From active affordance to active inference: vertical integration of cognition in the cerebral cortex through dual subcortical control systems.

In previous papers, we proposed that the dorsal attention system's top-down control is regulated by the dorsal division of the limbic system, providing a feedforward or impulsive form of control generating expectancies during active inference. In contrast, we proposed that the ventral attention system is regulated by the ventral limbic division, regulating feedback constraints and error-correction for active inference within the neocortical hierarchy. Here, we propose that these forms of cognitive control reflect vertical integration of subcortical arousal control systems that evolved for specific forms of behavior control. The feedforward impetus to action is regulated by phasic arousal, mediated by lemnothalamic projections from the reticular activating system of the lower brainstem, and then elaborated by the hippocampus and dorsal limbic division. In contrast, feedback constraint-based on environmental requirements-is regulated by the tonic activation furnished by collothalamic projections from the midbrain arousal control centers, and then sustained and elaborated by the amygdala, basal ganglia, and ventral limbic division. In an evolutionary-developmental analysis, understanding these differing forms of active affordance-for arousal and motor control within the subcortical vertebrate neuraxis-may help explain the evolution of active inference regulating the cognition of expectancy and error-correction within the mammalian 6-layered neocortex.

Continuity and change in neural plasticity through embryonic morphogenesis, fetal activity-dependent synaptogenesis, and infant memory consolidation.

There is an apparent continuity in human neural development that can be traced to venerable themes of vertebrate morphogenesis that have shaped the evolution of the reptilian telencephalon (including both primitive three-layered cortex and basal ganglia) and then the subsequent evolution of the mammalian six-layered neocortex. In this theoretical analysis, we propose that an evolutionary-developmental analysis of these general morphogenetic themes can help to explain the embryonic development of the dual divisions of the limbic system that control the dorsal and ventral networks of the human neocortex. These include the archicortical (dorsal limbic) Papez circuits regulated by the hippocampus that organize spatial, contextual memory, as well as the paleocortical (ventral limbic) circuits that organize object memory. We review evidence that these dorsal and ventral limbic divisions are controlled by the differential actions of brainstem lemnothalamic and midbrain collothalamic arousal control systems, respectively, thereby traversing the vertebrate subcortical neuraxis. These dual control systems are first seen shaping the phyletic morphogenesis of the archicortical and paleocortical foundations of the forebrain in embryogenesis. They then provide dual modes of activity-dependent synaptic organization in the active (lemnothalamic) and quiet (collothalamic) stages of fetal sleep. Finally, these regulatory systems mature to form the major systems of memory consolidation of postnatal development, including the rapid eye movement (lemnothalamic) consolidation of implicit memory and social attachment in the first year, and then-in a subsequent stage-the non-REM (collothalamic) consolidation of explicit memory that is integral to the autonomy and individuation of the second year of life.

Adaptive control of functional connectivity: dorsal and ventral limbic divisions regulate the dorsal and ventral neocortical networks.

The connectional anatomy of the primate cortex is now well-defined by the Structural Model, in which adjacent cortical areas are interconnected in an organized network hierarchy of communication and control. The computational theory of "active inference" can be aligned with this architecture, proposing that predictions descend from higher association areas to be updated by ascending prediction errors from lower (i.e. primary) sensory and motor areas. Given the connectivity, the limbic networks at the apex of the cerebral hierarchy must then be responsible for the most general expectancies, which are propagated through the hierarchy to organize the multiple component network levels of experience and behavior. Anatomical evidence suggests that there are dual limbic divisions, reflecting archicortical (dorsal) and paleocortical (ventral) derivations, resulting in fundamentally different neural mechanisms for managing expectancies across the corticolimbic hierarchy. In the functional connectivity literature, the dorsal attention network is seen to provide top-down or endogenous control of attention, whereas the ventral attention network provides stimulus bound or exogenous attentional control. We review evidence indicating that the dorsal, archicortical division of the limbic system provides a feedforward, impulsive, endogenous mode of motive control, whereas the ventral, paleocortical limbic division provides feedback constraint linked to exogenous events.

Neurophysiological mechanisms of implicit and explicit memory in the process of consciousness.

Neurophysiological mechanisms are increasingly understood to constitute the foundations of human conscious experience. These include the capacity for ongoing memory, achieved through a hierarchy of reentrant cross-laminar connections across limbic, heteromodal, unimodal, and primary cortices. The neurophysiological mechanisms of consciousness also include the capacity for volitional direction of attention to the ongoing cognitive process, through a reentrant fronto-thalamo-cortical network regulation of the inhibitory thalamic reticular nucleus. More elusive is the way that discrete objects of subjective experience, such as the color of deep blue or the sound of middle C, could be generated by neural mechanisms. Explaining such ineffable qualities of subjective experience is what Chalmers has called "the hard problem of consciousness," which has divided modern neuroscientists and philosophers alike. We propose that insight into the appearance of the hard problem can be gained through integrating classical phenomenological studies of experience with recent progress in the differential neurophysiology of consolidating explicit versus implicit memory. Although the achievement of consciousness, once it is reflected upon, becomes explicit, the underlying process of generating consciousness, through neurophysiological mechanisms, is largely implicit. Studying the neurophysiological mechanisms of adaptive implicit memory, including brain stem, limbic, and thalamic regulation of neocortical representations, may lead to a more extended phenomenological understanding of both the neurophysiological process and the subjective experience of consciousness.NEW & NOTEWORTHY The process of consciousness, generating the qualia that may appear to be irreducible qualities of experience, can be understood to arise from neurophysiological mechanisms of memory. Implicit memory, organized by the lemnothalamic brain stem projections and dorsal limbic consolidation in REM sleep, supports the unconscious field and the quasi-conscious fringe of current awareness. Explicit memory, organized by the collothalamic midbrain projections and ventral limbic consolidation of NREM sleep, supports the focal objects of consciousness.

Focal limbic sources create the large slow oscillations of the EEG in human deep sleep.

BACKGROUND: Initial observations with the human electroencephalogram (EEG) have interpreted slow oscillations (SOs) of the EEG during deep sleep (N3) as reflecting widespread surface-negative traveling waves that originate in frontal regions and propagate across the neocortex. However, mapping SOs with a high-density array shows the simultaneous appearance of posterior positive voltage fields in the EEG at the time of the frontal-negative fields, with the typical inversion point (apparent source) around the temporal lobe. METHODS: Overnight 256-channel EEG recordings were gathered from 10 healthy young adults. Individual head conductivity models were created using each participant's own structural MRI. Source localization of SOs during N3 was then performed. RESULTS: Electrical source localization models confirmed that these large waves were created by focal discharges within the ventral limbic cortex, including medial temporal and caudal orbitofrontal cortex. CONCLUSIONS: Although the functional neurophysiology of deep sleep involves interactions between limbic and neocortical networks, the large EEG deflections of deep sleep are not created by distributed traveling waves in lateral neocortex but instead by relatively focal limbic discharges.

Motive control of unconscious inference: The limbic base of adaptive Bayes.

Current computational models of neocortical processing, described as predictive coding theory, are providing new ways of understanding Helmholtz's classical insight that perception cannot proceed in a data-driven fashion, but instead requires unconscious inference based on prior experience. Predictive coding is a Bayesian process, in which the operations at each lower level of the cortical hierarchy are predicted by prior projections of expectancies from a higher level, and are then updated by error-correction with lower level evidence. To generalize the predictive coding model to the human neocortex as a whole requires aligning the Bayesian negotiation of prior expectancies with sensory and motor evidence not only within the connectional architecture of the neocortex (primary sensory/motor, unimodal association areas, and heteromodal association areas) but also with the limbic cortex that forms the base for the adaptive control of the heteromodal areas and thereby the cerebral hemisphere as a whole. By reviewing the current evidence on the anatomy of the human corticolimbic connectivity (now formalized as the Structural Model) we address the problem of how limbic cortex resonates to the homeostatic, personal significance of events to provide Bayesian priors to organize the operations of predictive coding across the multiple levels of the neocortex. By reviewing both classical evidence and current models of control exerted between limbic and neocortical networks, we suggest a neuropsychological theory of human cognition, the adaptive Bayes process model, in which prior expectancies are not simply rationalized propositions, but rather affectively-charged expectancies that bias the interpretation of sensory data and action affordances to support allostasis, the motive control of expectancies for future events.

Transcranial Electrical Stimulation targeting limbic cortex increases the duration of human deep sleep.

BACKGROUND: Researchers have proposed that impaired sleep may be a causal link in the progression from Mild Cognitive Impairment (MCI) to Alzheimer's Disease (AD). Several recent findings suggest that enhancing deep sleep (N3) may improve neurological health in persons with MCI, and buffer the risk for AD. Specifically, Transcranial Electrical Stimulation (TES) of frontal brain areas, the inferred source of the Slow Oscillations (SOs) of N3 sleep, can extend N3 sleep duration and improve declarative memory for recently learned information. Recent work in our laboratory using dense array Electroencephalography (dEEG) localized the sources of SOs to anterior limbic sites - suggesting that targeting these sites with TES may be more effective for enhancing N3. METHODS: For the present study, we recruited 13 healthy adults (M = 42 years) to participate in three all-night sleep EEG recordings where they received low level (0.5 mA) TES designed to target anterior limbic areas and a sham stimulation (placebo). We used a convolutional neural network, trained and tested on professionally scored EEG sleep staging, to predict sleep stages for each recording. RESULTS: When compared to the sham session, limbic-targeted TES significantly increased the duration of N3 sleep. TES also significantly increased spectral power in the 0.5-1 Hz frequency band (relative to pre-TES epochs) in left temporoparietal and left occipital scalp regions compared to sham. CONCLUSION: These results suggest that even low-level TES, when specifically targeting anterior limbic sites, can increase deep (N3) sleep and thereby contribute to healthy sleep quality.

Source localization of epileptic spikes using Multiple Sparse Priors.

OBJECTIVE: To evaluate epileptic source estimation using multiple sparse priors (MSP) inverse method and high-resolution, individual electrical head models. METHODS: Accurate source localization is dependent on accurate electrical head models and appropriate inverse solvers. Using high-resolution, individual electrical head models in fifteen epilepsy patients, with surgical resection and clinical outcome as criteria for accuracy, performance of MSP method was compared against standardized low-resolution brain electromagnetic tomography (sLORETA) and coherent maximum entropy on the mean (cMEM) methods. RESULTS: The MSP method performed similarly to the sLORETA method and slightly better than the cMEM method in terms of success rate. The MSP and cMEM methods were more focal than sLORETA with the advantage of not requiring an arbitrary selection of a hyperparameter or thresholding of reconstructed current density values to determine focus. MSP and cMEM methods were better than sLORETA in terms of spatial dispersion. CONCLUSIONS: Results suggest that the three methods are complementary and could be used together. In practice, the MSP method will be easier to use and interpret compared to sLORETA, and slightly more accurate and faster than the cMEM method. SIGNIFICANCE: Source localization of interictal spikes from dense-array electroencephalography data has been shown to be a reliable marker of epileptic foci and useful for pre-surgical planning. The advantages of MSP make it a useful complement to other inverse solvers in clinical practice.

Spatiotemporal Dynamics of Multiple Memory Systems During Category Learning.

The brain utilizes distinct neural mechanisms that ease the transition through different stages of learning. Furthermore, evidence from category learning has shown that dissociable memory systems are engaged, depending on the structure of a task. This can even hold true for tasks that are very similar to each other, which complicates the process of classifying brain activity as relating to changes that are associated with learning or reflecting the engagement of a memory system suited for the task. The primary goals of these studies were to characterize the mechanisms that are associated with category learning and understand the extent to which different memory systems are recruited within a single task. Two studies providing spatial and temporal distinctions between learning-related changes in the brain and category-dependent memory systems are presented. The results from these experiments support the notion that exemplar memorization, rule-based, and perceptual similarity-based categorization are flexibly recruited in order to optimize performance during a single task. We conclude that these three methods, along with the memory systems they rely on, aid in the development of expertise, but their engagement might depend on the level of familiarity with a category.

Safety of slow-pulsed transcranial electrical stimulation in acute spike suppression.

We examined the effects of slow-pulsed transcranial electrical stimulation (TES) in suppressing epileptiform discharges in seven adults with refractory epilepsy. An MRI-based realistic head model was constructed for each subject and co-registered with 256-channel dense EEG (dEEG). Interictal spikes were localized, and TES targeted the cortical source of each subject's principal spike population. Targeted spikes were suppressed in five subject's (29/35 treatment days overall), and nontargeted spikes were suppressed in four subjects. Epileptiform activity did not worsen. This study suggests that this protocol, designed to induce long-term depression (LTD), is safe and effective in acute suppression of interictal epileptiform discharges.

EEG source imaging of epileptic activity at seizure onset.

Surgical resection of the seizure onset zone (SOZ) requires that this region of the cortex is accurately localized. The onset of a seizure may be marked by transient discharges, but it also may be accompanied by oscillatory, sinusoidal electrographic activity, such as the EEG theta rhythm. However, because of the superposition of the seizure signal with other electrical signals, including noise artifacts and non-seizure brain activity, noninvasive Electrical Source Imaging (ESI) of the ictal EEG activity at seizure onset remains a challenging task for surgical planning. In the present study, we localize the SOZ from oscillatory features of the EEG at the ictal onset using 256-channel high density electroencephalography (HD-EEG), exact sensor positions, and individual electrical head models constructed from the patient's T1 magnetic resonance image (MRI). Epileptic activities at the seizure onset were characterized with joint time-frequency analysis and source estimated by standardized low resolution electromagnetic tomography (sLORETA) inverse method. The consistency of this localization was examined across multiple seizures for individual patients. For validation, results were compared to three clinical criteria: (1) epileptogenic lesions, (2) seizure onset observed in intracranial EEG, and (3) successful surgical outcomes. In this set of 84 seizures, the onsets of 56 seizures could be localized. For the lateralization measure, the results from HD-EEG with interictal spikes (8/10) and with ictal onset (10/10) were more accurate than international 10-20 EEG for interictal spikes (5/10) and ictal onset (5/10). ESI from HD-EEG with ictal onset (9/10) had greater concordance to the clinical criteria than HD-EEG with interictal spikes (6/10). Noninvasive ESI of oscillatory features at ictal onset using 256-channel HD-EEG and high-resolution individual head models can make a useful contribution to the clinical localization of the SOZ in presurgical planning.

Skull Modeling Effects in Conductivity Estimates Using Parametric Electrical Impedance Tomography.

OBJECTIVE: To estimate scalp, skull, compact bone, and marrow bone electrical conductivity values based on electrical impedance tomography (EIT) measurements, and to determine the influence of skull modeling details on the estimates. METHODS: We collected EIT data with 62 current injection pairs and built five 6-8 million finite element (FE) head models with different grades of skull simplifications for four subjects, including three whose head models serve as Atlases in the scientific literature and in commercial equipment (Colin27 and EGI's Geosource atlases). We estimated electrical conductivity of the scalp, skull, marrow bone, and compact bone tissues for each current injection pair, each model, and each subject. RESULTS: Closure of skull holes in FE models, use of simplified four-layer boundary element method-like models, and neglecting the CSF layer produce an overestimation of the skull conductivity of 10%, 10%-20%, and 20%-30%, respectively (accumulated overestimation of 50%-70%). The average extracted conductivities are 288 ± 53 (the scalp), 4.3 ± 0.08 (the compact bone), and 5.5 ± 1.25 (the whole skull) mS/m. The marrow bone estimates showed large dispersion. CONCLUSION: Present EIT estimates for the skull conductivity are lower than typical literature reference values, but previous in vivo EIT results are likely overestimated due to the use of simpler models. SIGNIFICANCE: Typical literature values of 7-10 mS/m for skull conductivity should be replaced by the present estimated values when using detailed skull head models. We also provide subject specific conductivity estimates for widely used Atlas head models.

Oscillatory Patterns of Phase Cone Formations near to Epileptic Spikes Derived from 256-Channel Scalp EEG Data.

Our objective was to determine if there are any distinguishable phase cone clustering patterns present near to epileptic spikes. These phase cones arise from episodic phase shifts due to the coordinated activity of cortical neurons at or near to state transitions and can be extracted from the high-density scalp EEG recordings. The phase cone clustering activities in the low gamma band (30-50 Hz) and in the ripple band (80-150 Hz) were extracted from the analytic phase after taking Hilbert transform of the 256-channel high density (dEEG) data of adult patients. We used three subjects in this study. Spatiotemporal contour plots of the unwrapped analytic phase with 1.0 ms intervals were constructed using a montage layout of 256 electrode positions. Stable phase cone patterns were selected based on the criteria that the sign of the spatial gradient did not change for at least three consecutive time samples and the frame velocity was within the range of propagation velocities of cortical axons. These plots exhibited dynamical formation of phase cones which were higher in the seizure area as compared with the nearby surrounding brain areas. Spatiotemporal oscillatory patterns were also visible during ±5 sec period from the location of the spike. These results suggest that the phase cone activity might be useful for noninvasive localization of epileptic sites and also for examining the cortical neurodynamics near to epileptic spikes.

Dynamic Responses in Brain Networks to Social Feedback: A Dual EEG Acquisition Study in Adolescent Couples.

Adolescence is a sensitive period for the development of romantic relationships. During this period the maturation of frontolimbic networks is particularly important for the capacity to regulate emotional experiences. In previous research, both functional magnetic resonance imaging (fMRI) and dense array electroencephalography (dEEG) measures have suggested that responses in limbic regions are enhanced in adolescents experiencing social rejection. In the present research, we examined social acceptance and rejection from romantic partners as they engaged in a Chatroom Interact Task. Dual 128-channel dEEG systems were used to record neural responses to acceptance and rejection from both adolescent romantic partners and unfamiliar peers (N = 75). We employed a two-step temporal principal component analysis (PCA) and spatial independent component analysis (ICA) approach to statistically identify the neural components related to social feedback. Results revealed that the early (288 ms) discrimination between acceptance and rejection reflected by the P3a component was significant for the romantic partner but not the unfamiliar peer. In contrast, the later (364 ms) P3b component discriminated between acceptance and rejection for both partners and peers. The two-step approach (PCA then ICA) was better able than either PCA or ICA alone in separating these components of the brain's electrical activity that reflected both temporal and spatial phases of the brain's processing of social feedback.

Slow-Frequency Pulsed Transcranial Electrical Stimulation for Modulation of Cortical Plasticity Based on Reciprocity Targeting with Precision Electrical Head Modeling.

In pain management as well as other clinical applications of neuromodulation, it is important to consider the timing parameters influencing activity-dependent plasticity, including pulsed versus sustained currents, as well as the spatial action of electrical currents as they polarize the complex convolutions of the cortical mantle. These factors are of course related; studying temporal factors is not possible when the spatial resolution of current delivery to the cortex is so uncertain to make it unclear whether excitability is increased or decreased with anodal vs. cathodal current flow. In the present study we attempted to improve the targeting of specific cortical locations by applying current through flexible source-sink configurations of 256 electrodes in a geodesic array. We constructed a precision electric head model for 12 healthy individuals. Extraction of the individual's cortical surface allowed computation of the component of the induced current that is normal to the target cortical surface. In an effort to replicate the long-term depression (LTD) induced with pulsed protocols in invasive animal research and transcranial magnetic stimulation studies, we applied 100 ms pulses at 1.9 s intervals either in cortical-surface-anodal or cortical-surface-cathodal directions, with a placebo (sham) control. The results showed significant LTD of the motor evoked potential as a result of the cortical-surface-cathodal pulses in contrast to the placebo control, with a smaller but similar LTD effect for anodal pulses. The cathodal LTD after-effect was sustained over 90 min following current injection. These results support the feasibility of pulsed protocols with low total charge in non-invasive neuromodulation when the precision of targeting is improved with a dense electrode array and accurate head modeling.

BrainK for Structural Image Processing: Creating Electrical Models of the Human Head.

BrainK is a set of automated procedures for characterizing the tissues of the human head from MRI, CT, and photogrammetry images. The tissue segmentation and cortical surface extraction support the primary goal of modeling the propagation of electrical currents through head tissues with a finite difference model (FDM) or finite element model (FEM) created from the BrainK geometries. The electrical head model is necessary for accurate source localization of dense array electroencephalographic (dEEG) measures from head surface electrodes. It is also necessary for accurate targeting of cerebral structures with transcranial current injection from those surface electrodes. BrainK must achieve five major tasks: image segmentation, registration of the MRI, CT, and sensor photogrammetry images, cortical surface reconstruction, dipole tessellation of the cortical surface, and Talairach transformation. We describe the approach to each task, and we compare the accuracies for the key tasks of tissue segmentation and cortical surface extraction in relation to existing research tools (FreeSurfer, FSL, SPM, and BrainVisa). BrainK achieves good accuracy with minimal or no user intervention, it deals well with poor quality MR images and tissue abnormalities, and it provides improved computational efficiency over existing research packages.

Transcranial Electrical Neuromodulation Based on the Reciprocity Principle.

A key challenge in multi-electrode transcranial electrical stimulation (TES) or transcranial direct current stimulation (tDCS) is to find a current injection pattern that delivers the necessary current density at a target and minimizes it in the rest of the head, which is mathematically modeled as an optimization problem. Such an optimization with the Least Squares (LS) or Linearly Constrained Minimum Variance (LCMV) algorithms is generally computationally expensive and requires multiple independent current sources. Based on the reciprocity principle in electroencephalography (EEG) and TES, it could be possible to find the optimal TES patterns quickly whenever the solution of the forward EEG problem is available for a brain region of interest. Here, we investigate the reciprocity principle as a guideline for finding optimal current injection patterns in TES that comply with safety constraints. We define four different trial cortical targets in a detailed seven-tissue finite element head model, and analyze the performance of the reciprocity family of TES methods in terms of electrode density, targeting error, focality, intensity, and directionality using the LS and LCMV solutions as the reference standards. It is found that the reciprocity algorithms show good performance comparable to the LCMV and LS solutions. Comparing the 128 and 256 electrode cases, we found that use of greater electrode density improves focality, directionality, and intensity parameters. The results show that reciprocity principle can be used to quickly determine optimal current injection patterns in TES and help to simplify TES protocols that are consistent with hardware and software availability and with safety constraints.

In vivo quantification of intraventricular hemorrhage in a neonatal piglet model using an EEG-layout based electrical impedance tomography array.

Intraventricular hemorrhage (IVH) is a common occurrence in the days immediately after premature birth. It has been correlated with outcomes such as periventricular leukomalacia (PVL), cerebral palsy and developmental delay. The causes and evolution of IVH are unclear; it has been associated with fluctuations in blood pressure, damage to the subventricular zone and seizures. At present, ultrasound is the most commonly used method for detection of IVH, but is used retrospectively. Without the presence of adequate therapies to avert IVH, the use of a continuous monitoring technique may be somewhat moot. While treatments to mitigate the damage caused by IVH are still under development, the principal benefit of a continuous monitoring technique will be in investigations into the etiology of IVH, and its associations with periventricular injury and blood pressure fluctuations. Electrical impedance tomography (EIT) is potentially of use in this context as accumulating blood displaces higher conductivity cerebrospinal fluid (CSF) in the ventricles. We devised an electrode array and EIT measurement strategy that performed well in detection of simulated ventricular blood in computer models and phantom studies. In this study we describe results of pilot in vivo experiments on neonatal piglets, and show that EIT has high sensitivity and specificity to small quantities of blood (<1 ml) introduced into the ventricle. EIT images were processed to an index representing the quantity of accumulated blood (the 'quantity index', QI). We found that QI values were linearly related to fluid quantity, and that the slope of the curve was consistent between measurements on different subjects. Linear discriminant analysis showed a false positive rate of 0%, and receiver operator characteristic analysis found area under curve values greater than 0.98 to administered volumes between 0.5, and 2.0 ml. We believe our study indicates that this method may be well suited to quantitative monitoring of IVH in newborns, simultaneously or interleaved with electroencephalograph assessments.

Optimization of focality and direction in dense electrode array transcranial direct current stimulation (tDCS).

OBJECTIVE: Transcranial direct current stimulation (tDCS) aims to alter brain function non-invasively via electrodes placed on the scalp. Conventional tDCS uses two relatively large patch electrodes to deliver electrical current to the brain region of interest (ROI). Recent studies have shown that using dense arrays containing up to 512 smaller electrodes may increase the precision of targeting ROIs. However, this creates a need for methods to determine effective and safe stimulus patterns as the number of degrees of freedom is much higher with such arrays. Several approaches to this problem have appeared in the literature. In this paper, we describe a new method for calculating optimal electrode stimulus patterns for targeted and directional modulation in dense array tDCS which differs in some important aspects with methods reported to date. APPROACH: We optimize stimulus pattern of dense arrays with fixed electrode placement to maximize the current density in a particular direction in the ROI. We impose a flexible set of safety constraints on the current power in the brain, individual electrode currents, and total injected current, to protect subject safety. The proposed optimization problem is convex and thus efficiently solved using existing optimization software to find unique and globally optimal electrode stimulus patterns. MAIN RESULTS: Solutions for four anatomical ROIs based on a realistic head model are shown as exemplary results. To illustrate the differences between our approach and previously introduced methods, we compare our method with two of the other leading methods in the literature. We also report on extensive simulations that show the effect of the values chosen for each proposed safety constraint bound on the optimized stimulus patterns. SIGNIFICANCE: The proposed optimization approach employs volume based ROIs, easily adapts to different sets of safety constraints, and takes negligible time to compute. An in-depth comparison study gives insight into the relationship between different objective criteria and optimized stimulus patterns. In addition, the analysis of the interaction between optimized stimulus patterns and safety constraint bounds suggests that more precise current localization in the ROI, with improved safety criterion, may be achieved by careful selection of the constraint bounds.

An improved artifacts removal method for high dimensional EEG.

BACKGROUND: Multiple noncephalic electrical sources superpose with brain signals in the recorded EEG. Blind source separation (BSS) methods such as independent component analysis (ICA) have been shown to separate noncephalic artifacts as unique components. However, robust and objective identification of artifact components remains a challenge in practice. In addition, with high dimensional data, ICA requires a large number of observations for stable solutions. Moreover, using signals from long recordings to provide the large observation set might violate the stationarity assumption of ICA due to signal changes over time. NEW METHOD: Instead of decomposing all channels simultaneously, subsets of channels are randomly selected and decomposed with ICA. With reduced dimensionality of the subsets, much less amount of data is required to derive stable components. To characterize each independent component, an artifact relevance index (ARI) is calculated by template matching each component with a model of the artifact. Automatic artifact identification is then implemented based on the statistical distribution of ARI of the numerous components generated. RESULTS: The proposed permutation resampling for identification matching (PRIM) method effectively removed eye blink artifacts from both simulated and real EEG. COMPARISON WITH EXISTING METHOD: The average topomap correlation coefficient between the cleaned EEG and the ground truth is 0.89±0.01 for PRIM, compared with 0.64±0.05 for conventional ICA based method. The average relative root-mean-square error is 0.40±0.01 for PRIM, compared with 0.66±0.10 for conventional method. CONCLUSIONS: The proposed method overcame limitations of conventional ICA based method and succeeded in removing eye blink artifacts automatically.