations

OBJECTIVE

The objective of this study was to investigate whether previously reported early blood oxygen level dependent (BOLD) changes in epilepsy could occur as a result of the modelling techniques rather than physiological changes.

METHODS

EEG-fMRI data were analysed from seven patients with focal epilepsy, six control subjects undergoing a visual experiment, in addition to simulations. In six separate analyses the event timing was shifted by either -9,-6,-3,+3,+6 or +9 s relative to the onset of the interictal epileptiform discharge (IED) or stimulus.

RESULTS

The visual dataset and simulations demonstrated an overlap between modelled haemodynamic response function (HRF) at event onset and at ± 3 s relative to onset, which diminished at ± 6s. Pre-spike analysis at -6s improved concordance with the assumed IED generating lobe relative to the standard HRF in 43% of patients.

CONCLUSION

The visual and simulated dataset findings indicate a form of "temporal bleeding", an overlap between the modelled HRF at time 0 and at ± 3s which attenuated at ± 6s. Pre-spike analysis at -6s may improve concordance.

SIGNIFICANCE

This form of analysis should be performed at 6s prior to onset of IED to minimise temporal bleeding effect. The results support the presence of relevant BOLD responses occurring prior to IEDs.

Reliability of information-based integration of EEG and fMRI data: a simulation study

Most studies involving simultaneous electroencephalographic (EEG) and functional magnetic resonance imaging (fMRI) data rely on the first-order, affine-linear correlation of EEG and fMRI features within the framework of the general linear model. An alternative is the use of information-based measures such as mutual information and entropy, which can also detect higher-order correlations present in the data. The estimate of information-theoretic quantities might be influenced by several parameters, such as the numerosity of the sample, the amount of correlation between variables, and the discretization (or binning) strategy of choice. While these issues have been investigated for invasive neurophysiological data and a number of bias-correction estimates have been developed, there has been no attempt to systematically examine the accuracy of information estimates for the multivariate distributions arising in the context of EEG-fMRI recordings. This is especially important given the differences between electrophysiological and EEG-fMRI recordings. In this study, we drew random samples from simulated bivariate and trivariate distributions, mimicking the statistical properties of EEG-fMRI data. We compared the estimated information shared by simulated random variables with its numerical value and found that the interaction between the binning strategy and the estimation method influences the accuracy of the estimate. Conditional on the simulation assumptions, we found that the equipopulated binning strategy yields the best and most consistent results across distributions and bias correction methods. We also found that within bias correction techniques, the asymptotically debiased (TPMC), the jackknife debiased (JD), and the best upper bound (BUB) approach give similar results, and those are consistent across distributions.

Single-Trial Visual Evoked Potential Extraction Using Partial Least-Squares-Based Approach

A single-trial extraction of a visual evoked potential (VEP) signal based on the partial least-squares (PLS) regression method has been proposed in this paper. This paper has focused on the extraction and estimation of the latencies of P100, P200, P300, N75, and N135 in the artificial electroencephalograph (EEG) signal. The real EEG signal obtained from the hospital was only concentrated on the P100. The performance of the PLS has been evaluated mainly on the basis of latency error rate of the peaks for the artificial EEG signal, and the mean peak detection and standard deviation for the real EEG signal. The simulation results show that the proposed PLS algorithm is capable of reconstructing the EEG signal into its desired shape of the ideal VEP. For P100, the proposed PLS algorithm is able to provide comparable results to the generalized eigenvalue decomposition (GEVD) algorithm, which alters (prewhitens) the EEG input signal using the prestimulation EEG signal. It has also shown better performance for later peaks (P200 and P300). The PLS outperformed not only in positive peaks but also in N75. In P100, the PLS was comparable with the GEVD although N135 was better estimated by GEVD. The proposed PLS algorithm is comparable to GEVD given that PLS does not alter the EEG input signal. The PLS algorithm gives the best estimate to multitrial ensemble averaging. This research offers benefits such as avoiding patient's fatigue during VEP test measurement in the hospital, in BCI applications and in EEG-fMRI integration.

Cortical response variation with different sound pressure levels: a combined event-related potentials and FMRI study

INTRODUCTION

Simultaneous recording of electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) provides high spatial and temporal resolution. In this study we combined EEG and fMRI to investigate the structures involved in the processing of different sound pressure levels (SPLs).

METHODS

EEG data were recorded simultaneously with fMRI from 16 healthy volunteers using MR compatible devices at 3 T. Tones with different SPLs were delivered to the volunteers and the N1/P2 amplitudes were included as covariates in the fMRI data analysis in order to compare the structures activated with high and low SPLs. Analysis of variance (ANOVA) and ROI analysis were also performed. Additionally, source localisation analysis was performed on the EEG data.

RESULTS

The integration of averaged ERP parameters into the fMRI analysis showed an extended map of areas exhibiting covariation with the BOLD signal related to the auditory stimuli. The ANOVA and ROI analyses also revealed additional brain areas other than the primary auditory cortex (PAC) which were active with the auditory stimulation at different SPLs. The source localisation analyses showed additional sources apart from the PAC which were active with the high SPLs.

DISCUSSION

The PAC and the insula play an important role in the processing of different SPLs. In the fMRI analysis, additional activation was found in the anterior cingulate cortex, opercular and orbito-frontal cortices with high SPLs. A strong response of the visual cortex was also found with the high SPLs, suggesting the presence of cross-modal effects.

Reproducible paired sources from concurrent EEG-fMRI data using BICAR

We introduce BICAR, an algorithm for obtaining robust, reproducible pairs of temporal and spatial components at the individual subject level from concurrent electroencephalographic and functional magnetic resonance imaging data. BICAR assigns a task-independent measure of component quality, reproducibility, to each paired source. Under BICAR a reproducibility cutoff is derived that can be used to objectively discard spuriously paired EEG-fMRI components. BICAR is run on minimally processed data: fMRI images undergo the standard preprocessing steps (alignment, motion correction, etc.) and EEG data, after scanner artifact removal, are simply bandpass filtered. This minimal processing allows the secondary scoring of the same set of BICAR components for a variety of different endpoint analyses; in this manuscript we propose a general method for scoring components for task event synchronization (evoked response analysis), but scoring using many other criteria, for example frequency content, are possible. BICAR is applied to five subjects performing a visual search task, and among the most reproducible components we find biologically relevant paired sources involved in visual processing, motor planning, execution, and attention.

EEG-fMRI integration for the study of human brain function

Electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) have proved to be extremely valuable tools for the non-invasive study of human brain function. Moreover, due to a notable degree of complementarity between the two modalities, the combination of EEG and fMRI data has been actively sought in the last two decades. Although initially focused on epilepsy, EEG-fMRI applications were rapidly extended to the study of healthy brain function, yielding new insights into its underlying mechanisms and pathways. Nevertheless, EEG and fMRI have markedly different spatial and temporal resolutions, and probe neuronal activity through distinct biophysical processes, many aspects of which are still poorly understood. The remarkable conceptual and methodological challenges associated with EEG-fMRI integration have motivated the development of a wide range of analysis approaches over the years, each relying on more or less restrictive assumptions, and aiming to shed further light on the mechanisms of brain function along with those of the EEG-fMRI coupling itself. Here, we present a review of the most relevant EEG-fMRI integration approaches yet proposed for the study of brain function, supported by a general overview of our current understanding of the biophysical mechanisms coupling the signals obtained from the two modalities.

Recording visual evoked potentials and auditory evoked P300 at 9.4T static magnetic field

Simultaneous recording of electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) has shown a number of advantages that make this multimodal technique superior to fMRI alone. The feasibility of recording EEG at ultra-high static magnetic field up to 9.4T was recently demonstrated and promises to be implemented soon in fMRI studies at ultra high magnetic fields. Recording visual evoked potentials are expected to be amongst the most simple for simultaneous EEG/fMRI at ultra-high magnetic field due to the easy assessment of the visual cortex. Auditory evoked P300 measurements are of interest since it is believed that they represent the earliest stage of cognitive processing. In this study, we investigate the feasibility of recording visual evoked potentials and auditory evoked P300 in a 9.4T static magnetic field. For this purpose, EEG data were recorded from 26 healthy volunteers inside a 9.4T MR scanner using a 32-channel MR compatible EEG system. Visual stimulation and auditory oddball paradigm were presented in order to elicit evoked related potentials (ERP). Recordings made outside the scanner were performed using the same stimuli and EEG system for comparison purposes. We were able to retrieve visual P100 and auditory P300 evoked potentials at 9.4T static magnetic field after correction of the ballistocardiogram artefact using independent component analysis. The latencies of the ERPs recorded at 9.4T were not different from those recorded at 0T. The amplitudes of ERPs were higher at 9.4T when compared to recordings at 0T. Nevertheless, it seems that the increased amplitudes of the ERPs are due to the effect of the ultra-high field on the EEG recording system rather than alteration in the intrinsic processes that generate the electrophysiological responses.

EEG acquisition in ultra-high static magnetic fields up to 9.4 T

The simultaneous acquisition of electroencephalographic (EEG) and functional magnetic resonance imaging (fMRI) data has gained momentum in recent years due to the synergistic effects of the two modalities with regard to temporal and spatial resolution. Currently, only EEG-data recorded in fields of up to 7 T have been reported. We investigated the feasibility of recording EEG inside a 9.4 T static magnetic field, specifically to determine whether meaningful EEG information could be recovered from the data after removal of the cardiac-related artefact. EEG-data were recorded reliably and reproducibly at 9.4 T and the cardiac-related artefact increased in amplitude with increasing B0, as expected. Furthermore, we were able to correct for the cardiac-related artefact and identify auditory event related responses at 9.4 T in 75% of subjects using independent component analysis (ICA). Also by means of ICA we detected event related spectral perturbations (ERSP) in subjects at 9.4 T in response to opening/closing the eyes comparable with the response at 0 T. Overall our results suggest that it is possible to record meaningful EEG data at ultra-high magnetic fields. The simultaneous EEG-fMRI approach at ultra-high-fields opens up the horizon for investigating brain dynamics at a superb spatial resolution and a temporal resolution in the millisecond domain.

Mapping interictal epileptic discharges using mutual information between concurrent EEG and fMRI

OBJECTIVE

The mapping of haemodynamic changes related to interictal epileptic discharges (IED) in simultaneous electroencephalography (EEG) and functional MRI (fMRI) studies is usually carried out by means of EEG-correlated fMRI analyses where the EEG information specifies the model to test on the fMRI signal. The sensitivity and specificity critically depend on the accuracy of EEG detection and the validity of the haemodynamic model. In this study we investigated whether an information theoretic analysis based on the mutual information (MI) between the presence of epileptic activity on EEG and the fMRI data can provide further insights into the haemodynamic changes related to interictal epileptic activity. The important features of MI are that: 1) both recording modalities are treated symmetrically; 2) no requirement for a-priori models for the haemodynamic response function, or assumption of a linear relationship between the spiking activity and BOLD responses, and 3) no parametric model for the type of noise or its probability distribution is necessary for the computation of MI.

METHODS

Fourteen patients with pharmaco-resistant focal epilepsy underwent EEG-fMRI and intracranial EEG and/or surgical resection with positive postoperative outcome (seizure freedom or considerable reduction in seizure frequency) was available in 7/14 patients. We used nonparametric statistical assessment of the MI maps based on a four-dimensional wavelet packet resampling method. The results of MI were compared to the statistical parametric maps obtained with two conventional General Linear Model (GLM) analyses based on the informed basis set (canonical HRF and its temporal and dispersion derivatives) and the Finite Impulse Response (FIR) models.

RESULTS

The MI results were concordant with the electro-clinically or surgically defined epileptogenic area in 8/14 patients and showed the same degree of concordance as the results obtained with the GLM-based methods in 12 patients (7 concordant and 5 discordant). In one patient, the information theoretic analysis improved the delineation of the irritative zone compared with the GLM-based methods.

DISCUSSION

Our findings suggest that an information theoretic analysis can provide clinically relevant information about the BOLD signal changes associated with the generation and propagation of interictal epileptic discharges. The concordance between the MI, GLM and FIR maps support the validity of the assumptions adopted in GLM-based analyses of interictal epileptic activity with EEG-fMRI in such a manner that they do not significantly constrain the localization of the epileptogenic zone.

EEG-fMRI based information theoretic characterization of the human perceptual decision system

The modern metaphor of the brain is that of a dynamic information processing device. In the current study we investigate how a core cognitive network of the human brain, the perceptual decision system, can be characterized regarding its spatiotemporal representation of task-relevant information. We capitalize on a recently developed information theoretic framework for the analysis of simultaneously acquired electroencephalography (EEG) and functional magnetic resonance imaging data (fMRI) (Ostwald et al. (2010), NeuroImage 49: 498-516). We show how this framework naturally extends from previous validations in the sensory to the cognitive domain and how it enables the economic description of neural spatiotemporal information encoding. Specifically, based on simultaneous EEG-fMRI data features from n = 13 observers performing a visual perceptual decision task, we demonstrate how the information theoretic framework is able to reproduce earlier findings on the neurobiological underpinnings of perceptual decisions from the response signal features' marginal distributions. Furthermore, using the joint EEG-fMRI feature distribution, we provide novel evidence for a highly distributed and dynamic encoding of task-relevant information in the human brain.

Analysis of time and space invariance of BOLD responses in the rat visual system

Neuroimaging studies of functional magnetic resonance imaging (fMRI) and electrophysiology provide the linkage between neural activity and the blood oxygenation level-dependent (BOLD) response. Here, BOLD responses to light flashes were imaged at 11.7T and compared with neural recordings from superior colliculus (SC) and primary visual cortex (V1) in rat brain--regions with different basal blood flow and energy demand. Our goal was to assess neurovascular coupling in V1 and SC as reflected by temporal/spatial variances of impulse response functions (IRFs) and assess, if any, implications for general linear modeling (GLM) of BOLD responses. Light flashes induced high magnitude neural/BOLD responses reproducibly from both regions. However, neural/BOLD responses from SC and V1 were markedly different. SC signals followed the boxcar shape of the stimulation paradigm at all flash rates, whereas V1 signals were characterized by onset/offset transients that exhibited different flash rate dependencies. We find that IRF(SC) is generally time-invariant across wider flash rate range compared with IRF(V1), whereas IRF(SC) and IRF(V1) are both space invariant. These results illustrate the importance of measured neural signals for interpretation of fMRI by showing that GLM of BOLD responses may lead to misinterpretation of neural activity in some cases.

Optimal band separation of extracellular field potentials

Local Field Potentials (LFPs) exhibit a broadband spectral structure that is traditionally partitioned into distinct frequency bands which are thought to originate from different types of neural events triggered by different processing pathways. However, the exact frequency boundaries of these processes are not known and, as a result, the frequency bands are often selected based on intuition, previous literature or visual inspection of the data. Here, we address these problems by developing a rigorous method for defining LFP frequency bands and their boundaries. The criterion introduced for determining the boundaries delimiting the bands is to maximize the information about an external correlate carried jointly by all bands in the partition. The method first partitions the LFP frequency range into two bands and then successively increases the number of bands in the partition. We applied the partitioning method to LFPs recorded from primary visual cortex of anaesthetized macaques, and we determined the optimal band partitioning that describes the encoding of naturalistic visual stimuli. The first optimal boundary partitioned the LFP response at 60 Hz into low and high frequencies, which had been previously found to convey independent information about the natural movie correlate. The second optimal boundary divided the high-frequency range at approximately 100 Hz into gamma and high-gamma frequencies, consistent with recent reports that these two bands reflect partly distinct neural processes. A third important boundary was at 25 Hz and it split the LFP range below 50 Hz into a stimulus-informative and a stimulus-independent band.

Multimodal functional network connectivity: an EEG-fMRI fusion in network space

EEG and fMRI recordings measure the functional activity of multiple coherent networks distributed in the cerebral cortex. Identifying network interaction from the complementary neuroelectric and hemodynamic signals may help to explain the complex relationships between different brain regions. In this paper, multimodal functional network connectivity (mFNC) is proposed for the fusion of EEG and fMRI in network space. First, functional networks (FNs) are extracted using spatial independent component analysis (ICA) in each modality separately. Then the interactions among FNs in each modality are explored by Granger causality analysis (GCA). Finally, fMRI FNs are matched to EEG FNs in the spatial domain using network-based source imaging (NESOI). Investigations of both synthetic and real data demonstrate that mFNC has the potential to reveal the underlying neural networks of each modality separately and in their combination. With mFNC, comprehensive relationships among FNs might be unveiled for the deep exploration of neural activities and metabolic responses in a specific task or neurological state.

Motion-related artefacts in EEG predict neuronally plausible patterns of activation in fMRI data

The simultaneous acquisition and subsequent analysis of EEG and fMRI data is challenging owing to increased noise levels in the EEG data. A common method to integrate data from these two modalities is to use aspects of the EEG data, such as the amplitudes of event-related potentials (ERP) or oscillatory EEG activity, to predict fluctuations in the fMRI data. However, this relies on the acquisition of high quality datasets to ensure that only the correlates of neuronal activity are being studied. In this study, we investigate the effects of head-motion-related artefacts in the EEG signal on the predicted T2*-weighted signal variation. We apply our analyses to two independent datasets: 1) four participants were asked to move their feet in the scanner to generate small head movements, and 2) four participants performed an episodic memory task. We created T2*-weighted signal predictors from indicators of abrupt head motion using derivatives of the realignment parameters, from visually detected artefacts in the EEG as well as from three EEG frequency bands (theta, alpha and beta). In both datasets, we found little correlation between the T2*-weighted signal and EEG predictors that were not convolved with the canonical haemodynamic response function (cHRF). However, all convolved EEG predictors strongly correlated with the T2*-weighted signal variation in various regions including the bilateral superior temporal cortex, supplementary motor area, medial parietal cortex and cerebellum. The finding that movement onset spikes in the EEG predict T2*-weighted signal intensity only when the time course of movements is convolved with the cHRF, suggests that the correlated signal might reflect a BOLD response to neural activity associated with head movement. Furthermore, the observation that broad-spectral EEG spikes tend to occur at the same time as abrupt head movements, together with the finding that abrupt movements and EEG spikes show similar correlations with the T2*-weighted signal, indicates that the EEG spikes are produced by abrupt movement and that continuous regressors of EEG oscillations contain motion-related noise even after stringent correction of the EEG data. If not properly removed, these artefacts complicate the use of EEG data as a predictor of T2*-weighted signal variation.

Investigating static nonlinearities in neurovascular coupling

Many statistical models of coupling between time changes of the band-limited power of neural signals and functional magnetic resonance imaging Blood Oxygenation Level Dependent (BOLD) signal time changes rely on linear convolution. The effect of nonlinear behaviors in single-trial relationships between neural signals and BOLD responses is rarely tested and included in models. Here we investigate whether using a static nonlinearity improves the prediction of single-trial BOLD responses from neural signals. A static nonlinearity is a nonlinear transformation of the convolution of neural responses which is implemented by the same nonlinear function for all time points. We evaluated this approach by applying it to simultaneous recordings of functional magnetic resonance imaging BOLD and band-limited neural signals (Local Field Potentials and Multi Unit Activity) from primary visual cortex of anaesthetized macaques. We found that using a simple polynomial static nonlinearity was sufficient to obtain highly significant improvements of the accuracy of single-trial BOLD prediction over the accuracy obtained with linear convolution. This suggests that static nonlinearities may be a useful tool for a compact and accurate statistical description of neurovascular coupling.

The relationship between the visual evoked potential and the gamma band investigated by blind and semi-blind methods

Gamma Band Activity (GBA) is increasingly studied for its relation with attention, change detection, maintenance of working memory and the processing of sensory stimuli. Activity around the gamma range has also been linked with early visual processing, although the relationship between this activity and the low frequency visual evoked potential (VEP) remains unclear. This study examined the ability of blind and semi-blind source separation techniques to extract sources specifically related to the VEP and GBA in order to shed light on the relationship between them. Blind (Independent Component Analysis—ICA) and semi-Blind (Functional Source Separation—FSS) methods were applied to dense array EEG data recorded during checkerboard stimulation. FSS was performed with both temporal and spectral constraints to identify specifically the generators of the main peak of the VEP (P100) and of the GBA. Source localisation and time-frequency analyses were then used to investigate the properties and co-dependencies between VEP/P100 and GBA. Analysis of the VEP extracted using the different methods demonstrated very similar morphology and localisation of the generators. Single trial time frequency analysis showed higher GBA when a larger amplitude VEP/P100 occurred. Further examination indicated that the evoked (phase-locked) component of the GBA was more related to the P100, whilst the induced component correlated with the VEP as a whole. The results suggest that the VEP and GBA may be generated by the same neuronal populations, and implicate this relationship as a potential mediator of the correlation between the VEP and the Blood Oxygenation Level Dependent (BOLD) effect measured with fMRI.

Combining EEG and fMRI

The combination of electroencephalography (EEG) with functional magnetic resonance imaging (fMRI) forms a powerful tool for the investigation of brain function, but concurrent implementation of EEG and fMRI poses many technical challenges. Here, the motivation for combining EEG and fMRI is explored and methods underlying the combination are described. After a brief introduction to the two different techniques, the advantages and disadvantages of different methods of data recording are detailed, followed by a description of the artefacts encountered when performing EEG and fMRI measurements simultaneously, and the methods which have been developed to eliminate these artefacts. Important safety considerations and potential pitfalls associated with simultaneous recording are also described. The ways in which EEG and fMRI data analysis can be integrated are then described along with examples of key work which illustrate the power of combined EEG/fMRI measurements. The chapter concludes with a brief discussion of future directions for combined EEG/fMRI research.

Voxel-wise information theoretic EEG-fMRI feature integration

We have recently proposed the evaluation of a set of information theoretic quantities (ITQs) for the integration of simultaneously acquired EEG-fMRI data (Ostwald, D., Porcaro, C., Bagshaw, A.P., 2010. An information theoretic approach to EEG-fMRI integration of visually evoked responses. Neuroimage. 49, 498-516). In our previous experimental evaluation of the information theoretic framework, we defined the data subsets from which to calculate the ITQs using a priori constraints. In the case of EEG, this meant that data were extracted from a single electrode, while for fMRI the analysed data came from voxels contained within a sphere surrounding the most responsive voxel of visual cortex. While this approach was a natural starting point for the evaluation of the framework in the application to combined EEG-fMRI data sets, a more principled approach to data selection is desirable. Here, we propose to combine standard fMRI data pre-processing and low-resolution electromagnetic tomography (LORETA) for the evaluation of ITQs across the entire three-dimensional brain space. We apply the proposed method to a simultaneous EEG-fMRI data set acquired during checkerboard stimulation and assess the topographical informativeness of EEG (time and frequency domain) and fMRI features with respect to the stimulus and each other. The resulting information theoretic effect size maps are supplemented with a statistical evaluation based on Gaussian null model simulations using a false-discovery rate procedure. Given the contamination of EEG recordings by artefacts induced by the MR scanning environment we further assessed the influence of different advanced EEG pre-processing methods (independent component analysis and functional source separation) on the information topography. The results of this analysis provide evidence for the topographically focussed informativeness of both EEG and fMRI features with respect to the stimulus, but for the current feature selection do not detect EEG-fMRI activity dependence. More advanced EEG data pre-processing rendered the feature distributions more stimulus-informative, but did not alter the EEG-fMRI activity and conditional dependencies.

MEG and fMRI Fusion for Non-Linear Estimation of Neural and BOLD Signal Changes

The combined analysis of magnetoencephalography (MEG)/electroencephalography and functional magnetic resonance imaging (fMRI) measurements can lead to improvement in the description of the dynamical and spatial properties of brain activity. In this paper we empirically demonstrate this improvement using simulated and recorded task related MEG and fMRI activity. Neural activity estimates were derived using a dynamic Bayesian network with continuous real valued parameters by means of a sequential Monte Carlo technique. In synthetic data, we show that MEG and fMRI fusion improves estimation of the indirectly observed neural activity and smooths tracking of the blood oxygenation level dependent (BOLD) response. In recordings of task related neural activity the combination of MEG and fMRI produces a result with greater signal-to-noise ratio, that confirms the expectation arising from the nature of the experiment. The highly non-linear model of the BOLD response poses a difficult inference problem for neural activity estimation; computational requirements are also high due to the time and space complexity. We show that joint analysis of the data improves the system's behavior by stabilizing the differential equations system and by requiring fewer computational resources.

Predicting EEG single trial responses with simultaneous fMRI and relevance vector machine regression

The combination of electroencephalography (EEG) and functional Magnetic Resonance Imaging (fMRI) has been proposed as a tool to study brain dynamics with both high temporal and high spatial resolution. Integration through EEG-fMRI trial-by-trial coupling has been proposed as a method to combine the different data sets and achieve temporal expansion of the fMRI data (Eichele et al., 2005). To fully benefit of this type of analysis simultaneous EEG-fMRI acquisitions are necessary (Debener et al., 2006). Here we address the issue of predicting the signal in one modality using information from the other modality. We use multivariate Relevance Vector Machine (RVM) regression to "learn" the relation between fMRI activation patterns and simultaneously acquired EEG responses in the context of a complex cognitive task entailing an auditory cue, visual mental imagery and a control visual target. We show that multivariate regression is a valuable approach for predicting evoked and induced oscillatory EEG responses from fMRI time series. Prediction of EEG from fMRI is largely influenced by the overall filtering effects of the hemodynamic response function. However, a detailed analysis of the auditory evoked responses shows that there is a small but significant contribution of single trial modulations that can be exploited for linking spatially-distributed patterns of fMRI activation to specific components of the simultaneously-recorded EEG signal.

Removal of ballistocardiogram artifacts using the cyclostationary source extraction method

Ballistocardiogram (BCG) artifact is considered here as the sum of a number of independent cyclostationary components having the same cycle frequency. Our proposed method, called cyclostationary source extraction (CSE), is able to extract these components without much destructive effect on the background electroencephalogram (EEG). It is shown that the proposed method outperforms other methods particularly in preserving the remaining signals. CSE is utilized to remove the BCG artifact from real EEG data recorded inside the magnetic resonance (MR) scanner, i.e., visual evoked potential (VEP). The results are compared to the results of benchmark BCG removal techniques. Analyzing the power spectral density of the cleaned EEG data, it is shown that CSE effectively removes the frequency components corresponding to the BCG artifact. It is also shown that VEPs recorded inside the scanner and processed using the proposed method are more correlated with the VEPs recorded outside the scanner. Moreover, there is no need for electrocardiogram (ECG) data in this method as the cycle frequency of the BCG is directly computed from the contaminated EEG signals.

Multimodal imaging: an evaluation of univariate and multivariate methods for simultaneous EEG/fMRI

The combination of electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) has been proposed as a tool to study brain dynamics with both high temporal and high spatial resolution. Multimodal imaging techniques rely on the assumption of a common neuronal source for the different recorded signals. In order to maximally exploit the combination of these techniques, one needs to understand the coupling (i.e., the relation) between electroencephalographic (EEG) and fMRI blood oxygen level-dependent (BOLD) signals. Recently, simultaneous EEG-fMRI measurements have been used to investigate the relation between the two signals. Previous attempts at the analysis of simultaneous EEG-fMRI data reported significant correlations between regional BOLD activations and modulation of both event-related potential (ERP) and oscillatory EEG power, mostly in the alpha but also in other frequency bands. Beyond the correlation of the two measured brain signals, the relevant issue we address here is the ability of predicting the signal in one modality using information from the other modality. Using multivariate machine learning-based regression, we show how it is possible to predict EEG power oscillations from simultaneously acquired fMRI data during an eyes-open/eyes-closed task using either the original channels or the underlying cortically distributed sources as the relevant EEG signal for the analysis of multimodal data.