EEG-Informed fMRI: A Review of Data Analysis Methods

Predictive value of functional MRI and EEG in epilepsy diagnosis after a first seizure

It is often difficult to predict seizure recurrence in subjects who have suffered a first-ever epileptic seizure. In this study, the predictive value of physiological signals measured using Electroencephalography (EEG) and functional MRI (fMRI) is assessed. In particular those patients developing epilepsy (i.e. a second unprovoked seizure) that were initially evaluated as having a low risk of seizure recurrence are of interest. In total, 26 epilepsy patients, of which 8 were initially evaluated as having a low risk of seizure recurrence (i.e. converters), and 17 subjects with only a single seizure were included. All subjects underwent routine EEG as well as fMRI measurements. For diagnostic classification, features related to the temporal dynamics were determined for both the processed EEG and fMRI data. Subsequently, a logistic regression classifier was trained on epilepsy and first-seizure subjects. The trained model was tested using the clinically relevant converters group. The sensitivity, specificity, and AUC (mean ± SD) of the regression model including metrics from both modalities were 74 ± 19%, 82 ± 18%, and 0.75 ± 0.12, respectively. Positive and negative predictive values (mean ± SD) of the regression model with both EEG and fMRI features are 84 ± 14% and 78 ± 12%. Moreover, this EEG/fMRI model showed significant improvements compared to the clinical diagnosis, whereas the models using metrics from either EEG or fMRI do not reach significance (p > 0.05). Temporal metrics computationally derived from EEG and fMRI time signals may clinically aid and synergistically improve the predictive value in a first-seizure sample.

Examining the effect of online advertisement cues on human responses using eye-tracking, EEG, and MRI

This study sought to emphasize how disciplines such as neuroscience and marketing can be applied in advertising and consumer behavior. The application of neuroscience methods in analyzing and understanding human behavior related to the Elaboration Likelihood Model (ELM) and brain activity has recently garnered attention. This study examines brain processes while participants attempted to elicit preferences for a product, and demonstrates factors that influence consumer behavior using eye-tracking, electroencephalography (EEG), and magnetic resonance imaging (MRI) from a neuroscience approach. We planned two conditions of online advertising, namely, peripheral cues without argument and central cues with argument strength. Thirty respondents participated in the experiment, consisting of eye-tracking, EEG, and MRI instruments to explore brain activity in central cue conditions. We investigated whether diffusion tensor imaging (DTI) analysis could detect regional brain changes. Using eye-tracking, we found that the responses were mainly in the mean fixation duration, number of fixations, mean saccade duration, and number of saccade durations for the central cue condition. Moreover, the findings show that the fusiform gyrus and frontal cortex are significantly associated with building a relationship by inferring central cues in the EEG assay. The MRI images show that the fusiform gyrus and frontal cortex are significantly active in the central cue condition. DTI analysis indicates that the corpus callosum has changed in the central cue condition. We used eye-tracking, EEG, MRI, and DTI to understand that these connections may apprehend responses when viewing advertisements, especially in the fusiform gyrus, frontal cortex, and corpus callosum.

EEG Microstates Predict Concurrent fMRI Dynamic Functional Connectivity States

Brain functional connectivity measured by resting-state fMRI varies over multiple time scales, and recurrent dynamic functional connectivity (dFC) states have been identified. These have been found to be associated with different cognitive and pathological states, with potential as disease biomarkers, but their neuronal underpinnings remain a matter of debate. A number of recurrent microstates have also been identified in resting-state EEG studies, which are thought to represent the quasi-simultaneous activity of large-scale functional networks reflecting time-varying brain states. Here, we hypothesized that fMRI-derived dFC states may be associated with these EEG microstates. To test this hypothesis, we quantitatively assessed the ability of EEG microstates to predict concurrent fMRI dFC states in simultaneous EEG-fMRI data collected from healthy subjects at rest. By training a random forests classifier, we found that the four canonical EEG microstates predicted fMRI dFC states with an accuracy of 90%, clearly outperforming alternative EEG features such as spectral power. Our results indicate that EEG microstates analysis yields robust signatures of fMRI dFC states, providing evidence of the electrophysiological underpinnings of dFC while also further supporting that EEG microstates reflect the dynamics of large-scale brain networks.

Electroencephalographic Correlates and Predictors of Treatment Outcome in OCD: A Brief Narrative Review

Electroencephalography (EEG) measures the brain's electrical activity with high temporal resolution. In comparison to neuroimaging modalities such as MRI or PET, EEG is relatively cheap, non-invasive, portable, and simple to administer, making it an attractive tool for clinical deployment. Despite this, studies utilizing EEG to investigate obsessive-compulsive disorder (OCD) are relatively sparse. This contrasts with a robust literature using other brain imaging methodologies. The present review examines studies that have used EEG to examine predictors and correlates of response in OCD and draws tentative conclusions that may guide much needed future work. Key findings include a limited literature base; few studies have attempted to predict clinical change from EEG signals, and they are confounded by the effects of both pharmacotherapy and psychotherapy. The most robust literature, consisting of several studies, has examined event-related potentials, including the P300, which several studies have reported to be abnormal at baseline in OCD and to normalize with treatment; but even here the literature is quite heterogeneous, and more work is needed. With more robust research, we suggest that the relatively low cost and convenience of EEG, especially in comparison to fMRI and PET, make it well-suited to the development of feasible personalized treatment algorithms.

Schrödinger filtering: a precise EEG despiking technique for EEG-fMRI gradient artifact

In EEG data acquired in the presence of fMRI, gradient-related spike artifacts contaminate the signal following the common preprocessing step of average artifact subtraction. Spike artifacts compromise EEG data quality since they overlap with the EEG signal in frequency, thereby confounding frequency-based inferences on activity. As well, spike artifacts can inflate or deflate correlations among time series, thereby confounding inferences on functional connectivity. We present Schrödinger filtering, which uses the Schrödinger equation to decompose the spike-containing input. The basis functions of the decomposition are localized and pulse-shaped, and selectively capture the various input peaks, with the spike components clustered at the beginning of the spectrum. Schrödinger filtering automatically subtracts the spike components from the data. On real and simulated data, we show that Schrödinger filtering (1) simultaneously accomplishes high spike removal and high signal preservation without affecting evoked activity, and (2) reduces spurious pairwise correlations in spontaneous activity. In these regards, Schrödinger filtering was significantly better than three other despiking techniques: median filtering, amplitude thresholding, and wavelet denoising. These results encourage the use of Schrödinger filtering in future EEG-fMRI pipelines, as well as in other spike-related applications (e.g., fMRI motion artifact removal or action potential extraction).

EEG Microstates Predict Concurrent fMRI Dynamic Functional Connectivity States

Brain functional connectivity measured by resting-state fMRI varies over multiple time scales, and recurrent dynamic functional connectivity (dFC) states have been identified. These have been found to be associated with different cognitive and pathological states, with potential as disease biomarkers, but their neuronal underpinnings remain a matter of debate. A number of recurrent microstates have also been identified in resting-state EEG studies, which are thought to represent the quasi-simultaneous activity of large-scale functional networks reflecting time-varying brain states. Here, we hypothesized that fMRI-derived dFC states may be associated with these EEG microstates. To test this hypothesis, we quantitatively assessed the ability of EEG microstates to predict concurrent fMRI dFC states in simultaneous EEG-fMRI data collected from healthy subjects at rest. By training a random forests classifier, we found that the four canonical EEG microstates predicted fMRI dFC states with an accuracy of 90%, clearly outperforming alternative EEG features such as spectral power. Our results indicate that EEG microstates analysis yields robust signatures of fMRI dFC states, providing evidence of the electrophysiological underpinnings of dFC while also further supporting that EEG microstates reflect the dynamics of large-scale brain networks.

Effects of Neural Mechanisms of Pretask Resting EEG Alpha Information on Situational Awareness: A Functional Connectivity Approach

OBJECTIVE

In this study, the influence of pretask resting neural mechanisms on situational awareness (SA)-task is studied.

BACKGROUND

Pretask electroencephalography (EEG) information and Stroop effect are known to influence task engagement independently. However, neural mechanisms of pretask resting absolute alpha (PRAA) and pretask resting alpha frontal asymmetry (PRAFA) in influencing SA-task which is undergoing Stroop effect is still not understood.

METHOD

The study involved pretask resting EEG measurements from 18 healthy individuals followed by functional magnetic resonance imaging (fMRI) acquisition during SA-task. To understand the effect of pretask alpha information and Stroop effect on SA, a robust correlation between mean reaction time, SA Index, PRAA, and PRAFA were assessed. Furthermore, neural underpinnings of PRAA, PRAFA in SA-task, and functional connectivity were analyzed through the EEG-informed fMRI approach.

RESULTS

Significant robust correlation of reaction time was observed with SA Index (Pearson: r = .50, pcorr = .05) and PRAFA (Pearson: r = .63; pcorr = .01), respectively. Similarly, SA Index significantly correlated with PRAFA (Pearson: r = .56, pcorr = .01; Spearman: r = .61, pcorr = .007), and PRAA (Pearson: r = .59, pcorr = .005; Spearman: r = .59, pcorr = .002). Neural underpinnings of SA-task revealed regions involved in visual-processing and higher-order cognition. PRAA was primarily underpinned at frontal-temporal areas and functionally connected to SA-task regions pertaining to the emotional regulation. PRAFA has correlated with limbic and parietal regions, which are involved in integration of visual, emotion, and memory information of SA-task.

CONCLUSION

The results suggest a strong association of reaction time with SA-task and PRAFA and strongly support the hypothesis that PRAFA, PRAA, and associated neural mechanisms significantly influence the outcome of SA-task.

APPLICATION

It is beneficial to study the effect of pretask resting information on SA-task to improve SA.

ESB: A low-cost EEG Synchronization Box

Electroencephalography (EEG) is a neuroimaging technique with a temporal resolution in the millisecond scale. Popular ERPs and ERD/ERS functions, as well as EEG-fRMI data and hyperscanning methods requires a proper temporal alignment (namely, synchronization) with stimulus onsets and other devices. Hardware-based synchronization, based on a SYNC signal injected into the device, ensures a reliable timing. In this paper we describe the design, test and validation of an EEG Synchronization Box (ESB), able to condition and distribute a SYNC signal (analog and digital) to different devices simultaneously. ESB can be easily built by individuals with basic soldering skills and represents a cost-effective solution to the available commercial synchronization boxes, while preserving similar electrical and functional features.

Modeling an auditory stimulated brain under altered states of consciousness using the generalized Ising model

Propofol is a short-acting medication that results in decreased levels of consciousness and is used for general anesthesia. Although it is the most commonly used anesthetic in the world, much remains unknown about the mechanisms by which it induces a loss of consciousness. Characterizing anesthesia-induced alterations to brain network activity might provide a powerful framework for understanding the neural mechanisms of unconsciousness. The aim of this work was to model brain activity in healthy brains during various stages of consciousness, as induced by propofol, in the auditory paradigm. We used the generalized Ising model (GIM) to fit the empirical fMRI data of healthy subjects while they listened to an audio clip from a movie. The external stimulus (audio clip) is believed to be at least partially driving a synchronization process of the brain activity and provides a similar conscious experience in different subjects. In order to observe the common synchronization among the subjects, a novel technique called the inter subject correlation (ISC) was implemented. We showed that the GIM-modified to incorporate the naturalistic external field-was able to fit the empirical task fMRI data in the awake state, in mild sedation, in deep sedation, and in recovery, at a temperature T* which is well above the critical temperature. To our knowledge this is the first study that captures human brain activity in response to real-life external stimuli at different levels of conscious awareness using mathematical modeling. This study might be helpful in the future to assess the level of consciousness of patients with disorders of consciousness and help in regaining their consciousness.

Loss of frontal regulator of vigilance during sleep inertia: A simultaneous EEG-fMRI study

Sleep inertia refers to a distinct physiological state of waking up from sleep accompanied by performance impairments and sleepiness. The neural substrates of sleep inertia are unknown, but growing evidence suggests that this inertia state maintains certain sleep features. To investigate the neurophysiological mechanisms of sleep inertia, a comparison of pre-sleep and post-sleep wakefulness with eyes-open resting-state was performed using simultaneous EEG-fMRI, which has the potential to reveal the dynamic details of neuroelectric and hemodynamic responses with high temporal resolution. Our data suggested sleep-like features of slow EEG power and decreased BOLD activity were persistent during sleep inertia. In the pre-sleep phase, participants with stronger EEG vigilance showed stronger activity in the fronto-parietal network (FPN), but this phenomenon disappeared during sleep inertia. A time course analysis confirmed a decreased correlation between EEG vigilance and the FPN activity during sleep inertia. This simultaneous EEG-fMRI study advanced our understanding of sleep inertia and revealed the importance of the FPN in maintaining awareness. This is the first study to reveal the dynamic brain network changes from multi-modalities perspective during sleep inertia.

Pushing the Limits of EEG: Estimation of Large-Scale Functional Brain Networks and Their Dynamics Validated by Simultaneous fMRI

Functional magnetic resonance imaging (fMRI) is the technique of choice for detecting large-scale functional brain networks and to investigate their dynamics. Because fMRI measures brain activity indirectly, electroencephalography (EEG) has been recently considered a feasible tool for detecting such networks, particularly the resting-state networks (RSNs). However, a truly unbiased validation of such claims is still missing, which can only be accomplished by using simultaneously acquired EEG and fMRI data, due to the spontaneous nature of the activity underlying the RSNs. Additionally, EEG is still poorly explored for the purpose of mapping task-specific networks, and no studies so far have been focused on investigating networks' dynamic functional connectivity (dFC) with EEG. Here, we started by validating RSNs derived from the continuous reconstruction of EEG sources by directly comparing them with those derived from simultaneous fMRI data of 10 healthy participants, and obtaining an average overlap (quantified by the Dice coefficient) of 0.4. We also showed the ability of EEG to map the facial expressions processing network (FEPN), highlighting regions near the posterior superior temporal sulcus, where the FEPN is anchored. Then, we measured the dFC using EEG for the first time in this context, estimated dFC brain states using dictionary learning, and compared such states with those obtained from the fMRI. We found a statistically significant match between fMRI and EEG dFC states, and determined the existence of two matched dFC states which contribution over time was associated with the brain activity at the FEPN, showing that the dynamics of FEPN can be captured by both fMRI and EEG. Our results push the limits of EEG toward being used as a brain imaging tool, while supporting the growing literature on EEG correlates of (dynamic) functional connectivity measured with fMRI, and providing novel insights into the coupling mechanisms underlying the two imaging techniques.

Functional Brain Connectivity of Language Functions in Children Revealed by EEG and MEG: A Systematic Review

The development of language functions is of great interest to neuroscientists, as these functions are among the fundamental capacities of human cognition. For many years, researchers aimed at identifying cerebral correlates of language abilities. More recently, the development of new data analysis tools has generated a shift toward the investigation of complex cerebral networks. In 2015, Weiss-Croft and Baldeweg published a very interesting systematic review on the development of functional language networks, explored through the use of functional magnetic resonance imaging (fMRI). Compared to fMRI and because of their excellent temporal resolution, magnetoencephalography (MEG) and electroencephalography (EEG) provide different and important information on brain activity. Both therefore constitute crucial neuroimaging techniques for the investigation of the maturation of functional language brain networks. The main objective of this systematic review is to provide a state of knowledge on the investigation of language-related cerebral networks in children, through the use of EEG and MEG, as well as a detailed portrait of relevant MEG and EEG data analysis methods used in that specific research context. To do so, we have summarized the results and systematically compared the methodological approach of 24 peer-reviewed EEG or MEG scientific studies that included healthy children and children with or at high risk of language disabilities, from birth up to 18 years of age. All included studies employed functional and effective connectivity measures, such as coherence, phase locking value, and Phase Slope Index, and did so using different experimental paradigms (e.g., at rest or during language-related tasks). This review will provide more insight into the use of EEG and MEG for the study of language networks in children, contribute to the current state of knowledge on the developmental path of functional connectivity in language networks during childhood and adolescence, and finally allow future studies to choose the most appropriate type of connectivity analysis.

Intrinsic connectome organization across temporal scales: New insights from cross-modal approaches

The discovery of a stable, whole-brain functional connectivity organization that is largely independent of external events has drastically extended our view of human brain function. However, this discovery has been primarily based on functional magnetic resonance imaging (fMRI). The role of this whole-brain organization in fast oscillation-based connectivity as measured, for example, by electroencephalography (EEG) and magnetoencephalography (MEG) is only beginning to emerge. Here, we review studies of intrinsic connectivity and its whole-brain organization in EEG, MEG, and intracranial electrophysiology with a particular focus on direct comparisons to connectome studies in fMRI. Synthesizing this literature, we conclude that irrespective of temporal scale over four orders of magnitude, intrinsic neurophysiological connectivity shows spatial similarity to the connectivity organization commonly observed in fMRI. A shared structural connectivity basis and cross-frequency coupling are possible mechanisms contributing to this similarity. Acknowledging that a stable whole-brain organization governs long-range coupling across all timescales of neural processing motivates researchers to take "baseline" intrinsic connectivity into account when investigating brain-behavior associations, and further encourages more widespread exploration of functional connectomics approaches beyond fMRI by using EEG and MEG modalities.

A Sparse EEG-Informed fMRI Model for Hybrid EEG-fMRI Neurofeedback Prediction

Measures of brain activity through functional magnetic resonance imaging (fMRI) or electroencephalography (EEG), two complementary modalities, are ground solutions in the context of neurofeedback (NF) mechanisms for brain rehabilitation protocols. While NF-EEG (in which real-time neurofeedback scores are computed from EEG signals) has been explored for a very long time, NF-fMRI (in which real-time neurofeedback scores are computed from fMRI signals) appeared more recently and provides more robust results and more specific brain training. Using fMRI and EEG simultaneously for bi-modal neurofeedback sessions (NF-EEG-fMRI, in which real-time neurofeedback scores are computed from fMRI and EEG) is very promising for the design of brain rehabilitation protocols. However, fMRI is cumbersome and more exhausting for patients. The original contribution of this paper concerns the prediction of bi-modal NF scores from EEG recordings only, using a training phase where EEG signals as well as the NF-EEG and NF-fMRI scores are available. We propose a sparse regression model able to exploit EEG only to predict NF-fMRI or NF-EEG-fMRI in motor imagery tasks. We compared different NF-predictors stemming from the proposed model. We showed that predicting NF-fMRI scores from EEG signals adds information to NF-EEG scores and significantly improves the correlation with bi-modal NF sessions compared to classical NF-EEG scores.

EPI distortion correction for concurrent human brain stimulation and imaging at 3T

BACKGROUND

Transcranial magnetic stimulation (TMS) can be paired with functional magnetic resonance imaging (fMRI) in concurrent TMS-fMRI experiments. These multimodal experiments enable causal probing of network architecture in the human brain which can complement alternative network mapping approaches. Critically, merely introducing the TMS coil into the scanner environment can sometimes produce substantial magnetic field inhomogeneities and spatial distortions which limit the utility of concurrent TMS-fMRI.

METHOD AND RESULTS

We assessed the efficacy of point spread function corrected echo planar imaging (PSF-EPI) in correcting for the field inhomogeneities associated with a TMS coil at 3 T. In phantom and brain scans, we quantitatively compared the coil-induced distortion artifacts measured in EPI scans with and without PSF correction. We found that the application of PSF corrections to the EPI data significantly improved signal-to-noise and reduced distortions. In phantom scans with the PSF-EPI sequence, we also characterized the temporal profile of dynamic artifacts associated with TMS delivery and found that image quality remained high as long as the TMS pulse preceded the RF excitation pulses by at least 50 ms. Lastly, we validated the PSF-EPI sequence in human brain scans involving TMS and motor behavior as well as resting state fMRI scans.

CONCLUSIONS

Our collective results demonstrate the potential benefits of PSF-EPI for concurrent TMS-fMRI when coil-related artifacts are a concern. The ability to collect high quality resting state fMRI data in the same session as the concurrent TMS-fMRI experiment offers a unique opportunity to interrogate network architecture in the human brain.

Simultaneous EEG-fMRI for Functional Neurological Assessment

The increasing incidence of neurodegenerative and psychiatric diseases requires increasingly sophisticated tools for their diagnosis and monitoring. Clinical assessment takes advantage of objective parameters extracted by electroencephalogram and magnetic resonance imaging (MRI) among others, to support clinical management of neurological diseases. The complementarity of these two tools can be now emphasized by the possibility of integrating the two technologies in a hybrid solution, allowing simultaneous acquisition of the two signals by the novel EEG-fMRI technology. This review will focus on simultaneous EEG-fMRI technology and related early studies, dealing about issues related to the acquisition and processing of simultaneous signals, and including critical discussion about clinical and technological perspectives.

Aberrant activity in conceptual networks underlies N400 deficits and unusual thoughts in schizophrenia

BACKGROUND

The N400 event-related potential (ERP) is triggered by meaningful stimuli that are incongruous, or unmatched, with their semantic context. Functional magnetic resonance imaging (fMRI) studies have identified brain regions activated by semantic incongruity, but their precise links to the N400 ERP are unclear. In schizophrenia (SZ), N400 amplitude reduction is thought to reflect overly broad associations in semantic networks, but the abnormalities in brain networks underlying deficient N400 remain unknown. We utilized joint independent component analysis (JICA) to link temporal patterns in ERPs to neuroanatomical patterns from fMRI and investigate relationships between N400 amplitude and neuroanatomical activation in SZ patients and healthy controls (HC).

METHODS

SZ patients (n = 24) and HC participants (n = 25) performed a picture-word matching task, in which words were either matched (APPLE→apple) by preceding pictures, or were unmatched by semantically related (in-category; IC, APPLE→lemon) or unrelated (out of category; OC, APPLE→cow) pictures, in separate ERP and fMRI sessions. A JICA "data fusion" analysis was conducted to identify the fMRI brain regions specifically associated with the ERP N400 component. SZ and HC loading weights were compared and correlations with clinical symptoms were assessed.

RESULTS

JICA identified an ERP-fMRI "fused" component that captured the N400, with loading weights that were reduced in SZ. The JICA map for the IC condition showed peaks of activation in the cingulate, precuneus, bilateral temporal poles and cerebellum, whereas the JICA map from the OC condition was linked primarily to visual cortical activation and the left temporal pole. Among SZ patients, fMRI activity from the IC condition was inversely correlated with unusual thought content.

CONCLUSIONS

The neural networks associated with the N400 ERP response to semantic violations depends on conceptual relatedness. These findings are consistent with a distributed network underlying neural responses to semantic incongruity including unimodal visual areas as well as integrative, transmodal areas. Unusual thoughts in SZ may reflect impaired processing in transmodal hub regions such as the precuneus, leading to overly broad semantic associations.

Simultaneous EEG-fMRI: Evaluating the Effect of the EEG Cap-Cabling Configuration on the Gradient Artifact

Electroencephalography (EEG) data recorded during simultaneous EEG-fMRI experiments are contaminated by large gradient artifacts (GA). The amplitude of the GA depends on the area of the wire loops formed by the EEG leads, as well as on the rate of switching of the magnetic field gradients, which are essential for MR imaging. Average artifact subtraction (AAS), the most commonly used method for GA correction, relies on the EEG amplifier having a large enough dynamic range to characterize the artifact voltages. Low-pass filtering (250 Hz cut-off) is generally used to attenuate the high-frequency voltage fluctuations of the GA, but even with this precaution channel saturation can occur, particularly during acquisition of high spatial resolution MRI data. Previous work has shown that the ribbon cable, used to connect the EEG cap and amplifier, makes a significant contribution to the GA, since the cable geometry produces large effective wire-loop areas. However, by appropriately connecting the wires of the ribbon cable to the EEG cap it should be possible to minimize the overall range and root mean square (RMS) amplitude of the GA by producing partial cancelation of the cap and cable contributions. Here by modifying the connections of the EEG cap to a 1 m ribbon cable we were able to reduce the range of the GA for a high-resolution coronal echo planar Imaging (EPI) acquisition by a factor of ∼ 1.6 and by a factor of ∼ 1.15 for a standard axial EPI acquisition. These changes could potentially be translated into a reduction in the required dynamic range, an increase in the EEG bandwidth or an increase in the achievable image resolution without saturation, all of which could be beneficially exploited in EEG-fMRI studies. The re-wiring could also prevent the system from saturating when small subject movements occur using the standard recording bandwidth.

Measurement of Neurovascular Coupling in Neonates

Neurovascular coupling refers to the mechanism that links the transient neural activity to the subsequent change in cerebral blood flow, which is regulated by both chemical signals and mechanical effects. Recent studies suggest that neurovascular coupling in neonates and preterm born infants is different compared to adults. The hemodynamic response after a stimulus is later and less pronounced and the stimulus might even result in a negative (hypoxic) signal. In addition, studies both in animals and neonates confirm the presence of a short hypoxic period after a stimulus in preterm infants. In clinical practice, different methodologies exist to study neurovascular coupling. The combination of functional magnetic resonance imaging or functional near-infrared spectroscopy (brain hemodynamics) with EEG (brain function) is most commonly used in neonates. Especially near-infrared spectroscopy is of interest, since it is a non-invasive method that can be integrated easily in clinical care and is able to provide results concerning longer periods of time. Therefore, near-infrared spectroscopy can be used to develop a continuous non-invasive measurement system, that could be used to study neonates in different clinical settings, or neonates with different pathologies. The main challenge for the development of a continuous marker for neurovascular coupling is how the coupling between the signals can be described. In practice, a wide range of signal interaction measures exist. Moreover, biomedical signals often operate on different time scales. In a more general setting, other variables also have to be taken into account, such as oxygen saturation, carbon dioxide and blood pressure in order to describe neurovascular coupling in a concise manner. Recently, new mathematical techniques were developed to give an answer to these questions. This review discusses these recent developments.

Investigating the variability of cardiac pulse artifacts across heartbeats in simultaneous EEG-fMRI recordings: A 7T study

Electroencephalography (EEG) recordings performed in magnetic resonance imaging (MRI) scanners are affected by complex artifacts caused by heart function, often termed pulse artifacts (PAs). PAs can strongly compromise EEG data quality, and remain an open problem for EEG-fMRI. This study investigated the properties and mechanisms of PA variability across heartbeats, which has remained largely unaddressed to date, and evaluated its impact on PA correction approaches. Simultaneous EEG-fMRI was performed at 7T on healthy participants at rest or under visual stimulation, with concurrent recordings of breathing and cardiac activity. PA variability was found to contribute to EEG variance with more than 500 μV² at 7T, which extrapolates to 92 μV² at 3T. Clustering analyses revealed that PA variability not only is linked to variations in head position/orientation, as previously hypothesized, but also, and more importantly, to the respiratory cycle and to heart rate fluctuations. The latter mechanisms are associated to short-timescale variability (even across consecutive heartbeats), and their importance varied across EEG channels. In light of this PA variability, three PA correction techniques were compared: average artifact subtraction (AAS), optimal basis sets (OBS), and an approach based on K-means clustering. All methods allowed the recovery of visual evoked potentials from the EEG data; nonetheless, OBS and K-means tended to outperform AAS, likely due to the inability of the latter in modeling short-timescale variability. Altogether, these results offer novel insights into the dynamics and underlying mechanisms of the pulse artifact, with important consequences for its correction, relevant to most EEG-fMRI applications.

Identification of epileptic brain states by dynamic functional connectivity analysis of simultaneous EEG-fMRI: a dictionary learning approach

Most fMRI studies of the brain's intrinsic functional connectivity (FC) have assumed that this is static; however, it is now clear that it changes over time. This is particularly relevant in epilepsy, which is characterized by a continuous interchange between epileptic and normal brain states associated with the occurrence of epileptic activity. Interestingly, recurrent states of dynamic FC (dFC) have been found in fMRI data using unsupervised learning techniques, assuming either their sparse or non-sparse combination. Here, we propose an l₁-norm regularized dictionary learning (l₁-DL) approach for dFC state estimation, which allows an intermediate and flexible degree of sparsity in time, and demonstrate its application in the identification of epilepsy-related dFC states using simultaneous EEG-fMRI data. With this l₁-DL approach, we aim to accommodate a potentially varying degree of sparsity upon the interchange between epileptic and non-epileptic dFC states. The simultaneous recording of the EEG is used to extract time courses representative of epileptic activity, which are incorporated into the fMRI dFC state analysis to inform the selection of epilepsy-related dFC states. We found that the proposed l₁-DL method performed best at identifying epilepsy-related dFC states, when compared with two alternative methods of extreme sparsity (k-means clustering, maximum; and principal component analysis, minimum), as well as an l₀-norm regularization framework (l₀-DL), with a fixed amount of temporal sparsity. We further showed that epilepsy-related dFC states provide novel insights into the dynamics of epileptic networks, which go beyond the information provided by more conventional EEG-correlated fMRI analysis, and which were concordant with the clinical profile of each patient. In addition to its application in epilepsy, our study provides a new dFC state identification method of potential relevance for studying brain functional connectivity dynamics in general.

Quantitative Analyses Help in Choosing Between Simultaneous vs. Separate EEG and fMRI

Simultaneous registration of scalp electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) is considered an attractive approach for studying brain function non-invasively. It combines the better spatial resolution of fMRI with the better temporal resolution of EEG, but comes at the cost of increased measurement artifact and the accompanying artifact preprocessing. This paper presents a study of the residual signal quality based on temporal signal to noise ratio (TSNR) for fMRI and fast Fourier transform (FFT) for EEG, after optimized conventional signal preprocessing. Measurements outside the magnetic resonance imaging scanner and inside the scanner prior to and during image acquisition were compared. For EEG, frequency and region dependent significant effects on FFT squared amplitudes were observed between separately vs. simultaneously recorded EEG and fMRI, with larger effects during image acquisition than without image acquisition inside the scanner bore. A graphical user interface was developed to aid in quality checking these measurements. For fMRI, separately recorded EEG-fMRI revealed relatively large areas with a significantly higher TSNR in right occipital and parietal regions and in the cingulum, compared to separately recorded EEG-fMRI. Simultaneously recorded EEG-fMRI showed significantly higher TSNR in inferior occipital cortex, diencephalon and brainstem, compared to separately recorded EEG-fMRI. Quantification of EEG and fMRI signals showed significant, but sometimes subtle, changes between separate compared to simultaneous EEG-fMRI measurements. To avoid interference with the experiment of interest in a simultaneous EEG-fMRI measurement, it seems warranted to perform a quantitative evaluation to ensure that there are no such uncorrectable effects present in regions or frequencies that are of special interest to the research question at hand.

Comparison of EEG microstates with resting state fMRI and FDG-PET measures in the default mode network via simultaneously recorded trimodal (PET/MR/EEG) data

Simultaneous trimodal positron emission tomography/magnetic resonance imaging/electroencephalography (PET/MRI/EEG) resting state (rs) brain data were acquired from 10 healthy male volunteers. The rs-functional MRI (fMRI) metrics, such as regional homogeneity (ReHo), degree centrality (DC) and fractional amplitude of low-frequency fluctuations (fALFFs), as well as 2-[18F]fluoro-2-desoxy-d-glucose (FDG)-PET standardised uptake value (SUV), were calculated and the measures were extracted from the default mode network (DMN) regions of the brain. Similarly, four microstates for each subject, showing the diverse functional states of the whole brain via topographical variations due to global field power (GFP), were estimated from artefact-corrected EEG signals. In this exploratory analysis, the GFP of microstates was nonparametrically compared to rs-fMRI metrics and FDG-PET SUV measured in the DMN of the brain. The rs-fMRI metrics (ReHO, fALFF) and FDG-PET SUV did not show any significant correlations with any of the microstates. The DC metric showed a significant positive correlation with microstate C (r_s  = 0.73, p = .01). FDG-PET SUVs indicate a trend for a negative correlation with microstates A, B and C. The positive correlation of microstate C with DC metrics suggests a functional relationship between cortical hubs in the frontal and occipital lobes. The results of this study suggest further exploration of this method in a larger sample and in patients with neuropsychiatric disorders. The aim of this exploratory pilot study is to lay the foundation for the development of such multimodal measures to be applied as biomarkers for diagnosis, disease staging, treatment response and monitoring of neuropsychiatric disorders.

Adaptive optimal basis set for BCG artifact removal in simultaneous EEG-fMRI

Electroencephalography (EEG) signals recorded during simultaneous functional magnetic resonance imaging (fMRI) are contaminated by strong artifacts. Among these, the ballistocardiographic (BCG) artifact is the most challenging, due to its complex spatio-temporal dynamics associated with ongoing cardiac activity. The presence of BCG residuals in EEG data may hide true, or generate spurious correlations between EEG and fMRI time-courses. Here, we propose an adaptive Optimal Basis Set (aOBS) method for BCG artifact removal. Our method is adaptive, as it can estimate the delay between cardiac activity and BCG occurrence on a beat-to-beat basis. The effective creation of an optimal basis set by principal component analysis (PCA) is therefore ensured by a more accurate alignment of BCG occurrences. Furthermore, aOBS can automatically estimate which components produced by PCA are likely to be BCG artifact-related and therefore need to be removed. The aOBS performance was evaluated on high-density EEG data acquired with simultaneous fMRI in healthy subjects during visual stimulation. As aOBS enables effective reduction of BCG residuals while preserving brain signals, we suggest it may find wide application in simultaneous EEG-fMRI studies.