Adaptive Mixture Independent Component Analysis Research Articles

Mobile Electroencephalography Reveals Differences in Cortical Processing During Exercises With Lower and Higher Cognitive Demands in Preadolescent Children.

The aim of this study was to examine whether cortical activity changes during exercise with increasing cognitive demands in preadolescent children. Twenty healthy children (8.75 [0.91]y) performed one movement game, which was conducted with lower and higher cognitive demands. During a baseline measurement and both exercise conditions, cortical activity was recorded using a 64-channel electroencephalographic system, and heart rate was assessed. Ratings of perceived excertion and perceived cognitive engagement were examined after each condition. To analyze power spectral density in the theta, alpha-1, and alpha-2 frequency bands, an adaptive mixture independent component analysis was used to determine the spatiotemporal sources of cortical activity, and brain components were clustered to identify spatial clusters. One-way repeated-measures analyses of variance revealed significant main effects for condition on theta in the prefrontal cluster, on alpha-1 in the prefrontal, central, bilateral motor, bilateral parieto-occipital, and occipital clusters, and on alpha-2 in the left motor, central, and left parieto-occipital clusters. Compared with the lower cognitive demand exercise, cortical activity was significantly higher in theta power in the prefrontal cluster and in alpha-1 power in the occipital cluster during the higher cognitive demand exercise. The present study shows that exercise complexity seems to influence cortical processing as it increased with increasing cognitive demands.

Unsupervised learning of brain state dynamics during emotion imagination using high-density EEG

This study applies adaptive mixture independent component analysis (AMICA) to learn a set of ICA models, each optimized by fitting a distributional model for each identified component process while maximizing component process independence within some subsets of time points of a multi-channel EEG dataset. Here, we applied 20-model AMICA decomposition to long-duration (1–2 h), high-density (128-channel) EEG data recorded while participants used guided imagination to imagine situations stimulating the experience of 15 specified emotions. These decompositions tended to return models identifying spatiotemporal EEG patterns or states within single emotion imagination periods. Model probability transitions reflected time-courses of EEG dynamics during emotion imagination, which varied across emotions. Transitions between models accounting for imagined “grief” and “happiness” were more abrupt and better aligned with participant reports, while transitions for imagined “contentment” extended into adjoining “relaxation” periods. The spatial distributions of brain-localizable independent component processes (ICs) were more similar within participants (across emotions) than emotions (across participants). Across participants, brain regions with differences in IC spatial distributions (i.e., dipole density) between emotion imagination versus relaxation were identified in or near the left rostrolateral prefrontal, posterior cingulate cortex, right insula, bilateral sensorimotor, premotor, and associative visual cortex. No difference in dipole density was found between positive versus negative emotions. AMICA models of changes in high-density EEG dynamics may allow data-driven insights into brain dynamics during emotional experience, possibly enabling the improved performance of EEG-based emotion decoding and advancing our understanding of emotion.

Which BSS method separates better the EEG Signals? A comparison of five different algorithms

A very common strategy for rejecting electroencephalographic (EEG) artifacts, includes the decomposition of filtered EEG signals using a Blind Source Separation (BSS) algorithm, the identification and removal of artifactual components and the reconstruction of the cleaned EEG signals. In this pipeline, the performance of the BSS algorithm, which is defined as its ability to separate properly the independent sources (like the EEG from artifactual sources), is very crucial for rejecting most of the artifacts, while maintaining the most part of EEG intact. The overwhelming majority of the published papers uses the extended INFOMAX version of Independent Component Analysis (ICA) for artifact rejection purposes. But is this the most efficient algorithm to separate EEG signals into independent components? This study comes to shed light to the aforementioned question by assessing the performance of the five most common BSS algorithms. The normalized entropy of the brain-related components, their correlation between independent components with the original sources and the amount of the overall mutual information reduction (MIR) achieved by each decomposition were computed in datasets with systematically varying numbers of electrodes (ranging from 19 tο 99), from 26 real human scalp EEG recordings. Additionally, 54 different datasets containing artificially contaminated EEG signals were also examined for the same purpose, on the basis of the Euclidean distance and the correlation, between the generated Independent Components (ICs) and the original vertical and horizontal eye signals, which were used for the contamination. The results support that the Adaptive Mixture ICA was the best performing BSS method.

0808 High Density EEG Correlates Of NREM Sleep Parasomnia Episodes

Abstract Introduction Parasomnia episodes (PE) consist of abnormal behaviors during sleep. Using high-density EEG (HDEEG), we sought to quantify topographical differences in spectral power during PE in comparison to wake and sleep. Methods 17 adult subjects with a history of NREM sleep parasomnia underwent 256-electrode HDEEG recordings during recovery sleep after 25h of sleep deprivation. PE occurred either spontaneously or when triggered by a sound. Data preprocessing of PE, sleep and wake data included filtering at 1-25 Hz, careful epoch and channel selection, and adaptive mixture independent component analysis (AMICA). We compared topographies of delta (slow wave activity, or SWA) and theta power, alpha power, and beta/delta ratio (a marker of cortical arousal) between states using paired t-tests. All results were thresholded at p<0.05 corrected for multiple comparison using statistical non parametric mapping (SNPM). Results Clean data were obtained in 26 PE arising out of N2/N3 sleep in 11 subjects. During PE, delta and theta power were significantly higher than during wake but lower than during sleep in central regions (at uncorrected p<0.05 for sleep vs. PE delta power). Occipital alpha was lower during PE compared to wake, but higher during PE compared to sleep. Finally, beta/delta ratio values during PE were globally higher than in wake, but globally lower than during sleep. Conclusion The present results confirm and extend our previous findings of decreased SWA in central areas during baseline sleep in patients with NREM sleep PE. They suggest that higher cortical arousal in central regions may precipitate motor behaviors during PE. Alpha power and beta-delta ratio during PE were intermediate between sleep and wake, suggesting that PE are transitional states with an admixture of cortical arousal and cortical sleep. Future analyses will use source reconstruction to identify the cortical generators of observed scalp differences. Support This work was funded by the Swiss National Science Foundation and the Tiny Blue Dot foundation.

Resting-State NIRS-EEG in Unresponsive Patients with Acute Brain Injury: A Proof-of-Concept Study.

Neurovascular-based imaging techniques such as functional MRI (fMRI) may reveal signs of consciousness in clinically unresponsive patients but are often subject to logistical challenges in the intensive care unit (ICU). Near-infrared spectroscopy (NIRS) is another neurovascular imaging technique but low cost, can be performed serially at the bedside, and may be combined with electroencephalography (EEG), which are important advantages compared to fMRI. Combined NIRS-EEG, however, has never been evaluated for the assessment of neurovascular coupling and consciousness in acute brain injury. We explored resting-state oscillations in eight-channel NIRS oxyhemoglobin and eight-channel EEG band-power signals to assess neurovascular coupling, the prerequisite for neurovascular-based imaging detection of consciousness, in patients with acute brain injury in the ICU (n = 9). Conscious neurological patients from step-down units and wards served as controls (n = 14). Unsupervised adaptive mixture-independent component analysis (AMICA) was used to correlate NIRS-EEG data with levels of consciousness and clinical outcome. Neurovascular coupling between NIRS oxyhemoglobin (0.07-0.13Hz) and EEG band-power (1-12Hz) signals at frontal areas was sensitive and prognostic to changing consciousness levels. AMICA revealed a mixture of five models from EEG data, with the relative probabilities of these models reflecting levels of consciousness over multiple days, although the accuracy was less than 85%. However, when combined with two channels of bilateral frontal neurovascular coupling, weighted k-nearest neighbor classification of AMICA probabilities distinguished unresponsive patients from conscious controls with > 90% accuracy (positive predictive value 93%, false discovery rate 7%) and, additionally, identified patients who subsequently failed to recover consciousness with > 99% accuracy. We suggest that NIRS-EEG for monitoring of acute brain injury in the ICU is worthy of further exploration. Normalization of neurovascular coupling may herald recovery of consciousness after acute brain injury.

Pathway-specific modulatory effects of neuromuscular electrical stimulation during pedaling in chronic stroke survivors

BackgroundNeuromuscular electrical stimulation (NMES) is extensively used in stroke motor rehabilitation. How it promotes motor recovery remains only partially understood. NMES could change muscular properties, produce altered sensory inputs, and modulate fluctuations of cortical activities; but the potential contribution from cortico-muscular couplings during NMES synchronized with dynamic movement has rarely been discussed.MethodWe investigated cortico-muscular interactions during passive, active, and NMES rhythmic pedaling in healthy subjects and chronic stroke survivors. EEG (128 channels), EMG (4 unilateral lower limb muscles) and movement parameters were measured during 3 sessions of constant-speed pedaling. Sensory-level NMES (20 mA) was applied to the muscles, and cyclic stimulation patterns were synchronized with the EMG during pedaling cycles. Adaptive mixture independent component analysis was utilized to determine the movement-related electro-cortical sources and the source dipole clusters. A directed cortico-muscular coupling analysis was conducted between representative source clusters and the EMGs using generalized partial directed coherence (GPDC). The bidirectional GPDC was compared across muscles and pedaling sessions for post-stroke and healthy subjects.ResultsDirected cortico-muscular coupling of NMES cycling was more similar to that of active pedaling than to that of passive pedaling for the tested muscles. For healthy subjects, sensory-level NMES could modulate GPDC of both ascending and descending pathways. Whereas for stroke survivors, NMES could modulate GPDC of only the ascending pathways.ConclusionsBy clarifying how NMES influences neuromuscular control during pedaling in healthy and post-stroke subjects, our results indicate the potential limitation of sensory-level NMES in promoting sensorimotor recovery in chronic stroke survivors.

Modeling brain dynamic state changes with adaptive mixture independent component analysis

There is a growing interest in neuroscience in assessing the continuous, endogenous, and nonstationary dynamics of brain network activity supporting the fluidity of human cognition and behavior. This non-stationarity may involve ever-changing formation and dissolution of active cortical sources and brain networks. However, unsupervised approaches to identify and model these changes in brain dynamics as continuous transitions between quasi-stable brain states using unlabeled, noninvasive recordings of brain activity have been limited. This study explores the use of adaptive mixture independent component analysis (AMICA) to model multichannel electroencephalographic (EEG) data with a set of ICA models, each of which decomposes an adaptively learned portion of the data into statistically independent sources. We first show that AMICA can segment simulated quasi-stationary EEG data and accurately identify ground-truth sources and source model transitions. Next, we demonstrate that AMICA decomposition, applied to 6–13 channel scalp recordings from the CAP Sleep Database, can characterize sleep stage dynamics, allowing 75% accuracy in identifying transitions between six sleep stages without use of EEG power spectra. Finally, applied to 30-channel data from subjects in a driving simulator, AMICA identifies models that account for EEG during faster and slower response to driving challenges, respectively. We show changes in relative probabilities of these models allow effective prediction of subject response speed and moment-by-moment characterization of state changes within single trials. AMICA thus provides a generic unsupervised approach to identifying and modeling changes in EEG dynamics. Applied to continuous, unlabeled multichannel data, AMICA may likely be used to detect and study any changes in cognitive states.

Exergames Inherently Contain Cognitive Elements as Indicated by Cortical Processing.

Exergames are increasingly used to train both physical and cognitive functioning, but direct evidence whether and how exergames affect cortical activity is lacking. Although portable electroencephalography (EEG) can be used while exergaming, it is unknown whether brain activity will be obscured by movement artifacts. The aims of this study were to assess whether electrophysiological measurements during exergaming are feasible and if so, whether cortical activity changes with additional cognitive elements. Twenty-four young adults performed self-paced sideways leaning movements, followed by two blocks of exergaming in which participants completed a puzzle by leaning left or right to select the correct piece. At the easy level, only the correct piece was shown, while two pieces were presented at the choice level. Brain activity was recorded using a 64-channel passive EEG system. After filtering, an adaptive mixture independent component analysis identified the spatio-temporal sources of brain activity. Results showed that it is feasible to record brain activity in young adults while playing exergames. Furthermore, five spatially different clusters were identified located frontal, bilateral central, and bilateral parietal. The frontal cluster had significantly higher theta power in the exergaming condition with choice compared to self-paced leaning movements and exergaming without choice, while both central clusters showed a significant increase in absolute alpha-2 power in the exergaming conditions compared to the self-paced movements. This is the first study to show that it is feasible to record brain activity while exergaming. Furthermore, results indicated that even a simple exergame without explicit cognitive demands inherently requires cognitive processing. These results pave the way for studying brain activity during various exergames in different populations to help improve their effective use in rehabilitation settings.

High-resolution EEG source imaging of one-year-old children.

Recently we described an iterative skull conductivity and source location estimation (SCALE) algorithm for simultaneously estimating head tissue conductivities and brain source locations. SCALE uses a realistic FEM forward problem head model and scalp maps of 10 or more near-dipolar sources identified by independent component analysis (ICA) decomposition of sufficient high-density EEG data. In this study, we applied SCALE to 20 minutes of 64-channel EEG data and magnetic resonance (MR) head images from four twelve-months-of-age infants. For each child, we selected 15-16 near-dipolar independent components from multiple-model adaptive mixture ICA (AMICA) decomposition of their EEG data. SCALE converged to brain-to-skull conductivity ratio (BSCR) estimates in the 10-12 range and mostly compact gyral or sulcal cortical distributions for the IC sources.

Simultaneous head tissue conductivity and EEG source location estimation

Accurate electroencephalographic (EEG) source localization requires an electrical head model incorporating accurate geometries and conductivity values for the major head tissues. While consistent conductivity values have been reported for scalp, brain, and cerebrospinal fluid, measured brain-to-skull conductivity ratio (BSCR) estimates have varied between 8 and 80, likely reflecting both inter-subject and measurement method differences. In simulations, mis-estimation of skull conductivity can produce source localization errors as large as 3cm. Here, we describe an iterative gradient-based approach to Simultaneous tissue Conductivity And source Location Estimation (SCALE). The scalp projection maps used by SCALE are obtained from near-dipolar effective EEG sources found by adequate independent component analysis (ICA) decomposition of sufficient high-density EEG data. We applied SCALE to simulated scalp projections of 15cm2-scale cortical patch sources in an MR image-based electrical head model with simulated BSCR of 30. Initialized either with a BSCR of 80 or 20, SCALE estimated BSCR as 32.6. In Adaptive Mixture ICA (AMICA) decompositions of (45-min, 128-channel) EEG data from two young adults we identified sets of 13 independent components having near-dipolar scalp maps compatible with a single cortical source patch. Again initialized with either BSCR 80 or 25, SCALE gave BSCR estimates of 34 and 54 for the two subjects respectively. The ability to accurately estimate skull conductivity non-invasively from any well-recorded EEG data in combination with a stable and non-invasively acquired MR imaging-derived electrical head model could remove a critical barrier to using EEG as a sub-cm2-scale accurate 3-D functional cortical imaging modality.

Beta- and gamma-range human lower limb corticomuscular coherence

Coherence between electroencephalography (EEG) recorded on the scalp above the motor cortex and electromyography (EMG) recorded on the skin of the limbs is thought to reflect corticospinal coupling between motor cortex and muscle motor units. Beta-range (13–30 Hz) corticomuscular coherence has been extensively documented during static force output while gamma-range (31–45 Hz) coherence has been linked to dynamic force output. However, the explanation for this beta-to-gamma coherence shift remains unclear. We recorded 264-channel EEG and 8-channel lower limb EMG while eight healthy subjects performed isometric and isotonic, knee, and ankle exercises. Adaptive mixture independent component analysis (AMICA) parsed EEG into models of underlying source signals. We computed magnitude squared coherence between electrocortical source signals and EMG. Significant coherence between contralateral motor cortex electrocortical signals and lower limb EMG was observed in the beta- and gamma-range for all exercise types. Gamma-range coherence was significantly greater for isotonic exercises than for isometric exercises. We conclude that active muscle movement modulates the speed of corticospinal oscillations. Specifically, isotonic contractions shift corticospinal oscillations toward the gamma-range while isometric contractions favor beta-range oscillations. Prior research has suggested that tasks requiring increased integration of visual and somatosensory information may shift corticomuscular coherence to the gamma-range. The isometric and isotonic tasks studied here likely required similar amounts of visual and somatosensory integration. This suggests that muscle dynamics, including the amount and type of proprioception, may play a role in the beta-to-gamma shift.

An EEG-based study of discrete isometric and isotonic human lower limb muscle contractions

BackgroundElectroencephalography (EEG) combined with independent component analysis enables functional neuroimaging in dynamic environments including during human locomotion. This type of functional neuroimaging could be a powerful tool for neurological rehabilitation. It could enable clinicians to monitor changes in motor control related cortical dynamics associated with a therapeutic intervention, and it could facilitate noninvasive electrocortical control of devices for assisting limb movement to stimulate activity dependent plasticity. Understanding the relationship between electrocortical dynamics and muscle activity will be helpful for incorporating EEG-based functional neuroimaging into clinical practice. The goal of this study was to use independent component analysis of high-density EEG to test whether we could relate electrocortical dynamics to lower limb muscle activation in a constrained motor task. A secondary goal was to assess the trial-by-trial consistency of the electrocortical dynamics by decoding the type of muscle action.MethodsWe recorded 264-channel EEG while 8 neurologically intact subjects performed isometric and isotonic, knee and ankle exercises at two different effort levels. Adaptive mixture independent component analysis (AMICA) parsed EEG into models of underlying source signals. We generated spectrograms for all electrocortical source signals and used a naïve Bayesian classifier to decode exercise type from trial-by-trial time-frequency data.ResultsAMICA captured different electrocortical source distributions for ankle and knee tasks. The fit of single-trial EEG to these models distinguished knee from ankle tasks with 80% accuracy. Electrocortical spectral modulations in the supplementary motor area were significantly different for isometric and isotonic tasks (p < 0.05). Isometric contractions elicited an event related desynchronization (ERD) in the α-band (8–12 Hz) and β-band (12–30 Hz) at joint torque onset and offset. Isotonic contractions elicited a sustained α- and β-band ERD throughout the trial. Classifiers based on supplementary motor area sources achieved a 4-way classification accuracy of 69% while classifiers based on electrocortical sources in multiple brain regions achieved a 4-way classification accuracy of 87%.ConclusionsIndependent component analysis of EEG reveals unique spatial and spectro-temporal electrocortical properties for different lower limb motor tasks. Using a broad distribution of electrocortical signals may improve classification of human lower limb movements from single-trial EEG.

Electrocortical source imaging of intracranial EEG data in epilepsy.

Here we report first results of numerical methods for modeling the dynamic structure and evolution of epileptic seizure activity in an intracranial subdural electrode recording from a patient with partial refractory epilepsy. A 16-min dataset containing two seizures was decomposed using up to five competing adaptive mixture independent component analysis (AMICA) models. Multiple models modeled early or late ictal, or pre- or post-ictal periods in the data, respectively. To localize sources, a realistic Boundary Element Method (BEM) head model was constructed for the patient with custom open skull and plastic (non-conductive) electrode holder features. Source localization was performed using Sparse Bayesian Learning (SBL) on a dictionary of overlapping multi-scale cortical patches constructed from 80,130 dipoles in gray matter perpendicular to the cortical surface. Remaining mutual information among seizure-model AMICA components was dominated by two dependent component subspaces with largely contiguous source domains localized to superior frontal gyrus and precentral gyrus; these accounted for most of the ictal activity. Similar though much weaker dependent subspaces were also revealed in pre-ictal data by the associated AMICA model. Electrocortical source imaging appears promising both for clinical epilepsy research and for basic cognitive neuroscience research using volunteer patients who must undergo invasive monitoring for medical purposes.

Theta power as an intermediate phenotype of ADHD

Event Abstract Back to Event Theta power as an intermediate phenotype of ADHD Gráinne McLoughlin1*, Jason Palmer2, Fruhling Rijsdijk3 and Scott Makeig2 1 Institute of Psychiatry, King's College London, United Kingdom 2 SCCN, University of California San Diego, United States 3 King’s College London, United Kingdom Theta power has been consistently associated with attention deficit hyperactivity disorder (ADHD) and is heritable and thus may be a possible endophenotype of the disorder. The term endophenotype (EPT) has been used extensively in recent genetic research into the aetiology of psychiatric disorders. EPTs must be genetically and phenotypically associated with the disorder, and are proposed to be closer to the site of gene action. However, conventional definitions of EPTs do not discriminate between a ‘liability-index’ (LI) and a ‘mediational’ (ME) model. The LI model specifies that a common set of genes increase risk both for the disorder and for the EPT. The ME model makes the stronger assumption that the causal pathway from genes to disorder passes through EPT. In this study, we aimed to use multivariate twin modeling to compare the LI model and the ME model for theta power in the etiology of ADHD. We collected EEG data in a sample of 69 male twin pairs (13 – 15 years) in two resting conditions, with three tasks in between. Theta power was identified using adaptive mixture independent component analysis (AMICA). AMICA indicated similar states in both resting conditions, which reduced measurement error and allowed more accurate estimations of the genetic architecture. Theta power in the two conditions was highly correlated (r = .90). Structural equation modeling of the data, which segregates the genetic and environmental variance, indicated that theta power is highly heritable (80%) and shares genetic overlap with ADHD (40%). Further, the ME model fitted the data better than the LI model, which indicates that theta activity mediates genetic influences on ADHD. Theta power is therefore important in the etiology of ADHD and may be a useful target for intervention. Funding: National Institute of Health Research, Royal Society. Keywords: EPT, Neuropsychiatry Conference: XI International Conference on Cognitive Neuroscience (ICON XI), Palma, Mallorca, Spain, 25 Sep - 29 Sep, 2011. Presentation Type: Poster Presentation Topic: Poster Sessions: Neuropsychiatric Applications Citation: McLoughlin G, Palmer J, Rijsdijk F and Makeig S (2011). Theta power as an intermediate phenotype of ADHD. Conference Abstract: XI International Conference on Cognitive Neuroscience (ICON XI). doi: 10.3389/conf.fnhum.2011.207.00120 Copyright: The abstracts in this collection have not been subject to any Frontiers peer review or checks, and are not endorsed by Frontiers. They are made available through the Frontiers publishing platform as a service to conference organizers and presenters. The copyright in the individual abstracts is owned by the author of each abstract or his/her employer unless otherwise stated. Each abstract, as well as the collection of abstracts, are published under a Creative Commons CC-BY 4.0 (attribution) licence (https://creativecommons.org/licenses/by/4.0/) and may thus be reproduced, translated, adapted and be the subject of derivative works provided the authors and Frontiers are attributed. For Frontiers’ terms and conditions please see https://www.frontiersin.org/legal/terms-and-conditions. Received: 17 Nov 2011; Published Online: 28 Nov 2011. * Correspondence: Dr. Gráinne McLoughlin, Institute of Psychiatry, King's College London, London, United Kingdom, grainne.mcloughlin@kcl.ac.uk Login Required This action requires you to be registered with Frontiers and logged in. To register or login click here. Abstract Info Abstract The Authors in Frontiers Gráinne McLoughlin Jason Palmer Fruhling Rijsdijk Scott Makeig Google Gráinne McLoughlin Jason Palmer Fruhling Rijsdijk Scott Makeig Google Scholar Gráinne McLoughlin Jason Palmer Fruhling Rijsdijk Scott Makeig PubMed Gráinne McLoughlin Jason Palmer Fruhling Rijsdijk Scott Makeig Related Article in Frontiers Google Scholar PubMed Abstract Close Back to top Javascript is disabled. Please enable Javascript in your browser settings in order to see all the content on this page.