Classifying Routine Clinical Electroencephalograms With Multivariate Iterative Filtering and Convolutional Neural Networks.

Electroencephalogram (EEG) is widely used in basic and clinical neuroscience to explore neural states in various populations, and classifying these EEG recordings is a fundamental challenge. While machine learning shows promising results in classifying long multivariate time series, optimal prediction models and feature extraction methods for EEG classification remain elusive. Our study addressed the problem of EEG classification under the framework of brain age prediction, applying a deep learning model on EEG time series. We hypothesized that decomposing EEG signals into oscillatory modes would yield more accurate age predictions than using raw or canonically frequency-filtered EEG. Specifically, we employed multivariate intrinsic mode functions (MIMFs), an empirical mode decomposition (EMD) variant based on multivariate iterative filtering (MIF), with a convolutional neural network (CNN) model. Testing a large dataset of routine clinical EEG scans (n = 6540) from patients aged 1 to 103 years, we found that an ad-hoc CNN model without fine-tuning could reasonably predict brain age from EEGs. Crucially, MIMF decomposition significantly improved performance compared to canonical brain rhythms (from delta to lower gamma oscillations). Our approach achieved a mean absolute error (MAE) of 13.76 ± 0.33 and a correlation coefficient of 0.64 ± 0.01 in brain age prediction over the entire lifespan. Our findings indicate that CNN models applied to EEGs, preserving their original temporal structure, remains a promising framework for EEG classification, wherein the adaptive signal decompositions such as the MIF can enhance CNN models' performance in this task.

Comparing neuronal oscillations during visual spatial attention orienting between normobaric and hypobaric hypoxia.

Normobaric hypoxia (NH) and hypobaric hypoxia (HH) are both used to train aircraft pilots to recognize symptoms of hypoxia. NH (low oxygen concentration) training is often preferred because it is more cost effective, simpler, and safer than HH. It is unclear, however, whether NH is neurophysiologically equivalent to HH (high altitude). Previous studies have shown that neural oscillations, particularly those in the alpha band (8-12 Hz), are impacted by hypoxia. Attention tasks have been shown to reliably modulate alpha oscillations, although the neurophysiological impacts of hypoxia during cognitive processing remains poorly understood. To address this we investigated induced and evoked power alongside physiological data while participants performed an attention task during control (normobaric normoxia or NN), NH (fraction of inspired oxygen = 12.8%, partial pressure of inspired oxygen = 87.2 mmHg), and HH (3962 m, partial pressure of inspired oxygen = 87.2 mmHg) conditions inside a hypobaric chamber. No significant differences between NH and HH were found in oxygen saturation, end tidal gases, breathing rate, middle cerebral artery velocity and blood pressure. Induced alpha power was significantly decreased in NH and HH when compared to NN. Participants in the HH condition showed significantly increased induced lower-beta power and evoked higher-beta power, compared with the NH and NN conditions, indicating that NH and HH differ in their impact on neurophysiological activity supporting cognition. NH and HH were found not to be neurophysiologically equivalent as electroencephalography was able to differentiate NH from HH.

Byte-Pair Encoding for classifying routine clinical electroencephalograms in adults over the lifespan.

Routine clinical EEG is a standard test used for the neurological evaluation of patients. A trained specialist interprets EEG recordings and classifies them into clinical categories. Given time demands and high inter-reader variability, there is an opportunity to facilitate the evaluation process by providing decision support tools that can classify EEG recordings automatically. Classifying clinical EEG is associated with several challenges: classification models are expected to be interpretable; EEGs vary in duration and EEGs are recorded by multiple technicians operating various devices. Our study aimed to test and validate a framework for EEG classification which satisfies these requirements by transforming EEG into unstructured text. We considered a highly heterogeneous and extensive sample of routine clinical EEGs (n = 5785), with a wide range of participants aged between 15 and 99 years. EEG scans were recorded at a public hospital, according to 10/20 electrode positioning with 20 electrodes. The proposed framework was based on symbolizing EEG signals and adapting a previously proposed method from natural language processing (NLP) to break symbols into words. Specifically, we symbolized the multichannel EEG time series and applied a byte-pair encoding (BPE) algorithm to extract a dictionary of the most frequent patterns (tokens) reflecting the variability of EEG waveforms. To demonstrate the performance of our framework, we used newly-reconstructed EEG features to predict patients' biological age with a Random Forest regression model. This age prediction model achieved a mean absolute error of 15.7 years. We also correlated tokens' occurrence frequencies with age. The highest correlations between the frequencies of tokens and age were observed at frontal and occipital EEG channels. Our findings demonstrated the feasibility of applying an NLP-based approach to classifying routine clinical EEG. Notably, the proposed algorithm could be instrumental in classifying clinical EEG with minimal preprocessing and identifying clinically-relevant short events, such as epileptic spikes.

Associations between spontaneous electroencephalogram oscillations and oxygen saturation across normobaric and hypobaric hypoxia.

High-altitude indoctrination (HAI) trains individuals to recognize symptoms of hypoxia by simulating high-altitude conditions using normobaric (NH) or hypobaric (HH) hypoxia. Previous studies suggest that despite equivalent inspired oxygen levels, physiological differences could exist between these conditions. In particular, differences in neurophysiological responses to these conditions are not clear. Our study aimed to investigate correlations between oxygen saturation (SpO2 ) and neural responses in NH and HH. We recorded 5-min of resting-state eyes-open electroencephalogram (EEG) and SpO2 during control, NH, and HH conditions from 13 participants. We applied a multivariate framework to characterize correlations between SpO2 and EEG measures (spectral power and multiscale entropy [MSE]), within each participant and at the group level. Participants were desaturating during the first 150 s of NH versus steadily desaturated in HH. We considered the entire time interval, first and second half intervals, separately. All the conditions were characterized by statistically significant participant-specific patterns of EEG-SpO2 correlations. However, at the group level, the desaturation period expressed a robust pattern of these correlations across frequencies and brain locations. Specifically, the first 150 s of NH during desaturation differed significantly from the other conditions with negative absolute alpha power-SpO2 correlations and positive MSE-SpO2 correlations. Once steadily desaturated, NH and HH had no significant differences in EEG-SpO2 correlations. Our findings indicate that the desaturating phase of hypoxia is a critical period in HAI courses, which would require developing strategies for mitigating the hypoxic stimulus in a real-world situation.

Children with autism spectrum disorder show atypical electroencephalographic response to processing contextual incongruencies.

Children with autism spectrum disorder (ASD) experience difficulties with social communication, making it challenging to interpret contextual information that aids in accurately interpreting language. To investigate how the brain processes the contextual information and how this is different in ASD, we compared event-related potentials (ERPs) in response to processing visual and auditory congruent and incongruent information. Two groups of children participated in the study: 37 typically developing children and 15 children with ASD (age range = 6 to 12). We applied a language task involving auditory sentences describing congruent or incongruent images. We investigated two ERP components associated with language processing: the N400 and P600. Our results showed how children with ASD present significant differences in their neural responses in comparison with the TD group, even when their reaction times and correct trials are not significantly different from the TD group.

Developmental Differences in Neuromagnetic Cortical Activation and Phase Synchrony Elicited by Scenes with Faces during Movie Watching.

The neural underpinnings of humans' ability to process faces and how it changes over typical development have been extensively studied using paradigms where face stimuli are oversimplified, isolated, and decontextualized. The prevalence of this approach, however, has resulted in limited knowledge of face processing in ecologically valid situations, in which faces are accompanied by contextual information at multiple time scales. In the present study, we use a naturalistic movie paradigm to investigate how neuromagnetic activation and phase synchronization elicited by faces from movie scenes in humans differ between children and adults. We used MEG data from 22 adults (6 females, 3 left handed; mean age, 27.7 ± 5.28 years) and 20 children (7 females, 1 left handed; mean age, 9.5 ± 1.52 years) collected during movie viewing. We investigated neuromagnetic time-locked activation and phase synchronization elicited by movie scenes containing faces in contrast to other movie scenes. Statistical differences between groups were tested using a multivariate data-driven approach. Our results revealed lower face-elicited activation and theta/alpha phase synchrony between 120 and 330 ms in children compared with adults. Reduced connectivity in children was observed between the primary visual areas as well as their connections with higher-order frontal and parietal cortical areas. This is the first study to map neuromagnetic developmental changes in face processing in a time-locked manner using a naturalistic movie paradigm. It supports and extends the existing evidence of core face-processing network maturation accompanied by the development of an extended system of higher-order cortical areas engaged in face processing.

A matter of perspective: Distinct brain mechanisms for evaluating positive and negative social feedback about oneself and another person.

Social rejection is a common experience in the life of young adults. Electroencephalographic (EEG) such as N1, P1 and P3 amplitude has been linked to experiencing social rejection; it remains unclear, whether these components are also influenced by the perspective, e.g., feedback directed to oneself or another person. We used EEG to investigate brain mechanisms associated with social feedback, directed either to oneself or another person. Female students (N = 57) engaged in a Chatroom Interact Task (CIT) during EEG. In this task participants received feedback as to whether themselves or someone else was accepted or rejected as a video chat partner. Mood was measured with the Positive Affect Negative Affect Schedule (PANAS). Participants showed more negative mood after rejection compared to acceptance. Spatiotemporal EEG cluster analysis revealed significant differences in P1, N1 and P3 ERP components associated with Acceptance vs. Rejection. The late positive potential (LPP) component was larger when processing self vs. other-related social feedback. Higher empathy, neuroticism, and lower age were associated with smaller LPP amplitude differences between Self and Other conditions. In this study we identified distinct brain dynamics associated with encoding social feedback and whether the feedback was targeted toward the self or to others.

Choice and Trade-offs: Parent Decision Making for Neurotechnologies for Pediatric Drug-Resistant Epilepsy.

This qualitative study investigated factors that guide caregiver decision making and ethical trade-offs for advanced neurotechnologies used to treat children with drug-resistant epilepsy. Caregivers with affected children were recruited to semi-structured focus groups or interviews at one of 4 major epilepsy centers in Eastern and Western Canada and the USA (n = 22). Discussions were transcribed and qualitative analytic methods applied to examine values and priorities (eg, risks, benefits, adherence, invasiveness, reversibility) of caregivers pertaining to novel technologies to treat drug-resistant epilepsy. Discussions revealed 3 major thematic branches for decision making: (1) features of the intervention-risks and benefits, with an emphasis on an aversion to perceived invasiveness; (2) decision drivers-trust in the clinical team, treatment costs; and (3) quality of available information about neurotechnological options. Overall, caregivers' definition of treatment success is more expansive than seizure freedom. The full involvement of their values and priorities must be considered in the decision-making process.

Alterations in coordinated EEG activity precede the development of seizures in comatose children.

OBJECTIVE: We aimed to test the hypothesis that computational features of the first several minutes of EEG recording can be used to estimate the risk for development of acute seizures in comatose critically-ill children. METHODS: In a prospective cohort of 118 comatose children, we computed features of the first five minutes of artifact-free EEG recording (spectral power, inter-regional synchronization and cross-frequency coupling) and tested if these features could help identify the 25 children who went on to develop acute symptomatic seizures during the subsequent 48 hours of cEEG monitoring. RESULTS: Children who developed acute seizures demonstrated higher average spectral power, particularly in the theta frequency range, and distinct patterns of inter-regional connectivity, characterized by greater connectivity at delta and theta frequencies, but weaker connectivity at beta and low gamma frequencies. Subgroup analyses among the 97 children with the same baseline EEG background pattern (generalized slowing) yielded qualitatively and quantitatively similar results. CONCLUSIONS: These computational features could be applied to baseline EEG recordings to identify critically-ill children at high risk for acute symptomatic seizures. SIGNIFICANCE: If confirmed in independent prospective cohorts, these features would merit incorporation into a decision support system in order to optimize diagnostic and therapeutic management of seizures among comatose children.

Atypical age-related changes in cortical thickness in autism spectrum disorder.

Recent longitudinal neuroimaging and neurophysiological studies have shown that tracking relative age-related changes in neural signals, rather than a static snapshot of a neural measure, could offer higher sensitivity for discriminating typically developing (TD) individuals from those with autism spectrum disorder (ASD). It is not clear, however, which aspects of age-related changes (trajectories) would be optimal for identifying atypical brain development in ASD. Using a large cross-sectional data set (Autism Brain Imaging Data Exchange [ABIDE] repository; releases I and II), we aimed to explore age-related changes in cortical thickness (CT) in TD and ASD populations (age range 6-30 years old). Cortical thickness was estimated from T1-weighted MRI images at three scales of spatial coarseness (three parcellations with different numbers of regions of interest). For each parcellation, three polynomial models of age-related changes in CT were tested. Specifically, to characterize alterations in CT trajectories, we compared the linear slope, curvature, and aberrancy of CT trajectories across experimental groups, which was estimated using linear, quadratic, and cubic polynomial models, respectively. Also, we explored associations between age-related changes with ASD symptomatology quantified as the Autism Diagnostic Observation Schedule (ADOS) scores. While no overall group differences in cortical thickness were observed across the entire age range, ASD and TD populations were different in terms of age-related changes, which were located primarily in frontal and tempo-parietal areas. These atypical age-related changes were also associated with ADOS scores in the ASD group and used to predict ASD from TD development. These results indicate that the curvature is the most reliable feature for localizing brain areas developmentally atypical in ASD with a more pronounced effect with symptomatology and is the most sensitive in predicting ASD development.

Atypical neuromagnetic resting activity associated with thalamic volume and cognitive outcome in very preterm children.

Children born very preterm, even in the absence of overt brain injury or major impairment, are at increased risk of cognitive difficulties. This risk is associated with developmental disruptions of the thalamocortical system during critical periods while in the neonatal intensive care unit. The thalamus is an important structure that not only relays sensory information but acts as a hub for integration of cortical activity which regulates cortical power across a range of frequencies. In this study, we investigate the association between atypical power at rest in children born very preterm at school age using magnetoencephalography (MEG), neurocognitive function and structural alterations related to the thalamus using MRI. Our results indicate that children born extremely preterm have higher power at slow frequencies (delta and theta) and lower power at faster frequencies (alpha and beta), compared to controls born full-term. A similar pattern of spectral power was found to be associated with poorer neurocognitive outcomes, as well as with normalized T1 intensity and the volume of the thalamus. Overall, this study provides evidence regarding relations between structural alterations related to very preterm birth, atypical oscillatory power at rest and neurocognitive difficulties at school-age children born very preterm.

Alterations in Local Connectivity and Their Developmental Trajectories in Autism Spectrum Disorder: Does Being Female Matter?

Autism spectrum disorder (ASD) is diagnosed more often in males with a ratio of 1:4 females/males. This bias is even stronger in neuroimaging studies. There is a growing evidence suggesting that local connectivity and its developmental trajectory is altered in ASD. Here, we aim to investigate how local connectivity and its age-related trajectories vary with ASD in both males and females. We used resting-state fMRI data from the ABIDE I and II repository: males (n = 102) and females (n = 92) with ASD, and typically developing males (n = 104) and females (n = 92) aged between 6 and 26. Local connectivity was quantified as regional homogeneity. We found increases in local connectivity in participants with ASD in the somatomotor and limbic networks and decreased local connectivity within the default mode network. These alterations were more pronounced in females with ASD. In addition, the association between local connectivity and ASD symptoms was more robust in females. Females with ASD had the most distinct developmental trajectories of local connectivity compared with other groups. Overall, our findings of more pronounced local connectivity alterations in females with ASD could indicate a greater etiological load for an ASD diagnosis in this group congruent with the female protective effect hypothesis.

Multiple constrained minimum variance beamformer (MCMV) performance in connectivity analyses.

Functional brain connectivity is increasingly being seen as critical for cognition, perception and motor control. Magnetoencephalography and electroencephalography are modalities that offer noninvasive mapping of electrophysiological interactions among brain regions, yet suffer from signal leakage and signal cancellation when estimating brain activity. This leads to biased connectivity values which complicate interpretation. In this study, we test the hypothesis that a Multiple Constrained Minimum Variance beamformer (MCMV) outperforms the more traditional Linearly Constrained Minimum Variance beamformer (LCMV) for estimation of electrophysiological connectivity. To this end, MCMV and LCMV performance is compared in task related analyses with both simulated data and human MEG recordings of visual steady state signals, and in resting state analyses with simulated data and human MEG data of 89 subjects. In task related scenarios connectivity was estimated using coherence and phase locking values, whereas envelope correlations were used for the resting state data. We also introduce a novel Augmented Pairwise MCMV (APW-MCMV) approach for signal leakage suppression in resting state analyses and assess its performance against LCMV and more conventional MCMV approaches. We demonstrate that with MCMV effects of signal mixing and coherent source cancellation are greatly reduced in both task related and resting state conditions, while in contrast to other approaches 0- and short time lag interactions are preserved. In addition, we demonstrate that in resting state analyses, APW-MCMV strongly reduces spurious connections while better controlling for false negatives compared to more conservative measures such as symmetrical orthogonalization.

Neuromagnetic activation and oscillatory dynamics of stimulus-locked processing during naturalistic viewing.

Naturalistic stimuli such as watching a movie while in the scanner provide an ecologically valid paradigm that has the potential of extracting valuable information on how the brain processes complex stimuli in realistic visual and auditory contexts. Naturalistic viewing is also easier to conduct with challenging participant groups including patients and children. Given the high temporal resolution of MEG, in the present study, we demonstrate how a short movie clip can be used to map distinguishable activation and connectivity dynamics underlying the processing of specific classes of visual stimuli such as face and hand manipulations, as well as contrasting activation dynamics for auditory words and non-words. MEG data were collected from 22 healthy volunteers (6 females, 3 left handed, mean age - 27.7 â€‹± â€‹5.28 years) during the presentation of naturalistic audiovisual stimuli. The MEG data were split into trials with the onset of the stimuli belonging to classes of interest (words, non-words, faces, hand manipulations). Based on the components of the averaged sensor ERFs time-locked to the visual and auditory stimulus onset, four and three time-windows, respectively, were defined to explore brain activation dynamics. Pseudo-Z, defined as the ratio of the source-projected time-locked power to the projected noise power for each vertex, was computed and used as a proxy of time-locked brain activation. Statistical testing using the mean-centered Partial Least Squares analysis indicated periods where a given visual or auditory stimuli had higher activation. Based on peak pseudo-Z differences between the visual conditions, time-frequency resolved analyses were performed to assess beta band desynchronization in motor-related areas, and inter-trial phase synchronization between face processing areas. Our results provide the first evidence that activation and connectivity dynamics in canonical brain regions associated with the processing of particular classes of visual and auditory stimuli can be reliably mapped using MEG during presentation of naturalistic stimuli. Given the strength of MEG for brain mapping in temporal and frequency domains, the use of naturalistic stimuli may open new techniques in analyzing brain dynamics during ecologically valid sensation and perception.

Sex differences in brain connectivity and male vulnerability in very preterm children.

Evidence indicates better cognitive and behavioral outcomes for females born very preterm (≤32 weeks gestation) compared to males, but the neurophysiology underlying this apparent resiliency of the female brain remains poorly understood. Here we test the hypothesis that very preterm males express more pronounced connectivity alterations as a reflection of higher male vulnerability. Resting state MEG recordings, neonatal and psychometric data were collected from 100 children at age 8 years: very preterm boys (n = 27), very preterm girls (n = 34), full-term boys (n = 15) and full-term girls (n = 24). Neuromagnetic source dynamics were reconstructed from 76 cortical brain regions. Functional connectivity was estimated using inter-regional phase-synchronization. We performed a series of multivariate analyses to test for differences across groups as well as to explore relationships between deviations in functional connectivity and psychometric scores and neonatal factors for very preterm children. Very preterm boys displayed significantly higher (p < .001) absolute deviation from average connectivity of same-sex full-term group, compared to very preterm girls versus full-term girls. In the connectivity comparison between very preterm and full-term groups separately for boys and girls, significant group differences (p < .05) were observed for boys, but not girls. Sex differences in connectivity (p < .01) were observed in very preterm children but not in full-term groups. Our findings indicate that very preterm boys have greater alterations in resting neurophysiological network communication than girls. Such uneven brain communication disruption in very preterm boys and girls suggests that stronger connectivity alterations might contribute to male vulnerability in long-term behavioral and cognitive outcome.

Electrophysiology of Inhibitory Control in the Context of Emotion Processing in Children With Autism Spectrum Disorder.

Autism Spectrum Disorder (ASD) is an increasingly common developmental disorder that affects 1 in 59 children. Despite this high prevalence of ASD, knowledge regarding the biological basis of its associated cognitive difficulties remains scant. In this study, we aimed to identify altered neurophysiological responses underlying inhibitory control and emotion processing difficulties in ASD, together with their associations with age and various domains of cognitive and social function. This was accomplished by assessing electroencephalographic recordings during an emotional go/nogo task alongside parent rating scales of behavior. Event related potential (ERP) N200 component amplitudes were reduced in children with ASD compared to typically developing (TD) children. No group differences were found, however, for task performance, P300 amplitude or latency, or N170 amplitude or latency, suggesting that individuals with ASD may only present conflict monitoring abnormalities, as reflected by the reduced N200 component, compared to TD individuals. Consistent with previous findings, increased age correlated with improved task performance scores and reduced N200 amplitude in the TD group, indicating that as these children develop, their neural systems become more efficient. These associations were not identified in the ASD group. Results also showed significant associations between increased N200 amplitudes and improved executive control abilities and decreased autism traits in TD children only. The newly discovered findings of decreased brain activation in children with ASD, alongside differences in correlations with age compared to TD children, provide a potential neurophysiological indicator of atypical development of inhibitory control mechanisms in these individuals.

Atypical resting state neuromagnetic connectivity and spectral power in very preterm children.

BACKGROUND: Children born very preterm often display selective cognitive difficulties at school age even in the absence of major brain injury. Alterations in neurophysiological activity underpinning such difficulties, as well as their relation to specific aspects of adverse neonatal experience, remain poorly understood. In the present study, we examined interregional connectivity and spectral power in very preterm children at school age, and their relationship with clinical neonatal variables and long-term outcomes (IQ, executive functions, externalizing/internalizing behavior, visual-motor integration). METHODS: We collected resting state magnetoencephalographic (MEG) and psychometric data from a cohort at the age of 8 years followed prospectively since birth, which included three groups: Extremely Low Gestational Age (ELGA, 24-28 weeks GA n = 24, age 7.7 ± 0.38, 10 girls), Very Low Gestational Age (VLGA, 29-32 weeks GA n = 37, age 7.7 ± 0.39, 24 girls), and full-term children (38-41 weeks GA n = 39, age 7.9 ± 1.02, 24 girls). Interregional phase synchrony and spectral power were tested for group differences, and associations with neonatal and outcome variables were examined using mean-centered and behavioral Partial Least Squares (PLS) analyses, respectively. RESULTS: We found greater connectivity in the theta band in the ELGA group compared to VLGA and full-term groups, primarily involving frontal connections. Spectral power analysis demonstrated overall lower power in the ELGA and VLGA compared to full-term group. PLS indicated strong associations between neurophysiological connectivity at school age, adverse neonatal experience and cognitive performance, and behavior. Resting spectral power was associated only with behavioral scores. CONCLUSIONS: Our findings indicate significant atypicalities of neuromagnetic brain activity and connectivity in very preterm children at school age, with alterations in connectivity mainly observed only in the ELGA group. We demonstrate a significant relationship between connectivity, adverse neonatal experience, and long-term outcome, indicating that the disruption of developing neurophysiological networks may mediate relationships between neonatal events and cognitive and behavioral difficulties at school age.

Extreme male developmental trajectories of homotopic brain connectivity in autism.

It has been proposed that autism spectrum disorder (ASD) may be characterized by an extreme male brain (EMB) pattern of brain development. Here, we performed the first investigation of how age-related changes in functional brain connectivity may be expressed differently in females and males with ASD. We analyzed resting-state functional magnetic resonance imaging data of 107 typically developing (TD) females, 114 TD males, 104 females, and 115 males with ASD (6-26 years) from the autism brain imaging data exchange repository. We explored how interhemispheric homotopic connectivity and its maturational curvatures change across groups. Differences between ASD and TD and between females and males with ASD were observed for the rate of changes in connectivity in the absence of overall differences in connectivity. The largest portion of variance in age-related changes in connectivity was described through similarities between TD males, ASD males, and ASD females, in contrast to TD females. We found that shape of developmental curvature is associated with symptomatology in both males and females with ASD. We demonstrated that females and males with ASD tended to follow the male pattern of developmental changes in interhemispheric connectivity, supporting the EMB theory of ASD.

Localizing Event-Related Potentials Using Multi-source Minimum Variance Beamformers: A Validation Study.

Adaptive and non-adaptive beamformers have become a prominent neuroimaging tool for localizing neural sources of electroencephalographic (EEG) and magnetoencephalographic (MEG) data. In this study, we investigated single-source and multi-source scalar beamformers with respect to their performances in localizing and reconstructing source activity for simulated and real EEG data. We compared a new multi-source search approach (multi-step iterative approach; MIA) to our previous multi-source search approach (single-step iterative approach; SIA) and a single-source search approach (single-step peak approach; SPA). In order to compare performances across these beamformer approaches, we manipulated various simulated source parameters, such as the amount of signal-to-noise ratio (0.1-0.9), inter-source correlations (0.3-0.9), number of simultaneously active sources (2-8), and source locations. Results showed that localization performance followed the order of MIA > SIA > SPA regardless of the number of sources, source correlations, and single-to-noise ratios. In addition, SIA and MIA were significantly better than SPA at localizing four or more sources. Moreover, MIA was better than SIA and SPA at identifying the true source locations when signal characteristics were at their poorest. Source waveform reconstructions were similar between MIA and SIA but were significantly better than that for SPA. A similar trend was also found when applying these beamformer approaches to a real EEG dataset. Based on our findings, we conclude that multi-source beamformers (MIA and SIA) are an improvement over single-source beamformers for localizing EEG. Importantly, our new search method, MIA, had better localization performance, localization precision, and source waveform reconstruction as compared to SIA or SPA. We therefore recommend its use for improved source localization and waveform reconstruction of event-related potentials.

Dominant Patterns of Information Flow in the Propagation of the Neuromagnetic Somatosensory Steady-State Response.

Methods of functional connectivity are applied ubiquitously in studies involving non-invasive whole-brain signals, but may be not optimal for exploring the propagation of the steady-state responses, which are strong oscillatory patterns of neurodynamics evoked by periodic stimulation. In our study, we explore a functional network underlying the somatosensory steady-state response using methods of effective connectivity. Human magnetoencephalographic (MEG) data were collected in 10 young healthy adults during 23-Hz vibro-tactile stimulation of the right hand index finger. The whole-brain dynamics of MEG source activity was reconstructed with a linearly-constrained minimum variance beamformer. We applied information-theoretic tools to quantify asymmetries in information flows between primary somatosensory area SI and the rest of the brain. Our analysis identified a pattern of coupling, leading from area SI to a source in the secondary somato-sensory area SII, thalamus, and motor cortex all contralateral to stimuli as well as to a source in the cerebellum ipsilateral to the stimuli. Our results support previously reported empirical evidence collected both in in vitro and in vivo, indicating critical areas of activation of the somatosensory system at the level of systems neuroscience.