The detection of absence seizures using cross-frequency coupling analysis with a deep learning network.

High frequency oscillations are important novel biomarkers of epileptogenic tissue. The interaction of oscillations across different time scales is revealed as cross-frequency coupling (CFC) representing a high-order structure in the functional organization of brain rhythms. New artificial intelligence methods such as deep learning neural networks can provide powerful tools for automated analysis of EEG. Here we present a Stacked Sparse Autoencoder (SSAE) trained to recognize absence seizure activity based on the cross-frequency patterns within scalp EEG. We used EEG records from the Temple University Hospital database. Absence seizures (n = 94) from 12 patients were taken into analysis along with segments of background activity. Half of the records were selected randomly for network training and the second half were used for testing. Power-to-power coupling was calculated between all frequencies 2-120 Hz pairwise using the EEGLAB toolbox. The resulting CFC matrices were used as training or testing inputs to the autoencoder. The trained network was able to recognize background and seizure segments (not used in training) with a sensitivity of 96.3%, specificity of 99.8% and overall accuracy of 98.5%. Our results provide evidence that the SSAE neural networks can be used for automated detection of absence seizures within scalp EEG.

Assessment of Cognitive Reserve using Near Infrared Spectroscopy.

Cognitive reserve (CR) is the ability to preserve cognitive functions in the presence of brain pathology. In the context of Alzheimer's disease (AD), patients with higher CR show better cognitive performance relative to brain damage therefore higher CR reduces the risk of dementia. There is a strong need to develop a neurophysiological biomarker of CR given the growing interest in understanding protective brain mechanisms in AD. FMRI studies indicate that frontoparietal network plays an important role in cognitive reserve. We calculated intraregional functional connectivity of lateral prefrontal cortex (FC LPFC) using functional near infrared spectroscopy (fNIRS) in the resting state of 13 healthy individuals who were also assessed for IQ and motoric skills (the Purdue Pegboard test, PPT). FC LPFC was found to positively correlate with IQ (a proxy measure of cognitive reserve) while showing a lack of or negative correlation with the PPT scores. The results demonstrate that the cost-effective, noninvasive and widely applicable fNIRS technology can be used to evaluate cognitive reserve in individuals at risk for and patients with AD with possible numerous applications in the context of healthy aging and other age-related cognitive disorders.

Motor engagement relates to accurate perception of phonemes and audiovisual words, but not auditory words.

A longstanding debate has surrounded the role of the motor system in speech perception, but progress in this area has been limited by tasks that only examine isolated syllables and conflate decision-making with perception. Using an adaptive task that temporally isolates perception from decision-making, we examined an EEG signature of motor activity (sensorimotor µ/beta suppression) during the perception of auditory phonemes, auditory words, audiovisual words, and environmental sounds while holding difficulty constant at two levels (Easy/Hard). Results revealed left-lateralized sensorimotor µ/beta suppression that was related to perception of speech but not environmental sounds. Audiovisual word and phoneme stimuli showed enhanced left sensorimotor µ/beta suppression for correct relative to incorrect trials, while auditory word stimuli showed enhanced suppression for incorrect trials. Our results demonstrate that motor involvement in perception is left-lateralized, is specific to speech stimuli, and it not simply the result of domain-general processes. These results provide evidence for an interactive network for speech perception in which dorsal stream motor areas are dynamically engaged during the perception of speech depending on the characteristics of the speech signal. Crucially, this motor engagement has different effects on the perceptual outcome depending on the lexicality and modality of the speech stimulus.

Temporal Derivative Distribution Repair (TDDR): A motion correction method for fNIRS.

Functional near-infrared spectroscopy (fNIRS) is an optical neuroimaging technique of growing interest as a tool for investigation of cortical activity. Due to the on-head placement of optodes, artifacts arising from head motion are relatively less severe than for functional magnetic resonance imaging (fMRI). However, it is still necessary to remove motion artifacts. We present a novel motion correction procedure based on robust regression, which effectively removes baseline shift and spike artifacts without the need for any user-supplied parameters. Our simulations show that this method yields better activation detection performance than 5 other current motion correction methods. In our empirical validation on a working memory task in a sample of children 7-15 years, our method produced stronger and more extensive activation than any of the other methods tested. The new motion correction method enhances the viability of fNIRS as a functional neuroimaging modality for use in populations not amenable to fMRI.

Sensitivity of fNIRS to cognitive state and load.

Functional near-infrared spectroscopy (fNIRS) is an emerging low-cost noninvasive neuroimaging technique that measures cortical bloodflow. While fNIRS has gained interest as a potential alternative to fMRI for use with clinical and pediatric populations, it remains unclear whether fNIRS has the necessary sensitivity to serve as a replacement for fMRI. The present study set out to examine whether fNIRS has the sensitivity to detect linear changes in activation and functional connectivity in response to cognitive load, and functional connectivity changes when transitioning from a task-free resting state to a task. Sixteen young adult subjects were scanned with a continuous-wave fNIRS system during a 10-min resting-state scan followed by a letter n-back task with three load conditions. Five optical probes were placed over frontal and parietal cortices, covering bilateral dorsolateral PFC (dlPFC), bilateral ventrolateral PFC (vlPFC), frontopolar cortex (FP), and bilateral parietal cortex. Activation was found to scale linearly with working memory load in bilateral prefrontal cortex. Functional connectivity increased with increasing n-back loads for fronto-parietal, interhemispheric dlPFC, and local connections. Functional connectivity differed between the resting state scan and the n-back scan, with fronto-parietal connectivity greater during the n-back, and interhemispheric vlPFC connectivity greater during rest. These results demonstrate that fNIRS is sensitive to both cognitive load and state, suggesting that fNIRS is well-suited to explore the full complement of neuroimaging research questions and will serve as a viable alternative to fMRI.

Does the resting state connectivity have hemispheric asymmetry? A near-infrared spectroscopy study.

Near-infrared spectroscopy (NIRS) is a novel technology for low-cost noninvasive brain imaging suitable for use in virtually all subject and patient populations. Numerous studies of brain functional connectivity using fMRI, and recently NIRS, suggest new tools for the assessment of cognitive functions during task performance and the resting state (RS). We analyzed functional connectivity and its possible hemispheric asymmetry measuring coherence of optical signals at low frequencies (0.01-0.1 Hz) in the prefrontal cortex in 13 right-handed (RH) and 2 left-handed (LH) healthy subjects at rest (4-8 min) using a continuous-wave NIRS instrument CW5 (TechEn, Milford, MA). Two optical probes were placed bilaterally over the inferior frontal gyrus (IFG) and the middle frontal gyrus (MFG) using anatomical landmarks of the 10-20 system. As a result, 28 optical channels (14 for each hemisphere) were recorded for changes in oxygenated (HbO) and de-oxygenated (HbR) hemoglobin. Global physiological signals (respiratory and cardiac) were removed using Principal and Independent Component Analyses. Inter-channel coherences for HbO and HbR signals were calculated using Morlet wavelets along with correlation coefficients. Connectivity matrices showed specific patterns of connectivity which was higher within each anatomical region (IFG and MFG) and between hemispheres (e.g., left IFG<->right IFG) than between IFG and MFG in the same hemisphere. Laterality indexes were calculated as t-values for the 'left>right' comparisons of intrinsic connectivity within each regional group of channels in each subject. Regardless of handedness, the group average laterality indexes were negative thus revealing significantly higher connectivity in the right hemisphere in the majority of RH subjects and in both LH subjects. The analysis of Granger causality between hemispheres has also shown a greater flow of information from the right to the left hemisphere which may point to an important role of the right hemisphere in the resting state. These data encourage further exploration of the NIRS connectivity and its application for the analysis of hemispheric relationships within the functional architecture of the brain.

High-Frequency Oscillations Recorded on the Scalp of Patients With Epilepsy Using Tripolar Concentric Ring Electrodes.

Epilepsy is the second most prevalent neurological disorder ([Formula: see text]% prevalence) affecting [Formula: see text] million people worldwide with up to 75% from developing countries. The conventional electroencephalogram is plagued with artifacts from movements, muscles, and other sources. Tripolar concentric ring electrodes automatically attenuate muscle artifacts and provide improved signal quality. We performed basic experiments in healthy humans to show that tripolar concentric ring electrodes can indeed record the physiological alpha waves while eyes are closed. We then conducted concurrent recordings with conventional disc electrodes and tripolar concentric ring electrodes from patients with epilepsy. We found that we could detect high frequency oscillations, a marker for early seizure development and epileptogenic zone, on the scalp surface that appeared to become more narrow-band just prior to seizures. High frequency oscillations preceding seizures were present in an average of 35.5% of tripolar concentric ring electrode data channels for all the patients with epilepsy whose seizures were recorded and absent in the corresponding conventional disc electrode data. An average of 78.2% of channels that contained high frequency oscillations were within the seizure onset or irritative zones determined independently by three epileptologists based on conventional disc electrode data and videos.

Gamma- and theta-band synchronization during semantic priming reflect local and long-range lexical-semantic networks.

Anterior and posterior brain areas are involved in the storage and retrieval of semantic representations, but it is not known how these areas dynamically interact during semantic processing. We hypothesized that long-range theta-band coherence would reflect coupling of these areas and examined the oscillatory dynamics of lexical-semantic processing using a semantic priming paradigm with a delayed letter-search task while recording subjects' EEG. Time-frequency analysis revealed facilitation of semantic processing for Related compared to Unrelated conditions, which resulted in a reduced N400 and reduced gamma power from 150 to 450ms. Moreover, we observed greater anterior-posterior theta coherence for Unrelated compared to Related conditions over the time windows 150-425ms and 600-900ms. We suggest that while gamma power reflects activation of local functional networks supporting semantic representations, theta coherence indicates dynamic coupling of anterior and posterior areas for retrieval and post-retrieval processing and possibly an interaction between semantic relatedness and working memory.

Word class and context affect alpha-band oscillatory dynamics in an older population.

Differences in the oscillatory EEG dynamics of reading open class (OC) and closed class (CC) words have previously been found (Bastiaansen et al., 2005) and are thought to reflect differences in lexical-semantic content between these word classes. In particular, the theta-band (4-7 Hz) seems to play a prominent role in lexical-semantic retrieval. We tested whether this theta effect is robust in an older population of subjects. Additionally, we examined how the context of a word can modulate the oscillatory dynamics underlying retrieval for the two different classes of words. Older participants (mean age 55) read words presented in either syntactically correct sentences or in a scrambled order ("scrambled sentence") while their EEG was recorded. We performed time-frequency analysis to examine how power varied based on the context or class of the word. We observed larger power decreases in the alpha (8-12 Hz) band between 200-700 ms for the OC compared to CC words, but this was true only for the scrambled sentence context. We did not observe differences in theta power between these conditions. Context exerted an effect on the alpha and low beta (13-18 Hz) bands between 0 and 700 ms. These results suggest that the previously observed word class effects on theta power changes in a younger participant sample do not seem to be a robust effect in this older population. Though this is an indirect comparison between studies, it may suggest the existence of aging effects on word retrieval dynamics for different populations. Additionally, the interaction between word class and context suggests that word retrieval mechanisms interact with sentence-level comprehension mechanisms in the alpha-band.

Feasibility of recording high frequency oscillations with tripolar concentric ring electrodes during pentylenetetrazole-induced seizures in rats.

As epilepsy remains a refractory condition in about 30% of patients with complex partial seizures, electrical stimulation of the brain has recently shown potential for additive seizure control therapy. Previously, we applied noninvasive transcranial focal stimulation via novel tripolar concentric ring electrodes (TCREs) on the scalp of rats after inducing seizures with pentylenetetrazole (PTZ). We developed a close-loop system to detect seizures and automatically trigger the stimulation and evaluated its effect on the electrographic activity recorded by TCREs in rats. In our previous work the detectors of seizure onset were based on seizure-induced changes in signal power in the frequency range up to 100 Hz, while in this preliminary study we assess the feasibility of recording high frequency oscillations (HFOs) in the range up to 300 Hz noninvasively with scalp TCREs during PTZ-induced seizures. Grand average power spectral density estimate and generalized likelihood ratio tests were used to compare power of electrographic activity at different stages of seizure development in a group of rats (n= 8). The results suggest that TCREs have the ability to record HFOs from the scalp as well as that scalp-recorded HFOs can potentially be used as features for seizure onset detection.

Abnormal interictal gamma activity may manifest a seizure onset zone in temporal lobe epilepsy.

Even though recent studies have suggested that seizures do not occur suddenly and that before a seizure there is a period with an increased probability of seizure occurrence, neurophysiological mechanisms of interictal and pre-seizure states are unknown. The ability of mathematical methods to provide much more sensitive tools for the detection of subtle changes in the electrical activity of the brain gives promise that electrophysiological markers of enhanced seizure susceptibility can be found even during interictal periods when EEG of epilepsy patients often looks 'normal'. Previously, we demonstrated in animals that hippocampal and neocortical gamma-band rhythms (30-100 Hz) intensify long before seizures caused by systemic infusion of kainic acid. Other studies in recent years have also drawn attention to the fast activity (>30 Hz) as a possible marker of epileptogenic tissue. The current study quantified gamma-band activity during interictal periods and seizures in intracranial EEG (iEEG) in 5 patients implanted with subdural grids/intracranial electrodes during their pre-surgical evaluation. In all our patients, we found distinctive (abnormal) bursts of gamma activity with a 3 to 100 fold increase in power at gamma frequencies with respect to selected by clinicians, quiescent, artifact-free, 7-20 min "normal" background (interictal) iEEG epochs 1 to 14 hours prior to seizures. Increases in gamma activity were largest in those channels which later displayed the most intensive electrographic seizure discharges. Moreover, location of gamma-band bursts correlated (with high specificity, 96.4% and sensitivity, 83.8%) with seizure onset zone (SOZ) determined by clinicians. Spatial localization of interictal gamma rhythms within SOZ suggests that the persistent presence of abnormally intensified gamma rhythms in the EEG may be an important tool for focus localization and possibly a determinant of epileptogenesis.

Transcutaneous focal electrical stimulation via concentric ring electrodes reduces synchrony induced by pentylenetetrazole in beta and gamma bands in rats.

Epilepsy is a neurological disorder that affects approximately one percent of the world population. Anti-epileptic drugs are ineffective in 25~30% of cases. Electrical stimulation to control seizures may be an additive therapy. We applied noninvasive transcutaneous focal electrical stimulation (TFES) via concentric ring electrodes on the scalp of rats after inducing seizures with pentylenetetrazole. We found a significant increase in synchrony within the beta-gamma bands during seizures and that TFES significantly reduced the synchrony of the beta-gamma activity and increased synchrony in the delta band.

Functional connectivity in the prefrontal cortex measured by near-infrared spectroscopy during ultrarapid object recognition.

Near-infrared spectroscopy (NIRS) is a developing technology for low-cost noninvasive functional brain imaging. With multichannel optical instruments, it becomes possible to measure not only local changes in hemoglobin concentrations but also temporal correlations of those changes in different brain regions which gives an optical analog of functional connectivity traditionally measured by fMRI. We recorded hemodynamic activity during the Go-NoGo task from 11 right-handed subjects with probes placed bilaterally over prefrontal areas. Subjects were detecting animals as targets in natural scenes pressing a mouse button. Data were low-pass filtered<1 Hz and cardiac∕respiration∕superficial layers artifacts were removed using Independent Component Analysis. Fisher's transformed correlations of poststimulus responses (30 s) were averaged over groups of channels unilaterally in each hemisphere (intrahemispheric connectivity) and the corresponding channels between hemispheres (interhemispheric connectivity). The hemodynamic response showed task-related activation (an increase∕decrease in oxygenated∕deoxygenated hemoglobin, respectively) greater in the right versus left hemisphere. Intra- and interhemispheric functional connectivity was also significantly stronger during the task compared to baseline. Functional connectivity between the inferior and the middle frontal regions was significantly stronger in the right hemisphere. Our results demonstrate that optical methods can be used to detect transient changes in functional connectivity during rapid cognitive processes.

Possible therapeutic effects of transcutaneous electrical stimulation via concentric ring electrodes.

Even with the latest advancements in antiepileptic drugs (AEDs) there are still many persons whose seizures are not controlled. There are also side effects reported associated with the AEDs. Electrical stimulation of the brain has shown promise toward controlling seizures. However, most brain stimulation techniques involve invasive procedures to implant electrodes and electronic stimulators. There are no conclusive descriptions of where to place the implanted electrodes to control seizures. Noninvasive electrical stimulation does not require the risks of implantation, and the electrodes can be moved easily as needed to determine where they may be the most effective in reducing seizure activity. Herein we review the progress of our group in the development of noninvasive electrical stimulation via concentric ring electrodes to control seizures in rats induced by penicillin G, pilocarpine, and pentylenetetrazole (PTZ).

"Seeing" electroencephalogram through the skull: imaging prefrontal cortex with fast optical signal.

Near-infrared spectroscopy is a novel imaging technique potentially sensitive to both brain hemodynamics (slow signal) and neuronal activity (fast optical signal, FOS). The big challenge of measuring FOS noninvasively lies in the presumably low signal-to-noise ratio. Thus, detectability of the FOS has been controversially discussed. We present reliable detection of FOS from 11 individuals concurrently with electroencephalogram (EEG) during a Go-NoGo task. Probes were placed bilaterally over prefrontal cortex. Independent component analysis (ICA) was used for artifact removal. Correlation coefficient in the best correlated FOS-EEG ICA pairs was highly significant (p < 10(-8)), and event-related optical signal (EROS) was found in all subjects. Several EROS components were similar to the event-related potential (ERP) components. The most robust "optical N200" at t = 225 ms coincided with the N200 ERP; both signals showed significant difference between targets and nontargets, and their timing correlated with subject's reaction time. Correlation between FOS and EEG even in single trials provides further evidence that at least some FOS components "reflect" electrical brain processes directly. The data provide evidence for the early involvement of prefrontal cortex in rapid object recognition. EROS is highly localized and can provide cost-effective imaging tools for cortical mapping of cognitive processes.

Event-related fast optical signal in a rapid object recognition task: improving detection by the independent component analysis.

Noninvasive recording of fast optical signals presumably reflecting neuronal activity is a challenging task because of a relatively low signal-to-noise ratio. To improve detection of those signals in rapid object recognition tasks, we used the independent component analysis (ICA) to reduce "global interference" (heartbeat and contribution of superficial layers). We recorded optical signals from the left prefrontal cortex in 10 right-handed participants with a continuous-wave instrument (DYNOT, NIRx, Brooklyn, NY). Visual stimuli were pictures of urban, landscape and seashore scenes with various vehicles as targets (target-to-non-target ratio 1:6) presented at ISI=166 ms or 250 ms. Subjects mentally counted targets. Data were filtered at 2-30 Hz and artifactual components were identified visually (for heartbeat) and using the ICA weight matrix (for superficial layers). Optical signals were restored from the ICA components with artifactual components removed and then averaged over target and non-target epochs. After ICA processing, the event-related response was detected in 70%-100% of subjects. The refined signal showed a significant decrease from baseline within 200-300 ms after targets and a slight increase after non-targets. The temporal profile of the optical signal corresponded well to the profile of a "differential ERP response", the difference between targets and non-targets which peaks at 200 ms in similar object detection tasks. These results demonstrate that the detection of fast optical responses with continuous-wave instruments can be improved through the ICA method capable to remove noise, global interference and the activity of superficial layers. Fast optical signals may provide further information on brain processing during higher-order cognitive tasks such as rapid categorization of objects.

Communication call-evoked gamma-band activity in the auditory cortex of awake bats is modified by complex acoustic features.

Mustached bats emit an acoustically rich variety of calls for social communication. In the posterior primary auditory cortex, activity of neural ensembles measured as local field potentials (LFPs) can uniquely encode each call type. Here we report that LFPs recorded in response to calls contain oscillatory activity in the gamma-band frequency range (>20 Hz). The power spectrum of these high-frequency oscillations shows either two peaks of energy (at 40 Hz and 100 Hz), or just one peak at 40 Hz. The relative power of gamma-band activity in the power spectrum of a call-evoked LFP correlates significantly with the 'harmonic complexity' of a call. Gamma-band activity is attenuated with reversal of frequency-modulated calls. Amplitude modulation, even when asymmetric across call reversals, has no significant effect on gamma-band activity. These results provide the first experimental evidence that complex features within different groups of species-specific calls modify the power spectrum of evoked gamma-band activity.

Local field potentials and spiking activity in the primary auditory cortex in response to social calls.

The mustached bat, Pteronotus parnellii, uses complex communication sounds ("calls") for social interactions. We recorded both event-related local field potentials (LFPs) and single/few-unit (SU) spike activity from the same electrode in the posterior region of the primary auditory cortex (AIp) during presentation of simple syllabic calls to awake bats. Temporal properties of the LFPs, which reflect activity within local neuronal clusters, and spike discharges from SUs were studied at 138 recording sites in six bats using seven variants each of 14 simple syllables presented at intensity levels of 40-90 dB SPL. There was no clear spatial selectivity to different call types within the AIp area. Rather, as shown previously, single units responded to multiple call types with similar values of the peak response rate in the peri-stimulus time histogram (PSTH). The LFPs and SUs, however, showed a rich temporal structure that was unique for each call type. Multidimensional scaling (MDS) of the averaged waveforms of call-evoked LFPs and PSTHs revealed that calls were better segregated in the two-dimensional space based on the LFP compared with the PSTH data. A representation within the "LFP-space" revealed that one of the dimensions correlated with the predominant and fundamental frequency of a call. The other dimension showed a high correlation with "harmonic complexity" ("fine" spectral structure of a call). We suggest that the temporal pattern of LFP and spiking activity reflects call-specific dynamics at any locus within the AIp area. This dynamic contributes to a distributed (population-based) representation of calls. Alternatively stated, the fundamental frequency and harmonic structure of calls, and not the recording location within the AIp, determines the temporal structure of the call-evoked LFP.

Neurodynamics for auditory stream segregation: tracking sounds in the mustached bat's natural environment.

During navigation and the search phase of foraging, mustached bats emit approximately 25 ms long echolocation pulses (at 10-40 Hz) that contain multiple harmonics of a constant frequency (CF) component followed by a short (3 ms) downward frequency modulation. In the context of auditory stream segregation, therefore, bats may either perceive a coherent pulse-echo sequence (PEPE...), or segregated pulse and echo streams (P-P-P... and E-E-E...). To identify the neural mechanisms for stream segregation in bats, we developed a simple yet realistic neural network model with seven layers and 420 nodes. Our model required recurrent and lateral inhibition to enable output nodes in the network to 'latch-on' to a single tone (corresponding to a CF component in either the pulse or echo), i.e., exhibit differential suppression by the alternating two tones presented at a high rate (> 10 Hz). To test the applicability of our model to echolocation, we obtained neurophysiological data from the primary auditory cortex of awake mustached bats. Event-related potentials reliably reproduced the latching behaviour observed at output nodes in the network. Pulse as well as nontarget (clutter) echo CFs facilitated this latching. Individual single unit responses were erratic, but when summed over several recording sites, they also exhibited reliable latching behaviour even at 40 Hz. On the basis of these findings, we propose that a neural correlate of auditory stream segregation is present within localized synaptic activity in the mustached bat's auditory cortex and this mechanism may enhance the perception of echolocation sounds in the natural environment.

Modeling complex tone perception: grouping harmonics with combination-sensitive neurons.

Perception of complex communication sounds is a major function of the auditory system. To create a coherent precept of these sounds the auditory system may instantaneously group or bind multiple harmonics within complex sounds. This perception strategy simplifies further processing of complex sounds and facilitates their meaningful integration with other sensory inputs. Based on experimental data and a realistic model, we propose that associative learning of combinations of harmonic frequencies and nonlinear facilitation of responses to those combinations, also referred to as "combination-sensitivity," are important for spectral grouping. For our model, we simulated combination sensitivity using Hebbian and associative types of synaptic plasticity in auditory neurons. We also provided a parallel tonotopic input that converges and diverges within the network. Neurons in higher-order layers of the network exhibited an emergent property of multifrequency tuning that is consistent with experimental findings. Furthermore, this network had the capacity to "recognize" the pitch or fundamental frequency of a harmonic tone complex even when the fundamental frequency itself was missing.