Intrinsic coupling between gamma oscillations, neuronal discharges, and slow cortical oscillations during human slow-wave sleep

Enhanced phase-amplitude coupling of human electrocorticography selectively in the posterior cortical region during rapid eye movement sleep

The spatiotemporal dynamics of interaction between slow (delta or infraslow) waves and fast (gamma) activities during wakefulness and sleep are yet to be elucidated in human electrocorticography (ECoG). We evaluated phase-amplitude coupling (PAC), which reflects neuronal coding in information processing, using ECoG in 11 patients with intractable focal epilepsy. PAC was observed between slow waves of 0.5-0.6 Hz and gamma activities, not only during light sleep and slow-wave sleep (SWS) but even during wakefulness and rapid eye movement (REM) sleep. While PAC was high over a large region during SWS, it was stronger in the posterior cortical region around the temporoparietal junction than in the frontal cortical region during REM sleep. PAC tended to be higher in the posterior cortical region than in the frontal cortical region even during wakefulness. Our findings suggest that the posterior cortical region has a functional role in REM sleep and may contribute to the maintenance of the dreaming experience.

Gamma-band modulation in the human amygdala during reaching movements

OBJECTIVE

Motor brain-computer interface (BCI) represents a new frontier in neurological surgery that could provide significant benefits for patients living with motor deficits. Both the primary motor cortex and posterior parietal cortex have successfully been used as a neural source for human motor BCI, leading to interest in exploring other brain areas involved in motor control. The amygdala is one area that has been shown to have functional connectivity to the motor system; however, its role in movement execution is not well studied. Gamma oscillations (30-200 Hz) are known to be prokinetic in the human cortex, but their role is poorly understood in subcortical structures. Here, the authors use direct electrophysiological recordings and the classic "center-out" direct-reach experiment to study amygdaloid gamma-band modulation in 8 patients with medically refractory epilepsy.

METHODS

The study population consisted of 8 epilepsy patients (2 men; age range 21-62 years) who underwent implantation of micro-macro depth electrodes for seizure localization and EEG monitoring. Data from the macro contacts sampled at 2000 Hz were used for analysis. The classic center-out direct-reach experiment was used, which consists of an intertrial interval phase, a fixation phase, and a response phase. The authors assessed the statistical significance of neural modulation by inspecting for nonoverlapping areas in the 95% confidence intervals of spectral power for the response and fixation phases.

RESULTS

In 5 of the 8 patients, power spectral analysis showed a statistically significant increase in power within regions of the gamma band during the response phase compared with the fixation phase. In these 5 patients, the 95% bootstrapped confidence intervals of trial-averaged power in contiguous frequencies of the gamma band during the response phase were above, and did not overlap with, the confidence intervals of trial-averaged power during the fixation phase.

CONCLUSIONS

To the authors' knowledge, this is the first time that direct neural recordings have been used to show gamma-band modulation in the human amygdala during the execution of voluntary movement. This work indicates that gamma-band modulation in the amygdala could be a contributing source of neural signals for use in a motor BCI system.

Neural oscillations in the infralimbic cortex after electrical stimulation of the amygdala. Relevance to acute stress processing

The stress system coordinates the adaptive reactions of the organism to stressors. Therefore, dysfunctions in this circuit may correlate to anxiety-related disorders, including depression. Comprehending the dynamics of this network may lead to a better understanding of the mechanisms that underlie these diseases. The central nucleus of the amygdala (CeA) activates the hypothalamic-pituitary-adrenal axis and brainstem nodes by triggering endocrine, autonomic and behavioral stress responses. The medial prefrontal cortex plays a significant role in regulating reactions to stressors, and is specifically important for limiting fear responses. Brain oscillations reflect neural systems activity. Synchronous neuronal assemblies facilitate communication and synaptic plasticity, mechanisms that cooperatively support the temporal representation and long-term consolidation of information. The purpose of this article was to delve into the interactions between these structures in stress contexts by evaluating changes in oscillatory activity. We particularly analyzed the local field potential in the infralimbic region of the medial prefrontal cortex (IL) in urethane-anesthetized rats after the electrical activation of the central nucleus of the amygdala by mimicking firing rates induced by acute stress. Electrical CeA activation induced a delayed, but significant, change in the IL, with prominent slow waves accompanied by an increase in the theta and gamma activities, and spindles. The phase-amplitude coupling of both slow waves and theta oscillations significantly increased with faster oscillations, including theta-gamma coupling and the nesting of spindles, theta and gamma oscillations in the slow wave cycle. These results are further discussed in neural processing terms of the stress response and memory formation.

High-Frequency Oscillations in the Normal Human Brain

OBJECTIVE

High-frequency oscillations (HFOs) are a promising biomarker for the epileptogenic zone. It has not been possible, however, to differentiate physiological from pathological HFOs, and baseline rates of HFO occurrence vary substantially across brain regions. This project establishes region-specific normative values for physiological HFOs and high-frequency activity (HFA).

METHODS

Intracerebral stereo-encephalographic recordings with channels displaying normal physiological activity from nonlesional tissue were selected from 2 tertiary epilepsy centers. Twenty-minute sections from N2/N3 sleep were selected for automatic detection of ripples (80-250Hz), fast ripples (>250Hz), and HFA defined as long-lasting activity > 80Hz. Normative values are provided for 17 brain regions.

RESULTS

A total of 1,171 bipolar channels with normal physiological activity from 71 patients were analyzed. The highest rates of ripples were recorded in the occipital cortex, medial and basal temporal region, transverse temporal gyrus and planum temporale, pre- and postcentral gyri, and medial parietal lobe. The mean rate of fast ripples was very low (0.038/min). Only 5% of channels had a rate > 0.2/min HFA was observed in the medial occipital lobe, pre- and postcentral gyri, transverse temporal gyri and planum temporale, and lateral occipital lobe.

INTERPRETATION

This multicenter atlas is the first to provide region-specific normative values for physiological HFO rates and HFA in common stereotactic space; rates above these can now be considered pathological. Physiological ripples are frequent in eloquent cortex. In contrast, physiological fast ripples are very rare, making fast ripples a good candidate for defining the epileptogenic zone. Ann Neurol 2018;84:374-385.

Sleep spindle and K-complex detection using tunable Q-factor wavelet transform and morphological component analysis

A novel framework for joint detection of sleep spindles and K-complex events, two hallmarks of sleep stage S2, is proposed. Sleep electroencephalography (EEG) signals are split into oscillatory (spindles) and transient (K-complex) components. This decomposition is conveniently achieved by applying morphological component analysis (MCA) to a sparse representation of EEG segments obtained by the recently introduced discrete tunable Q-factor wavelet transform (TQWT). Tuning the Q-factor provides a convenient and elegant tool to naturally decompose the signal into an oscillatory and a transient component. The actual detection step relies on thresholding (i) the transient component to reveal K-complexes and (ii) the time-frequency representation of the oscillatory component to identify sleep spindles. Optimal thresholds are derived from ROC-like curves (sensitivity vs. FDR) on training sets and the performance of the method is assessed on test data sets. We assessed the performance of our method using full-night sleep EEG data we collected from 14 participants. In comparison to visual scoring (Expert 1), the proposed method detected spindles with a sensitivity of 83.18% and false discovery rate (FDR) of 39%, while K-complexes were detected with a sensitivity of 81.57% and an FDR of 29.54%. Similar performances were obtained when using a second expert as benchmark. In addition, when the TQWT and MCA steps were excluded from the pipeline the detection sensitivities dropped down to 70% for spindles and to 76.97% for K-complexes, while the FDR rose up to 43.62 and 49.09%, respectively. Finally, we also evaluated the performance of the proposed method on a set of publicly available sleep EEG recordings. Overall, the results we obtained suggest that the TQWT-MCA method may be a valuable alternative to existing spindle and K-complex detection methods. Paths for improvements and further validations with large-scale standard open-access benchmarking data sets are discussed.

Learning machines and sleeping brains: Automatic sleep stage classification using decision-tree multi-class support vector machines

BACKGROUND

Sleep staging is a critical step in a range of electrophysiological signal processing pipelines used in clinical routine as well as in sleep research. Although the results currently achievable with automatic sleep staging methods are promising, there is need for improvement, especially given the time-consuming and tedious nature of visual sleep scoring.

NEW METHOD

Here we propose a sleep staging framework that consists of a multi-class support vector machine (SVM) classification based on a decision tree approach. The performance of the method was evaluated using polysomnographic data from 15 subjects (electroencephalogram (EEG), electrooculogram (EOG) and electromyogram (EMG) recordings). The decision tree, or dendrogram, was obtained using a hierarchical clustering technique and a wide range of time and frequency-domain features were extracted. Feature selection was carried out using forward sequential selection and classification was evaluated using k-fold cross-validation.

RESULTS

The dendrogram-based SVM (DSVM) achieved mean specificity, sensitivity and overall accuracy of 0.92, 0.74 and 0.88 respectively, compared to expert visual scoring. Restricting DSVM classification to data where both experts' scoring was consistent (76.73% of the data) led to a mean specificity, sensitivity and overall accuracy of 0.94, 0.82 and 0.92 respectively.

COMPARISON WITH EXISTING METHODS

The DSVM framework outperforms classification with more standard multi-class "one-against-all" SVM and linear-discriminant analysis.

CONCLUSION

The promising results of the proposed methodology suggest that it may be a valuable alternative to existing automatic methods and that it could accelerate visual scoring by providing a robust starting hypnogram that can be further fine-tuned by expert inspection.

On noninvasive source imaging of the human K-complex

OBJECTIVE

To assess whether existing noninvasive source localization techniques can provide valid solutions for large extended cortical sources we tested the capability of various methods of EEG source imaging (ESI) and magnetic source imaging (MSI) to localize the large superficial cortical generator of the human K-complex.

METHODS

We recently determined the intracranial distribution of the K-complex in a study of 6 patients with epilepsy (Clin. Neurophysiol. 121 (2010) 1176). Here we use the simultaneously acquired scalp EEG data to evaluate the validity and reliability of different ESI techniques. MEG recordings were acquired in 3 of the 6 patients, and K-complexes were recorded with high density EEG and MEG in an additional subject without epilepsy. ESI forward models included finite element method and boundary element method (BEM) volume conductors; for MSI, single sphere and BEM models were assessed. Inverse models included equivalent current dipole mapping and distributed current source modeling algorithms.

RESULTS

ESI and MSI provided physiologically invalid source solutions in all subjects, incorrectly localizing K-complex generators to deep midline structures. ESI provided consistent localization results across subjects for individual and averaged K-complexes, indicating solutions were not influenced by random noise or choice of model parameters. MEG K-complexes were lower in amplitude relative to baseline than EEG K-complexes, with less consistent localization results even after signal averaging, likely due to MEG-specific signal cancellation and sensitivity to source orientation. Distributed source modeling did not resolve the known problem of excessively deep fitting of single dipole locations for extended cortical sources.

CONCLUSIONS

Various noninvasive ESI and MSI techniques tested did not provide localization results for individual or averaged K-complexes that were physiologically meaningful or concordant with source locations indicated by intracranial recordings. Distributed source algorithms, though theoretically more appropriate for localizing extended cortical sources, showed the same propensity as dipole mapping to provide deep midline solutions for an extended superficial cortical source. Further studies are needed to determine appropriate modeling approaches for these large electrographic events.

SIGNIFICANCE

Existing noninvasive source localization techniques may not provide valid solutions for large extended cortical sources such as the human K-complex.

Human gamma oscillations during slow wave sleep

Neocortical local field potentials have shown that gamma oscillations occur spontaneously during slow-wave sleep (SWS). At the macroscopic EEG level in the human brain, no evidences were reported so far. In this study, by using simultaneous scalp and intracranial EEG recordings in 20 epileptic subjects, we examined gamma oscillations in cerebral cortex during SWS. We report that gamma oscillations in low (30-50 Hz) and high (60-120 Hz) frequency bands recurrently emerged in all investigated regions and their amplitudes coincided with specific phases of the cortical slow wave. In most of the cases, multiple oscillatory bursts in different frequency bands from 30 to 120 Hz were correlated with positive peaks of scalp slow waves ("IN-phase" pattern), confirming previous animal findings. In addition, we report another gamma pattern that appears preferentially during the negative phase of the slow wave ("ANTI-phase" pattern). This new pattern presented dominant peaks in the high gamma range and was preferentially expressed in the temporal cortex. Finally, we found that the spatial coherence between cortical sites exhibiting gamma activities was local and fell off quickly when computed between distant sites. Overall, these results provide the first human evidences that gamma oscillations can be observed in macroscopic EEG recordings during sleep. They support the concept that these high-frequency activities might be associated with phasic increases of neural activity during slow oscillations. Such patterned activity in the sleeping brain could play a role in off-line processing of cortical networks.

Large-scale heterogeneous representation of sound attributes in rat primary auditory cortex: from unit activity to population dynamics

Recent evidence indicates the existence of pyramidal cells (PCs) and interneurons with nontrivial tuning characteristics for sound attributes in the primary auditory cortex (A1) of mammals. These neurons are functionally distributed into layers and sparsely organized at a small scale. However, their topological locations at a large scale in A1 have not yet been investigated. Furthermore, these neurons are usually classified from fine maps of attribute-dependent spiking activity, and not much attention is paid to population postsynaptic potentials related to their activity. We used extracellular recordings obtained from multiple sites in A1 of adult rats to determine neuronal codifiers for sound attributes defined by coarse representations of the population dose-response curves. We demonstrated that these codifiers, majorly involving PCs, are heterogeneously distributed along A1. Spiking activity in these neurons during stimulation was correlated to β (12-25 Hz) and low γ (25-70 Hz) postsynaptic oscillations in the infragranular layer, whereas in the supragranular layer, better correlations were found with high γ (70-170 Hz) oscillations. The time-frequency analysis of the postsynaptic potentials showed a transient broadband power increase in all layers after the stimulus onset that was followed by a sustained high γ oscillation in the supragranular layer, fluctuations in the laminar content of the low-frequency oscillations, and a global attenuation in the low-frequency powers after the stimulus offset that happened together with a long-lasting strengthening of the β oscillations. We concluded that, for rats, sounds are codified in A1 by segregated networks of specialized PCs whose postsynaptic activity impinges on the emergence of sparse/dense spiking patterns.