Generative Adversarial Network (GAN) for Simulating Electroencephalography.

Electroencephalographs record the electrical activity of your brain through the scalp. Electroencephalography is difficult to obtain due to its sensitivity and variability. Applications of electroencephalography such as for diagnosis, education, brain-computer interfaces require large samples of electroencephalography recording, however, it is often difficult to obtain the required datasets. Generative adversarial networks are robust deep learning framework which have proven themselves to be capable of synthesizing data. The robust nature of a generative adversarial network was used to generate multi-channel electroencephalography data in order to see if generative adversarial networks could reconstruct the spatio-temporal aspects of multi-channel electroencephalography signals. We were able to find that the synthetic electroencephalography data was able to replicate fine details of electroencephalography data and could potentially help us to generate large sample synthetic resting-state electroencephalography data for use in simulation testing of neuroimaging analyses. Generative adversarial networks (GANs) are robust deep-learning frameworks that can be trained to be convincing replicants of real data GANs were capable of generating "fake" EEG data that replicated fine details and topographies of "real" resting-state EEG data.

SimMEEG software for simulating event-related MEG and EEG data with underlying functional connectivity.

BACKGROUND: Simulating electromagnetic brain signals is becoming more and more prevalent because of the need to verify and validate electrophysiological analysis methods, especially with regards to measurements of functional connectivity and phase-amplitude coupling (PAC). NEW METHOD: A Matlab software package with a graphical user interface, called SimMEEG, was developed so that novice and expert users could input basic signal parameters (e.g., multiple frequencies, amplitudes, durations, phase interactions, and phase-amplitude couplings) in order to simulate signals that approximate event-related brain activity. A major objective for developing this software was to create software to simulate interacting brain signals that could be used to verify functional connectivity analyses, specifically phase-locking value (PLV), phase-lag index (PLI) algorithms, and PAC. SimMEEG simulates three sources with various parameters defined by users. Users can define, multiple frequencies, event-related changes in amplitude, synchronizations (i.e., PLV and PLI), and PAC. RESULTS: SimMEEG simulated signals that estimated the signal input parameters within specific target limits for PLV, PLI, and PAC. In addition, SimMEEG's synthetic signals closely approximated real human visual event-related responses. The Monte-Carlo module was used to simulate ten MEG and EEG datasets, conduct function connectivity analyses for comparison among several inverse solutions, and generate errors from ground truth. SimMEEG identified inverse solutions that had good and poor performances for functional connectivity analyses. COMPARISON WITH EXISTING METHODS: Comparison between other simulation methods and SimMEEG are discussed. CONCLUSION: SimMEEG software could be used to simulate real human EEG and MEG to aid in verification of analysis pipelines by comparing analyzed results to the simulated ground truth of the user-defined event-related brain signals. A core program within SimMEEG, called SimSignals, might also be used in other fields that would benefit from simulations of three interacting signals.

How can no change in an auditory stimulus generate an N2b-P3a?

The purpose of this study was to investigate the occurrence of an endogenously-evoked no-go N2b. Previous literature focused on the N2b being evoked by exogenous auditory stimuli. In this study, no-go stimuli were the absence of a gap in a 1000-ms noise burst (i.e., no-gap trials). ERPs were measured from 35 participants while performing a gap-detection task and passively listening to the same stimuli. Participants were asked to press a button when they heard a gap in the noise burst (go trials) and to withhold their button press when they did not perceive a gap in the noise burst (no-go trials). The current study's gap-detection task had predictable timing (gaps always occurred at 500 ms after noise burst onset) and high probability of gaps occurring (10:1); therefore, participants built up an expectancy that gaps would occur on most trials at 500 ms. For no-gap trials, this meant that a participant's expectancy was violated and thus a N2b-P3a response was generated. We found that all participants had N2b-P3a responses to no-gap trials. Overall, this study demonstrated that the no-go N2b-P3a response can be evoked by an endogenous signal in the form of the omission of an expected gap in noise.

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.

Event Related Potentials Reveal Early Phonological and Orthographic Processing of Single Letters in Letter-Detection and Letter-Rhyme Paradigms.

INTRODUCTION: When and where phonological processing occurs in the brain is still under some debate. Most paired-rhyme and phonological priming studies used word stimuli, which involve complex neural networks for word recognition and semantics. This study investigates early (<300 ms) and late (>300 ms) orthographic and phonological processing of letters. METHODS: Fifteen participants aged 20-35 engaged in three two-forced choice experiments, one letter-detection (LetterID) and two letter-rhyme (Paired-Rhyme and Letter-Rhyme) tasks. From the EEG recordings, event related potential (ERP) differences within and across task stimuli were found. We also calculated the global field power (GFP) for each participant. Accuracies and reaction times were also measured from their button presses for each task. RESULTS: Behavioral: Reaction times were 18 ms faster to letter than pseudoletter stimuli, and 27 ms faster to rhyme than nonrhyme stimuli. ERP/GFP: In the LetterID task, grand-mean evoked potentials (EPs) showed typical P1, N1, P2, and P3 waveform morphologies to letter and pseudoletter stimuli, with GFPs to pseudoletters being greater than letters from 160-600 ms. Across both rhyme tasks, there were greater negativities for nonrhyme than for rhyme stimuli at 145 ms and 426 ms. The P2 effect for rhyme stimuli was smaller than letter stimuli when compared across tasks. CONCLUSION: Differences in early processing of letters vs. pseudoletters between 130-190 ms suggest that letters are processed earlier and perhaps faster in the brain than pseudoletters. The P2 effect between letter and rhyme stimuli likely reflect sublexical phonological processing. Together, findings from our study fill in evidence for the temporal dynamics of orthographic and phonological processing of single letters.

Altered Network Oscillations and Functional Connectivity Dynamics in Children Born Very Preterm.

Structural brain connections develop atypically in very preterm children, and altered functional connectivity is also evident in fMRI studies. Such alterations in brain network connectivity are associated with cognitive difficulties in this population. Little is known, however, about electrophysiological interactions among specific brain networks in children born very preterm. In the present study, we recorded magnetoencephalography while very preterm children and full-term controls performed a visual short-term memory task. Regions expressing task-dependent activity changes were identified using beamformer analysis, and inter-regional phase synchrony was calculated. Very preterm children expressed altered regional recruitment in distributed networks of brain areas, across standard physiological frequency ranges including the theta, alpha, beta and gamma bands. Reduced oscillatory synchrony was observed among task-activated brain regions in very preterm children, particularly for connections involving areas critical for executive abilities, including middle frontal gyrus. These findings suggest that inability to recruit neurophysiological activity and interactions in distributed networks including frontal regions may contribute to difficulties in cognitive development in children born very preterm.

Region-Specific Slowing of Alpha Oscillations is Associated with Visual-Perceptual Abilities in Children Born Very Preterm.

Children born very preterm (≤32 weeks gestational age) without major intellectual or neurological impairments often express selective deficits in visual-perceptual abilities. The alterations in neurophysiological development underlying these problems, however, remain poorly understood. Recent research has indicated that spontaneous alpha oscillations are slowed in children born very preterm, and that atypical alpha-mediated functional network connectivity may underlie selective developmental difficulties in visual-perceptual ability in this group. The present study provides the first source-resolved analysis of slowing of spontaneous alpha oscillations in very preterm children, indicating alterations in a distributed set of brain regions concentrated in areas of posterior parietal and inferior temporal regions associated with visual perception, as well as prefrontal cortical regions and thalamus. We also uniquely demonstrate that slowing of alpha oscillations is associated with selective difficulties in visual-perceptual ability in very preterm children. These results indicate that region-specific slowing of alpha oscillations contribute to selective developmental difficulties prevalent in this population.

Paying attention to orthography: a visual evoked potential study.

In adult readers, letters, and words are rapidly identified within visual networks to allow for efficient reading abilities. Neuroimaging studies of orthography have mostly used words and letter strings that recruit many hierarchical levels in reading. Understanding how single letters are processed could provide further insight into orthographic processing. The present study investigated orthographic processing using single letters and pseudoletters when adults were encouraged to pay attention to or away from orthographic features. We measured evoked potentials (EPs) to single letters and pseudoletters from adults while they performed an orthographic-discrimination task (letters vs. pseudoletters), a color-discrimination task (red vs. blue), and a target-detection task (respond to #1 and #2). Larger and later peaking N1 responses (~170 ms) and larger P2 responses (~250 ms) occurred to pseudoletters as compared to letters. This reflected greater visual processing for pseudoletters. Dipole analyses localized this effect to bilateral fusiform and inferior temporal cortices. Moreover, this letter-pseudoletter difference was not modulated by task and thus indicates that directing attention to or away from orthographic features did not affect early visual processing of single letters or pseudoletters within extrastriate regions. Paying attention to orthography or color as compared to disregarding the stimuli (target-detection task) elicited selection negativities at about 175 ms, which were followed by a classical N2-P3 complex. This indicated that the tasks sufficiently drew participant's attention to and away from the stimuli. Together these findings revealed that visual processing of single letters and pseudoletters, in adults, appeared to be sensory-contingent and independent of paying attention to stimulus features (e.g., orthography or color).

Neonatal pain-related stress, functional cortical activity and visual-perceptual abilities in school-age children born at extremely low gestational age.

Children born very prematurely (< or =32 weeks) often exhibit visual-perceptual difficulties at school-age, even in the absence of major neurological impairment. The alterations in functional brain activity that give rise to such problems, as well as the relationship between adverse neonatal experience and neurodevelopment, remain poorly understood. Repeated procedural pain-related stress during neonatal intensive care has been proposed to contribute to altered neurocognitive development in these children. Due to critical periods in the development of thalamocortical systems, the immature brain of infants born at extremely low gestational age (ELGA; < or =28 weeks) may have heightened vulnerability to neonatal pain. In a cohort of school-age children followed since birth we assessed relations between functional brain activity measured using magnetoencephalogragy (MEG), visual-perceptual abilities and cumulative neonatal pain. We demonstrated alterations in the spectral structure of spontaneous cortical oscillatory activity in ELGA children at school-age. Cumulative neonatal pain-related stress was associated with changes in background cortical rhythmicity in these children, and these alterations in spontaneous brain oscillations were negatively correlated with visual-perceptual abilities at school-age, and were not driven by potentially confounding neonatal variables. These findings provide the first evidence linking neonatal pain-related stress, the development of functional brain activity, and school-age cognitive outcome in these vulnerable children.

Multi-core beamformers: derivation, limitations and improvements.

Minimum variance beamformers are popular tools used in EEG and MEG for analysis of brain activity. In recent years new multi-source beamformer methods were developed, including the Dual-Core Beamformer (DCBF) and its enhanced version (eDCBF). Both techniques should allow modeling of correlated brain activity under a wide range of conditions. However, the mathematical justification given is based on single-source results and computer simulations, which do not provide an insight into the assumptions involved and the limits of their applicability. Current work addresses this problem. Analytical expressions relating actual source parameters to those obtained with the DCBF and eDCBF are derived, and rigorous conclusions regarding the accuracy of the DCBF/eDCBF reconstructions are made. In particular, it is shown that DCBF accurately identifies source coordinates, but amplitudes and orientations are only correct for high SNRs and fully correlated sources. In contrast, eDCBF source localization is inaccurate, but if the source positions are found precisely, eDCBF allows perfect reconstruction for arbitrary SNRs. If the source positions are approximate, the reconstruction errors are generally larger for higher SNR values. The eDCBF results can be improved by using global unbiased localizer functions and an alternative way of estimating source orientations.

Functional communication within a perceptual network processing letters and pseudoletters.

Many studies have identified regions within human ventral visual stream to be important for object identification and categorization; however, knowledge of how perceptual information is communicated within the visual network is still limited. Current theories posit that if a high correspondence between incoming sensory information and internal representations exists, then the object is rapidly identified, and if there is not, then the object requires extra detailed processing. Event-related responses from the present magnetoencephalography study showed two main effects. The N1m peak latencies were approximately 15 milliseconds earlier to familiar letters than to unfamiliar pseudoletters, and the N2m was more negative to pseudoletters than to letters. Event-related beamforming analyses identified these effects to be within bilateral visual cortices with a right lateralization for the N2m effect. Furthermore, functional connectivity analyses revealed that gamma-band (50-80 Hz) oscillatory phase synchronizations among occipital regions were greater to letters than to pseudoletters (around 85 milliseconds). However, during a later time interval between 245 and 375 milliseconds, pseudoletters elicited greater gamma-band phase synchronizations among a more distributed occipital network than did letters. These findings indicate that familiar object processing begins by at least 85 milliseconds, which could represent an initial match to an internal template. In addition, unfamiliar object processing persisted longer than that for familiar objects, which could reflect greater attention to inexperienced objects to determine their identity and/or to consolidate a new template to aid in future identification.

Application of multi-source minimum variance beamformers for reconstruction of correlated neural activity.

Linearly constrained minimum variance beamformers are highly effective for analysis of weakly correlated brain activity, but their performance degrades when correlations become significant. Multiple constrained minimum variance (MCMV) beamformers are insensitive to source correlations but require a priori information about the source locations. Besides the question whether unbiased estimates of source positions and orientations can be obtained remained unanswered. In this work, we derive MCMV-based source localizers that can be applied to both induced and evoked brain activity. They may be regarded as a generalization of scalar minimum-variance beamformers for the case of multiple correlated sources. We show that for arbitrary noise covariance these beamformers provide simultaneous unbiased estimates of multiple source positions and orientations and remain bounded at singular points. We also propose an iterative search algorithm that makes it possible to find sources approximately without a priori assumptions about their locations and orientations. Simulations and analyses of real MEG data demonstrate that presented approach is superior to traditional single-source beamformers in situations where correlations between the sources are significant.

Magnetoencephalography reveals slowing of resting peak oscillatory frequency in children born very preterm.

Resting cortical activity is characterized by a distinct spectral peak in the alpha frequency range. Slowing of this oscillatory peak toward the upper theta-band has been associated with a variety of neurological and neuropsychiatric conditions and has been attributed to altered thalamocortical dynamics. Children born very preterm exhibit altered development of thalamocortical systems. To test the hypothesis that peak oscillatory frequency is slowed in children born very preterm, we recorded resting magnetoencephalography (MEG) from school age children born very preterm (≤ 32 wk gestation) without major intellectual or neurological impairment and age-matched full-term controls. Very preterm children exhibit a slowing of peak frequency toward the theta-band over bilateral frontal cortex, together with reduced alpha-band power over bilateral frontal and temporal cortex, suggesting that mildly dysrhythmic thalamocortical interactions may contribute to altered spontaneous cortical activity in children born very preterm.

Neuroimaging evidence for top-down maturation of selective auditory attention.

This study investigated maturational differences of selective auditory attention effects on transient evoked responses and 40-Hz auditory steady-state responses between children and adults. Magnetoencephalography (MEG) was recorded from children and adults performing a task where they attended to 40-Hz amplitude-modulated (AM) tones of 1,200 Hz while ignoring 40-Hz AM tones of 800 Hz. By using standard dipole-modeling procedures, the N1m of the transient evoked fields and the 40-Hz ASSRs were localized to secondary and primary auditory cortices, respectively. Source waveforms for the transient evoked fields and ASSRs were reconstructed at these locations and compared between attended and unattended tones. Source waveforms revealed attention enhances the sustained negativity of the transient evoked responses in both adults and children around 250 and 400 ms. ASSRs were also found to be enhanced within this time range but only for adults. The results provide evidence for a limited role of attention modification of the 40-Hz ASSRs in children around the age of 12 years old. Because ASSRs are generated in a lower auditory processing stage as compared to the transient auditory evoked responses, findings from the present study could indicate that the maturation of attention progresses in top-to-bottom manner. These findings fit with the notion that as a person gains sensory experience selective gating of relevant from irrelevant information likely occurs at earlier and earlier processing levels in order to become more automatic and efficient.

Altered long-range alpha-band synchronization during visual short-term memory retention in children born very preterm.

Children born very preterm, even when intelligence is broadly normal, often experience selective difficulties in executive function and visual-spatial processing. Development of structural cortical connectivity is known to be altered in this group, and functional magnetic resonance imaging (fMRI) evidence indicates that very preterm children recruit different patterns of functional connectivity between cortical regions during cognition. Synchronization of neural oscillations across brain areas has been proposed as a mechanism for dynamically assigning functional coupling to support perceptual and cognitive processing, but little is known about what role oscillatory synchronization may play in the altered neurocognitive development of very preterm children. To investigate this, we recorded magnetoencephalographic (MEG) activity while 7-8 year old children born very preterm and age-matched full-term controls performed a visual short-term memory task. Very preterm children exhibited reduced long-range synchronization in the alpha-band during visual short-term memory retention, indicating that cortical alpha rhythms may play a critical role in altered patterns functional connectivity expressed by this population during cognitive and perceptual processing. Long-range alpha-band synchronization was also correlated with task performance and visual-perceptual ability within the very preterm group, indicating that altered alpha oscillatory mechanisms mediating transient functional integration between cortical regions may be relevant to selective problems in neurocognitive development in this vulnerable population at school age.

Altered Long-Range Phase Synchronization and Cortical Activation in Children Born Very Preterm.

Children born very preterm, even with broadly normal IQ, commonly show selective difficulties in visuospatial processing and executive functioning. Very little, however, is known what alterations in cortical processing underlie these deficits. We recorded MEG while eight children born very preterm (≤32 weeks gestational age) and eight full-term controls performed a visual short-term memory task at mean age 7.5 years (range 6.4 - 8.4). Previously, we demonstrated increased long-range alpha and beta band phase synchronization between MEG sensors during STM retention in a group of 17 full-term children age 6-10 years. Here we present preliminary evidence that long-range phase synchronization in very preterm children, relative to controls, is reduced in the alpha-band but increased in the theta-band. In addition, we investigated cortical activation during STM retention employing synthetic aperture magnetometry (SAM) beamformer to localize changes in gamma-band power. Preliminary results indicate sequential activation of occipital, parietal and frontal cortex in control children, as well as reduced activation in very preterm children relative to controls. These preliminary results suggest that children born very preterm exhibit altered inter-regional functional connectivity and cortical activation during cognitive processing.

Long-range synchronization and local desynchronization of alpha oscillations during visual short-term memory retention in children.

Local alpha-band synchronization has been associated with both cortical idling and active inhibition. Recent evidence, however, suggests that long-range alpha synchronization increases functional coupling between cortical regions. We demonstrate increased long-range alpha and beta band phase synchronization during short-term memory retention in children 6-10 years of age. Furthermore, whereas alpha-band synchronization between posterior cortex and other regions is increased during retention, local alpha-band synchronization over posterior cortex is reduced. This constitutes a functional dissociation for alpha synchronization across local and long-range cortical scales. We interpret long-range synchronization as reflecting functional integration within a network of frontal and visual cortical regions. Local desynchronization of alpha rhythms over posterior cortex, conversely, likely arises because of increased engagement of visual cortex during retention.

External landmark and head-shape-based functional data normalization.

Functional data, as obtained from magnetoencephalography (MEG) techniques, is frequently transformed to a template brain space for pooling data across the population. This transformation is often performed via an intermediate magnetic resonance image (MRI) of the subject's brain that is first registered to the template brain MRI. However, in many instances, it is difficult, expensive or undesirable to acquire MRIs, and since reconstructed functional data is lower resolution, the full information in MRIs is not required for registration. We present here two alternative options for computing the transformation of functional data to a common template space. These alternatives compute the registration based on external landmarks placed on the head, and the external shape of the head, features that are considerably simpler and inexpensive to acquire than MRIs. We present quantification of the accuracy of using these alternative features for registration, and show that they give accuracy for functional data registration that is sufficient given the functional data resolution, and is comparable to existing methods that are commonly used.

Cortical oscillatory power changes during auditory oddball task revealed by spatially filtered magnetoencephalography.

OBJECTIVE: To investigate the neural sources and associated changes in oscillatory activity involved in auditory attention and memory updating processing using spatially filtered magnetoencephalography. METHODS: We recorded magnetic responses during an auditory oddball task in 12 normal subjects. Synthetic aperture magnetometry (SAM)-permutation analysis was used to visualize the multiple brain regions associated with event-related magnetic fields (ERFs), and event-related oscillations during target detection processing. RESULTS: SAM-permutation results showed the topographical distribution of N1m over the bilateral primary auditory cortex. Post-stimulus delta (1.5-4 Hz) activity sources, likely related to the P300 slow-waveform, were distributed over the right frontocentral and parietal regions. Source locations of theta (4-8 Hz) and alpha (8-13 Hz) event-related synchronization (ERS) were identified over the dorsolateral and medial prefrontal cortex. We visualized bilateral central-Rolandic suppresions for mu (8-15 Hz), beta (15-30 Hz), and low-gamma (30-60 Hz) activities, more dominant in the hemisphere contralateral to the moving hand (button-pressing in response to target stimuli). CONCLUSIONS: Prefrontal theta and alpha ERS, and frontocentral-parietal delta ERS are functionally engaged in auditory attention and memory updating process. SIGNIFICANCE: Spatially filtered MEG is valuable for detection and source localization of task-related changes in the ongoing oscillatory activity during oddball tasks.

A complementary analytic approach to examining medial temporal lobe sources using magnetoencephalography.

Neuropsychological and neuroimaging findings reveal that the hippocampus is important for recognition memory. However, it is unclear when and whether the hippocampus contributes differentially to recognition of previously studied items (old) versus novel items (new), or contributes to a general processing requirement that is necessary for recognition of both types of information. To address this issue, we examined the temporal dynamics and spectral frequency underlying hippocampal activity during recognition of old/new complex scenes using magnetoencephalography (MEG). In order to provide converging evidence to existing literature in support of the potential of MEG to localize the hippocampus, we reconstructed brain source activity using the beamformer method and analyzed three types of processing-related signal changes by applying three different analysis methods: (1) Synthetic aperture magnetometry (SAM) revealed event related and non-event-related spectral power changes; (2) Inter-trial coherence (ITC) revealed time-locked changes in neural synchrony; and (3) Event-related SAM (ER-SAM) revealed averaged event-related responses over time. Hippocampal activity was evident for both old and new information within the theta frequency band and during the first 250 ms following stimulus onset. The early onset of hippocampal responses suggests that general comparison processes related to recognition of new/old information may occur obligatorily.