EEG Montage Research Articles

Sensitivity of detecting interictal epileptiform activity using rapid reduced montage EEG

ObjectiveRapid EEG devices (REDs) have demonstrated substantial benefit regarding reduced time to performance of study and diagnosis in cases where urgent EEG is needed to evaluate patients for potentially revealing nonconvulsive status epilepticus and seizures. However, urgent EEG is also important in identifying cases regarding the need for initiation of antiseizure medication as well as triaging the use of continuous EEG monitoring. Some forms of REDs have a reduced montage (RRME) with electrode derivations that are one-half of standard recordings. This could impact spatial resolution and therefore potentially limit recovery of epileptiform abnormalities. MethodsIn this study we evaluated the use of the Ceribell® rapid response EEG system and compared it to conventional video EEG (CvEEG). After applying inclusion and exclusion criteria, a total of 20 subjects were included in our analysis. ResultsRRME was highly sensitive in detecting abundant and periodic discharges (p = 0.013) as well as discharges with a broad spatial distribution on CvEEG (p = 0.039). Sensitivity for detecting less prevalent discharges or those with more restricted spatial distribution was lower. SignificanceGiven the possibility of less frequent and more restricted epileptiform discharges eluding detection on RRME, we propose a protocol for the approach of using RRME and when to consider CvEEG when RRME is negative for epileptiform activity and highlight that urgent CvEEG may still be warranted following RRME.

Density Spectral Array EEG for Sleep Staging in Pediatric Patients.

Sleep is an essential physiologic process, which is frequently disrupted in children with illness and/or injury. Accurate identification and quantification of sleep may provide insights to improve long-term clinical outcomes. Traditionally, however, the identification of sleep stages has relied on the resource-intensive and time-consuming gold standard polysomnogram. We sought to use limited EEG data, converted into density spectrum array EEG, to accurately identify sleep stages in a clinical pediatric population. We reviewed 87 clinically indicated pediatric polysomnographic studies with concurrent full montage EEG, between March 2017 and June 2020, of which 11 had normal polysomnogram and EEG interpretations. We then converted the EEG data of those normal studies into density spectral array EEG trends and had five blinded raters classify sleep stage (wakefulness, nonrapid eye movement [NREM] 1, NREM 2, NREM 3, and rapid eye movement) in 5-minute epochs. We compared the classified sleep stages from density spectral array EEG to the gold standard polysomnogram. Inter-rater reliability was highest (κ = 0.745, P < 0.0001) when classifying state into wakefulness, NREM sleep, and rapid eye movement sleep. Agreement between group classification and polysomnogram was highest (κ = 0.873, [0.819, 0.926], P < 0.0001) when state was classified into wakefulness and sleep and was lowest (κ = 0.674 [0.645, 0.703], P < 0.0001) when classified into wakefulness, NREM 1, NREM 2, NREM 3, and rapid eye movement. The most common error that raters made was overscoring of NREM 1. Density spectral array EEG can be used to identify sleep stages in clinical pediatric patients without relying on traditional polysomnography.

Characterization of Cochlear Implant Artifact and Removal Based on Multi-Channel Wiener Filter in Unilateral Child Patients.

Cochlear implants (CI) allow deaf patients to improve language perception and improving their emotional valence assessment. Electroencephalographic (EEG) measures were employed so far to improve CI programming reliability and to evaluate listening effort in auditory tasks, which are particularly useful in conditions when subjective evaluations are scarcely appliable or reliable. Unfortunately, the presence of CI on the scalp introduces an electrical artifact coupled to EEG signals that masks physiological features recorded by electrodes close to the site of implant. Currently, methods for CI artifact removal have been developed for very specific EEG montages or protocols, while others require many scalp electrodes. In this study, we propose a method based on the Multi-channel Wiener filter (MWF) to overcome those shortcomings. Nine children with unilateral CI and nine age-matched normal hearing children (control) participated in the study. EEG data were acquired on a relatively low number of electrodes (n = 16) during resting condition and during an auditory task. The obtained results obtained allowed to characterize CI artifact on the affected electrode and to significantly reduce, if not remove it through MWF filtering. Moreover, the results indicate, by comparing the two sample populations, that the EEG data loss is minimal in CI users after filtering, and that data maintain EEG physiological characteristics.

Prognosis of comatose patients with reduced EEG montage by combining quantitative EEG features in various domains.

As the frontoparietal network underlies recovery from coma, a limited frontoparietal montage was used, and the prognostic values of EEG features for comatose patients were assessed. Collected with a limited frontoparietal EEG montage, continuous EEG recordings of 81 comatose patients in ICU were used retrospectively. By the 60-day Glasgow outcome scale (GOS), the patients were dichotomized into favorable and unfavorable outcome groups. Temporal-, frequency-, and spatial-domain features were automatically extracted for comparison. Partial correlation analysis was applied to eliminate redundant factors, and multiple correspondence analysis was used to explore discrimination between groups. Prognostic characteristics were calculated to assess the performance of EEG feature-based predictors established by logistic regression. Analyses were performed on all-patients group, strokes subgroup, and traumatic brain injury (TBI) subgroup. By analysis of all patients, raised burst suppression ratio (BSR), suppressed root mean square (RMS), raised power ratio of β to α rhythm (β/α), and suppressed phase-lag index between F3 and P4 (PLI [F3, P4]) were associated with unfavorable outcome, and yielded AUC of 0.790, 0.811, 0.722, and 0.844, respectively. For the strokes subgroup, the significant variables were BSR, RMS, θ/total, θ/δ, and PLI (F3, P4), while for the TBI subgroup, only PLI (F3, P4) was significant. BSR combined with PLI (F3, P4) gave the best predictor by cross-validation analysis in the all-patients group (AUC = 0.889, 95% CI: 0.819-0.960). Features extracted from limited frontoparietal montage EEG served as valuable coma prognostic tools, where PLI (F3, P4) was always significant. Combining PLI (F3, P4) with features in other domains may achieve better performance. A limited-montage EEG coupled with an automated algorithm is valuable for coma prognosis.

Evaluating EEG complexity metrics as biomarkers for depression.

Nonlinear EEG analysis offers the potential for both increased diagnostic accuracy and deeper mechanistic understanding of psychopathology. EEG complexity measures have previously been shown to positively correlate with clinical depression. In this study, resting state EEG recordings were taken across multiple sessions and days with both eyes open and eyes closed conditions from a total of 306 subjects, 62 of which were in a current depressive episode, and 81 of which had a history of diagnosed depression but were not currently depressed. Three different EEG montages (mastoids, average, and Laplacian) were also computed. Higuchi fractal dimension (HFD) and sample entropy (SampEn) were calculated for each unique condition. The complexity metrics showed high internal consistency within session and high stability across days. Higher complexity was found in open-eye recordings compared to closed eyes. The predicted correlation between complexity and depression was not found. However, an unexpected sex effect was observed, in which males and females exhibited different topographic patterns of complexity.

Supporting the Detection of Early Alzheimer's Disease with a Four-Channel EEG Analysis.

Alzheimer's disease (AD) is the most prevalent form of dementia. Although there is no current cure, medical treatment can help to control its progression. Hence, early-stage diagnosis is crucial to maximize the living standards of the patients. Biochemical markers and medical imaging in combination with neuropsychological tests represent the most extended diagnosis procedure. However, these techniques require specialized personnel and long processing time. Furthermore, the access to some of these techniques is often limited in crowded healthcare systems and rural areas. In this context, electroencephalography (EEG), a non-invasive technique to obtain endogenous brain information, has been proposed for the diagnosis of early-stage AD. Despite the valuable information provided by clinical EEG and high density montages, these approaches are impractical in conditions such as those described above. Consequently, in this study, we evaluated the feasibly of using a reduced EEG montage with only four channels to detect early-stage AD. For this purpose, we involved eight clinically diagnosed AD patients and eight healthy controls. The results we obtained reveal similar accuracies ([Formula: see text]-value[Formula: see text]0.66) for the reduced montage (0.86) and a 16-channel montage (0.87). This suggests that a four-channel wearable EEG system could be an effective tool for supporting early-stage AD detection.

Rapid response EEG with needle electrodes in an intensive care unit with limited resources

ObjectiveContinuous EEG (cEEG) is the gold standard for detecting seizures and rhythmic and periodic patterns (RPPs) in critically ill patients but is often not available in health systems with limited resources. The current study aims to determine the feasibility and utility of low-cost, practical, limited montage, sub-dermal needle electrode EEG in a setting where otherwise no EEG would be available. MethodsThe study included all adult patients admitted to the intensive care unit of a single center over a 24-month period. Members of the existing ICU care team, mostly nurses, were trained to place 8 sub-dermal needle EEG electrodes to achieve rapid, limited montage-EEG recording. Clinical outcomes were recorded, including any reported major complications; and the EEG findings documented, including background characterization, RPPs, and seizures. ResultsOne hundred twenty-three patients, mean age 68 years, underwent an average of 15.6 min of EEG recording. There were no complications of electrode placement. Overall, 13.0% had seizures (8.1% qualifying as status epilepticus [SE]), 18.7 % had generalized periodic discharges (GPDs), 4.9% had lateralized periodic discharges (LPDs), and 11.4 % sporadic epileptiform discharges (sEDs). Greater mortality was observed in patients with worse background EEGs, seizures, LPDs, or sEDs. ConclusionsRapid, limited montage EEG could be achieved safely and inexpensively in a broad population of critically ill patients following minimal training of existing care teams. SignificanceFor resource poor centers or centers outside of major metropolitan areas who otherwise have no access to EEG, this may prove a useful method for screening for non-convulsive seizures and status epilepticus.

Corticospinal excitability is highest at the early rising phase of sensorimotor µ-rhythm

Alpha oscillations are thought to reflect alternating cortical states of excitation and inhibition. Studies of perceptual thresholds and evoked potentials have shown the scalp EEG negative phase of the oscillation to correspond to a short-lasting low-threshold and high-excitability state of underlying visual, somatosensory, and primary motor cortex. The negative peak of the oscillation is assumed to correspond to the state of highest excitability based on biophysical considerations and considerable effort has been made to improve the extraction of a predictive signal by individually optimizing EEG montages. Here, we investigate whether it is the negative peak of sensorimotor µ-rhythm that corresponds to the highest corticospinal excitability, and whether this is consistent between individuals.In 52 adult participants, a standard 5-channel surface Laplacian EEG montage was used to extract sensorimotor µ-rhythm during transcranial magnetic stimulation (TMS) of primary motor cortex. Post-hoc trials were sorted from 800 TMS-evoked motor potentials (MEPs) according to the pre-stimulus EEG (estimated instantaneous phase) and MEP amplitude (as an index of corticospinal excitability). Different preprocessing transformations designed to improve the accuracy by which µ-alpha phase predicts excitability were also tested.By fitting a sinusoid to the MEP amplitudes, sorted according to pre-stimulus EEG-phase, we found that excitability was highest during the early rising phase, at a significant delay with respect to the negative peak by on average 45° or 10 ms. The individual phase of highest excitability was consistent across study participants and unaffected by two different EEG-cleaning methods that utilize 64 channels to improve signal quality by compensating for individual noise level and channel covariance. Personalized transformations of the montage did not yield better prediction of excitability from µ-alpha phase.The relationship between instantaneous phase of a brain oscillation and fluctuating cortical excitability appears to be more complex than previously hypothesized. In TMS of motor cortex, a standard surface Laplacian 5-channel EEG montage is effective in extracting a predictive signal and the phase corresponding to the highest excitability appears to be consistent between individuals. This is an encouraging result with respect to the clinical potential of therapeutic personalized brain interventions in the motor system. However, it remains to be investigated, whether similar results can be obtained for other brain areas and brain oscillations targeted with EEG and TMS.

Artificial Intelligence-based Detection of Epileptic Discharges from Pediatric Scalp Electroencephalograms: A Pilot Study.

We developed an artificial intelligence (AI) technique to identify epileptic discharges (spikes) in pediatric scalp electroencephalograms (EEGs). We built a convolutional neural network (CNN) model to automatically classify steep potential images into spikes and background activity. For the CNN model' training and validation, we examined 100 children with spikes in EEGs and another 100 without spikes. A different group of 20 children with spikes and 20 without spikes were the actual test subjects. All subjects were ≥ 3 to < 18 years old. The accuracy, sensitivity, and specificity of the analysis were >0.97 when referential and combination EEG montages were used, and < 0.97 with a bipolar montage. The correct classification of background activity in individual patients was significantly better with a referential montage than with a bipolar montage (p=0.0107). Receiver operating characteristic curves yielded an area under the curve > 0.99, indicating high performance of the classification method. EEG patterns that interfered with correct classification included vertex sharp transients, sleep spindles, alpha rhythm, and low-amplitude ill-formed spikes in a run. Our results demonstrate that AI is a promising tool for automatically interpreting pediatric EEGs. Some avenues for improving the technique were also indicated by our findings.

Analysis And Evaluation Of Consumers’ Cognitive Responses To Visual And Gustatory Stimuli For Neuromarketing Application

Neuromarketing merges viewpoints of marketing, neuroscience, economics, choice hypothesis that are required to analyze the psychology of consumers’ preference to product development. The traditional methods involve product ratings, conducting questionnaire surveys that stumble upon verbal declarations provided by the vendees. Consumer Neuroscience describes the emotional, cognitive aspects that form the base of human decision making. Our study aims to utilize the neuroscientific information that distinguishes contrasts between healthy subjects’ EEG signals for examining the brain activity during visual and gustatory stimuli of different flavours of a beverage brand. The EEG montage assigned according to brain-region-specific localization draws out the subjects’ true elicited subconscious response regardless of whether the subject attempts to control his/her affective state. The results showed the activation of theta and delta bands of EEG signals during the given stimuli. These elicited signal variations can be used to identify the best favoured item for successful product dispatch and reduction in loss. Another major application is directed towards the customization of liquid food intake of locked-in, comatose, vegetative state patients by observing their brain response to the various fluid intake and determining the best response among them. This aids physicians to put the patients on a path to recovery.

Quantitative detection of seizures with minimal-density EEG montage using phase synchrony and cross-channel coherence amplitude in critical care.

Seizures frequently occur in paediatric emergency and critical care, with up to 74% being sub-clinical seizures making detection difficult. Delays in seizure detection and treatment worsen the neurological outcome of critically-ill patients. Gold-standard seizure detections using multi-channels electroencephalograms (EEG) require trained clinical physiologists to apply scalp electrodes and highly specialised neurologists to interpret and identify seizures. In this study, we extracted phase synchrony and cross-channel coherence amplitude across 4 and 8 pre-selected scalp EEG signals. Binary classification is used to determine whether the signal segment is seizure or non-seizure, and the predictions were compared against the gold-standard seizure onset markings. The application of the algorithm on a cohort of forty routinely collected EEGs from paediatric patients showed an average accuracy of 77.2 % and 76.5% using 4 and 8 channels, respectively. Clinical Relevance- This work demonstrates the feasibility of seizure detection with pre-defined 4 and 8 EEG electrodes with an average accuracy of 77%. This means for the first time seizure detection is possible using an EEG montage that can be applied readily at the bedside independent of expert input.

µ-rhythm phase from somatosensory but not motor cortex correlates with corticospinal excitability in EEG-triggered TMS

BackgroundSensorimotor µ-rhythm phase is correlated with corticospinal excitability. Transcranial magnetic stimulation (TMS) of motor cortex results in larger motor evoked potentials (MEPs) during the negative peak of the EEG oscillation as extracted with a surface Laplacian. However, the anatomical source of the relevant oscillation is not clear and demonstration of the relationship is sensitive to the choice of EEG montage. Objective/hypothesisHere, we compared two EEG montages preferentially sensitive to oscillations originating from the crown of precentral gyrus (dorsal premotor cortex) vs. postcentral gyrus (secondary somatosensory cortex). We hypothesized that the EEG signal from precentral gyrus would correlate more strongly with MEP amplitude, given that the corticospinal neurons are located in the anterior wall of the sulcus and the corticospinal tract has input from premotor cortex. New methodReal-time EEG-triggered TMS of motor cortex was applied in 6 different conditions in randomly interleaved order, 3 phase conditions (positive peak, negative peak, random phase of the ongoing µ-oscillation), and each phase condition for 2 different EEG montages corresponding to oscillations preferentially originating in precentral gyrus (premotor cortex) vs. postcentral gyrus (somatosensory cortex), extracted using FCC3h vs. C3 centered EEG montages. ResultsThe negative vs. positive peak of sensorimotor µ-rhythm as extracted from the C3 montage (postcentral gyrus, somatosensory cortex) correlated with states of high vs. low corticospinal excitability (p < 0.001), replicating previous findings. However, no significant correlation was found for sensorimotor µ-rhythm as extracted from the neighboring FCC3 montage (precentral gyrus, premotor cortex). This implies that EEG-signals from the somatosensory cortex are better predictors of corticospinal excitability than EEG-signals from the motor areas. ConclusionsThe extraction of a brain oscillation whose phase corresponds to corticospinal excitability is highly sensitive to the selected EEG montage and the location of the EEG sensors on the scalp. Here, the cortical source of EEG oscillations predicting response amplitude does not correspond to the cortical target of the stimulation, indicating that even in this simple case, a specific neuronal pathway from somatosensory cortex to primary motor cortex is involved.

Automated methodology for optimal selection of minimum electrode subsets for accurate EEG source estimation based on Genetic Algorithm optimization

High-density Electroencephalography (HD-EEG) has proven to be the EEG montage that estimates the neural activity inside the brain with highest accuracy. Multiple studies have reported the effect of electrode number on source localization for specific sources and specific electrode configurations. The electrodes for these configurations are often manually selected to uniformly cover the entire head, going from 32 to 128 electrodes, but electrode configurations are not often selected according to their contribution to estimation accuracy. In this work, an optimization-based study is proposed to determine the minimum number of electrodes that can be used and to identify the optimal combinations of electrodes that can retain the localization accuracy of HD-EEG reconstructions. This optimization approach incorporates scalp landmark positions of widely used EEG montages. In this way, a systematic search for the minimum electrode subset is performed for single- and multiple-source localization problems. The Non-dominated Sorting Genetic Algorithm II (NSGA-II) combined with source reconstruction methods is used to formulate a multi-objective optimization problem that concurrently minimizes (1) the localization error for each source and (2) the number of required EEG electrodes. The method can be used for evaluating the source localization quality of low-density EEG systems (e.g. consumer-grade wearable EEG). We performed an evaluation over synthetic and real EEG datasets with known ground-truth. The experimental results show that optimal subsets with 6 electrodes can attain an equal or better accuracy than HD-EEG (with more than 200 channels) for a single source case. This happened when reconstructing a particular brain activity in more than 88% of the cases in synthetic signals and 63% in real signals, and in more than 88% and 73% of cases when considering optimal combinations with 8 channels. For a multiple-source case of three sources (only with synthetic signals), it was found that optimized combinations of 8, 12 and 16 electrodes attained an equal or better accuracy than HD-EEG with 231 electrodes in at least 58%, 76%, and 82% of cases respectively. Additionally, for such electrode numbers, lower mean errors and standard deviations than with 231 electrodes were obtained.

Effect of Transcranial direct current stimulation (tDCS) on altered conscious patients after traumatic brain injury &amp; cerebrovascular accident: A randomized clinical control trial

Background &amp; Objective: Traumatic brain injury and cerebrovascular accidents can result in altered levels of consciousness. This study is aimed at finding the effect of transcranial direct current stimulation (tDCS) on the level of consciousness in these patients. Methods: A total of 100 patients admitted to the neurological ICU of the hospital were screened and 40 subjects after satisfying inclusion criteria were recruited within the first one to two weeks of injury. They were randomly divided into two groups after written consent from a caretaker, Group A (experimental) (n=20) and Group B (control) (n=20) by computerized randomization. Group A received Anodal tDCS to the motor area (C3/C4 ipsilesional), sensory area (P3/P4 ipsilesional) and left dorsolateral prefrontal cortex (F3) according to the 10/20 EEG montage for two sessions of 20min/day for 7 consecutive days and routine physiotherapy. Group B only received routine physiotherapy similar to Group A. Glasgow coma scale (GCS) and Rancho Los Amigos scale (RLAS) was taken pre and post- intervention to assess the level of consciousness. Results: The pretest and post- test GCS and RLAS scores in groups A and B showed statistical significance at p&lt;0.01. The differences of mean GCS and RLAS between pretest and posttest in group A showed better improvement than that of group B. The results were statistically significant at p&lt;0.01. The effect size was large, calculated by Cohen’s d. Conclusion: The tDCS can be effective in improving GCS and RLAS in altered consciousness patients in the acute period after injury. It is non-invasive, cost-effective with minimal contraindications, and doesnot interfere with other modalities in the intensive care unit.

Seizure classification with selected frequency bands and EEG montages: a Natural Language Processing approach

Individualized treatment is crucial for epileptic patients with different types of seizures. The differences among patients impact the drug choice as well as the surgery procedure. With the advance in machine learning, automatic seizure detection can ease the manual time-consuming and labor-intensive procedure for diagnose seizure in the clinical setting. In this paper, we present an electroencephalography (EEG) frequency bands (sub-bands) and montages selection (sub-zones) method for classifier training that exploits Natural Language Processing from individual patients’ clinical report. The proposed approach is targeting for individualized treatment. We integrated the prior knowledge from patient’s reports into the classifier-building process, mimicking the authentic thinking process of experienced neurologist’s when diagnosing seizure using EEG. The keywords from clinical documents are mapped to the EEG data in terms of frequency bands and scalp EEG electrodes. The data of experiments are from the Temple University Hospital EEG seizure corpus, and the dataset is divided based on each group of patients with same seizure type and same recording electrode references. The classifier includes Random Forest, Support Vector Machine and Multi-Layer Perceptron. The classification performance indicates that competitive results can be achieve with a small portion of EEG the data. Using the sub-zones selection for Generalized Seizures (GNSZ) on all three electrodes, data are reduced by nearly 50% while the performance metrics remain at the same level with the whole frequency and zones. Moreover, when selecting by sub-zones and sub-bands together for GNSZ with Linked Ears reference, the data range reduced to 0.3% of whole range, and the performance deviates less than 3% from the results with whole range of data. Results show that using proposed approach may lead to more efficient implementations of the seizure classifier to be executed on power-efficient devices for long lasting real-time seizures detection.

Electroencephalogram in low-risk term newborns predicts neurodevelopmental metrics at age two years

ObjectiveTo determine whether neurodevelopmental biomarkers at 2 years of age are already present in the newborns’ EEG at birth. MethodsLow-risk term newborns were enrolled and studied utilizing EEG prior to discharge from the birth hospital. A 14-channel EEG montage (scalp-level) and source signals were calculated using the EEG. Their spectral power was calculated for each of the five frequency bands. Cognitive, language and motor skills were assessed using the Bayley Scales of Infant Development-III at age 2 years. The relationship between the spectral power in each frequency band and neurodevelopmental scores were quantified using the Spearman’s r. The role of gender, gestational age (GA) and delivery mode, if found significant (P < 0.05), were controlled by analyzing partial correlation. ResultsWe studied 47 newborns and found a significant association between gender, and delivery mode with EEG power. Scalp- and source-level spectral powers were positively associated with cognitive and language scores. At the source level, significant associations were identified in the parietal and occipital regions. ConclusionsElectrophysiological biomarkers of neurodevelopment at age 2 years are already present at birth in low-risk term infants. SignificanceLow-risk newborns’ EEG utility as a screening tool to optimize neurodevelopmental outcome warrants further evaluation.

Effects of a reduction of the number of electrodes in the EEG montage on the number of identified seizure patterns

Continuous EEG monitoring (cEEG) is frequently used in neurocritical care. The detection of seizures is one of the main objectives. The placement of the EEG electrodes is time consuming, therefore a reduced montage might lead to an increased availability in the ICU setting. It is unknown whether such a reduction of electrodes reduces the number of seizure patterns that are detected. A total of 95 seizure and 95 control EEG sequences from a pediatric epilepsy monitoring unit (EMU) were anonymized and reduced to an eight-lead montage. Two experts evaluated the recordings and the seizure detection rates using the reduced and the full montage were compared. Sensitivity and specificity for the seizure detection were calculated using the original EMU findings as gold standard. The sensitivity to detect seizures was 0.65 for the reduced montage compared to 0.76 for the full montage (p = 0.031). The specificities (0.97 and 0.96) were comparable (p = 1). A total of 4/9 (44%) of the generalized, 12/44 (27%) of the frontal, 6/14 (43%) of the central, 0/1 (0%) of the occipital, 6/20 (30%) of the temporal, and 5/7 (71%) of the parietal seizure patterns were not detected using the reduced montage. The median time difference between the onset of the seizure pattern in the full and reduced montage was 0.026s (IQR 5.651s). In this study the reduction of the EEG montage from 21 to eight electrodes reduced the sensitivity to detect seizure patterns from 0.76 to 0.65. The specificity remained virtually unchanged.

Robust learning from corrupted EEG with dynamic spatial filtering

Building machine learning models using EEG recorded outside of the laboratory setting requires methods robust to noisy data and randomly missing channels. This need is particularly great when working with sparse EEG montages (1–6 channels), often encountered in consumer-grade or mobile EEG devices. Neither classical machine learning models nor deep neural networks trained end-to-end on EEG are typically designed or tested for robustness to corruption, and especially to randomly missing channels. While some studies have proposed strategies for using data with missing channels, these approaches are not practical when sparse montages are used and computing power is limited (e.g., wearables, cell phones). To tackle this problem, we propose dynamic spatial filtering (DSF), a multi-head attention module that can be plugged in before the first layer of a neural network to handle missing EEG channels by learning to focus on good channels and to ignore bad ones. We tested DSF on public EEG data encompassing ∼4000 recordings with simulated channel corruption and on a private dataset of ∼100 at-home recordings of mobile EEG with natural corruption. Our proposed approach achieves the same performance as baseline models when no noise is applied, but outperforms baselines by as much as 29.4% accuracy when significant channel corruption is present. Moreover, DSF outputs are interpretable, making it possible to monitor the effective channel importance in real-time. This approach has the potential to enable the analysis of EEG in challenging settings where channel corruption hampers the reading of brain signals.

EEG Abnormalities and Their Radiographic Correlates in a COVID-19 Inpatient Cohort.

To identify the prevalence of EEG abnormalities in patients with coronavirus disease 2019 (COVID-19) with neurologic changes, their associated neuroimaging abnormalities, and rates of mortality. A retrospective case series of 192 adult COVID-19-positive inpatients with EEG performed between March and June 2020 at 4 hospitals: 161 undergoing continuous, 24 routine, and 7 reduced montage EEG. Study indication, epilepsy history, intubation status, administration of sedatives or antiseizure medications (ASMs), metabolic abnormalities, neuroimaging pathology associated with epileptiform abnormalities, and in-hospital mortality were analyzed. EEG indications included encephalopathy (54.7%), seizure (18.2%), coma (17.2%), focal deficit (5.2%), and abnormal movements (4.6%). Epileptiform abnormalities occurred in 39.6% of patients: focal intermittent epileptiform discharges in 25.0%, lateralized periodic discharges in 6.3%, and generalized periodic discharges in 19.3%. Seizures were recorded in 8 patients, 3 with status epilepticus. ASM administration, epilepsy history, and older age were associated with epileptiform abnormalities. Only 26.3% of patients presented with any epileptiform abnormality, 37.5% with electrographic seizures, and 25.7% patients with clinical seizures had known epilepsy. Background findings included generalized slowing (88.5%), focal slowing (15.6%), burst suppression (3.6%), attenuation (3.1%), and normal EEG (3.1%). Neuroimaging pathology was identified in 67.1% of patients with epileptiform abnormalities, over two-thirds acute. In-hospital mortality was 39.5% for patients with epileptiform abnormalities and 36.2% for those without. Risk factors for mortality were coma and ventilator support at time of EEG. This article highlights the range of EEG abnormalities frequently associated with acute neuroimaging abnormalities in COVID-19. Mortality rates were high, particularly for patients in coma requiring mechanical ventilation. These findings may guide the prognosis and management of patients with COVID-19 and neurologic changes.

Feasibility of Reconstructing Source Functional Connectivity with Low-Density EEG

ObjectivesFunctional connectivity (FC) is increasingly used as target for neuromodulation and enhancement of performance. A reliable assessment of FC with electroencephalography (EEG) currently requires a laboratory environment with high-density montages and a long preparation time. This study investigated the feasibility of reconstructing source FC with a low-density EEG montage towards a usage in real life applications.MethodsSource FC was reconstructed with inverse solutions and quantified as node degree of absolute imaginary coherence in alpha frequencies. We used simulated coherent point sources as well as two real datasets to investigate the impact of electrode density (19 vs. 128 electrodes) and usage of template vs. individual MRI-based head models on localization accuracy. In addition, we checked whether low-density EEG is able to capture inter-individual variations in coherence strength.ResultsIn numerical simulations as well as real data, a reduction of the number of electrodes led to less reliable reconstructions of coherent sources and of coupling strength. Yet, when comparing different approaches to reconstructing FC from 19 electrodes, source FC obtained with beamformers outperformed sensor FC, FC computed after independent component analysis, and source FC obtained with sLORETA. In particular, only source FC based on beamformers was able to capture neural correlates of motor behavior.ConclusionReconstructions of FC from low-density EEG is challenging, but may be feasible when using source reconstructions with beamformers.