George K. Kostopoulos’s research while affiliated with University of Patras and other places

In this paper we present a novel approach to the person identification problem using rhythmic brain activity of spindles from whole night EEG recordings. The proposed system consists of a feature extraction module and a K-NN based classifier. Different types of features from time, frequency and wavelet domain are used to highlight the topographic, temporal, morphological, spectral and statistical discriminative information of sleep spindles. The feature set’s efficacy is exhaustively tested in order to find the most significant descriptors that maximize intra-subject separability. Extensive experiments resulted in the optimal number of sensors and features that must be used to form the subject-specific unique descriptors. The proposed system showed significant identification accuracy of 99% ~ 90% for 2–20 subjects, and not lower than 86% when identifying 28 persons, indicating that this new type of modality should be further investigated to be used in EEG based identification applications.

Understanding absence seizures demands complementary research approaches using diverse animal models, including ones induced by a variety of chemical agents. The characteristics, induction methods, and validation criteria met by several animal models are reviewed, along with a critical comparison of their advantages, limitations, and main contributions to epileptology. The distinction between typical and atypical absence seizures (TAS and AAS) in humans is considered in the respective animal analogs. Emphasis is given to the TAS models produced acutely by systemic administration of γ-hydroxybutyrate (GHB) in the rat, and penicillin in the cat (FGPE), while the validation of other TAS models is considered less complete. Among their contributions to epileptology, GHB first showed the importance of GABAB receptors and tonic inhibition on neurons and glia, while the FGPE first established the current generators and corticothalamic circuit pacing of generalized 3 Hz spike-and-wave discharges. The lifelong occurrence of absences in rats after prepubertal injection of AY-9944 is presented as a chronic model of AAS, allowing experimentation of the limbic involvement, the cognitive impairment, and the worse outcome characterizing the AAS, compared to TAS.

Real-time neurofeedback plays an increasing role in today’s clinical and basic neuroscience research. In this work, we present a real-time sleep EEG spindles detection algorithm fast enough to be used for real time acoustic feedback stimulation. We further highlight the architecture of a system that implements the algorithm and its experimental evaluation. This system can handle EEG data acquired by various means (i.e. conventional EEG systems, wireless sensors) and a response time of a few msecs has been achieved. The presented algorithm is dynamically adaptive and has accuracy similar to other well-known non real-time algorithms. Comparison and evaluation was performed using EEG data from an open database.

One of the current frontier research themes in informatics relates to the extent to which computers and machines in general can become capable of learning and teaching each other. Hopes have been raised that their education could benefit from emulating mechanisms underlying learning in animal brains. An overview of these mechanisms will be briefly presented with a focus on the recently revealed fundamental role of sleep in memory consolidation and learning., Compared to brains, computers are found very much inferior when it comes to learning. Several road signs are suggested for enriching computers’ repertory in the direction of increasing their capacity to learn by becoming more brain-like. However, the prospect of achieving such goal with state of art technology appears extremely dim.

We studied slow (≤2.5 Hz) nonevolving generalized periodic epileptiform discharges (GPEDs) in the electroencephalogram (EEG) of comatose patients after cardiac arrest (CA) in search of evidence that could assist early diagnosis of possible hypoxic nonconvulsive status epilepticus (NCSE) and its differentiation from terminal brain anoxia (BA), which can present with a similar EEG pattern. We investigated the topography of the GPEDs in the first post‐CA EEGs of 13 patients, using voltage‐mapping, and compared findings between two patients with NCSE and GPEDs > 2.5 Hz (group 1), and 11 with GPEDs ≤ 2 Hz, of whom six had possible NCSE (group 2) and five had terminal BA (group 3). Voltage mapping showed frontal maximum for the negative phase of the GPEDs in all patients of groups 1 and 2, but not in any of the patients of group 3, who invariably showed maximization of the negative phase posteriorly. Morphology, amplitude, and duration of the GPEDs varied across the groups, without distinctive features for possible NCSE. These findings provide evidence that, in hypoxic coma after CA with slow GPEDs, anterior topography of the maximum GPED negativity on voltage mapping may be a distinctive biomarker for possible NCSE contributing to the coma.

We used tomographic analysis of MEG signals to characterize regional spectral changes in the brain at sleep onset and during light sleep. We identified two key processes that may causally link to loss of consciousness during the quiet or “core” periods of NREM1. First, active inhibition in the frontal lobe leads to delta and theta spectral power increases. Second, activation suppression leads to sharp drop of spectral power in alpha and higher frequencies in posterior parietal cortex. During NREM2 core periods, the changes identified in NREM1 become more widespread, but focal increases also emerge in alpha and low sigma band power in frontal midline cortical structures, suggesting reemergence of some monitoring of internal and external environment. Just before spindles and K-complexes (KCs), the hallmarks of NREM2, we identified focal spectral power changes in pre-frontal cortex, mid cingulate, and areas involved in environmental and internal monitoring, i.e., the rostral and sub-genual anterior cingulate. During both spindles and KCs, alpha and low sigma bands increases. Spindles emerge after further active inhibition (increase in delta power) of the frontal areas responsible for environmental monitoring, while in posterior parietal cortex, power increases in low and high sigma bands. KCs are correlated with increase in alpha power in the monitoring areas. These specific regional changes suggest strong and varied vigilance changes for KCs, but vigilance suppression and sharpening of cognitive processing for spindles. This is consistent with processes designed to ensure accurate and uncorrupted memory consolidation. The changes during KCs suggest a sentinel role: evaluation of the salience of provoking events to decide whether to increase processing and possibly wake up, or to actively inhibit further processing of intruding influences. The regional spectral patterns of NREM1, NREM2, and their dynamic changes just before spindles and KCs reveal an edge effect facilitating the emergence of spindles and KCs and defining the precise loci where they might emerge. In the time domain, the spindles are seen in widespread areas of the cortex just as reported from analysis of intracranial data, consistent with the emerging consensus of a differential topography that depends on the kind of memory stored.

Objective This work investigates the spatial distribution in time of generalized ictal spikes in the typical absences of childhood absence epilepsy (CAE). Methods We studied twelve children with CAE, who had more than two typical absences during their routine video-EEG. Seizures were identified, and ictal spikes were marked over the maximum electronegative peak, clustered, waveform-averaged and spatiotemporaly analyzed in 2D electrode space. Results Consistency of spatiotemporal patterns of ictal spikes was high between the absences of the same child, but low between children. Three main discharge patterns were identified: of anterio-posterior propagation, of posterio-anterior propagation and confined to the frontal/prefrontal regions. In 4 patients, the propagation patterns transformed during the seizure into either a lateralized diminished or a non-lateralized reverse direction form. Most spikes originated fronto-temporaly, all maximized over the frontal/prefrontal electrodes and mostly decayed prefrontaly. In 4 patients, lateralized propagation patterns were identified. Conclusions Ictal spike propagation patterns suggest that epileptogenic CAE networks are personalized, interconnect distal areas in the brain - not the entire cortex - with a tendency to generate bilateral symmetrical discharges, sometimes unsuccessfully. The transformation of propagation patterns during the seizure indicates the existence of dynamic interplay within epileptogenic networks. Significance Our results support the revised concept of ictogenesis of ILAE definition in genetic (also known as idiopathic) generalized epilepsies. Understanding the focal features in CAE avoids misdiagnosis as focal epilepsy and inappropriate treatment.

In the frame of an historical section of the 2015 Featured Regional meeting of FENS in Thessaloniki, I was asked to talk on the "History of Neuroscience in Greece". Realizing the impossibility of the task, I focused on only two points: the dawn of neuroscience in Greece and the current situation which threatens the survival of neuroscience in Greece. Both points are closer to provocative questions than to evidence-based conclusions. This article is protected by copyright. All rights reserved.

During Non-Rapid Eye Movement sleep (NREM) the brain is relatively disconnected from the environment, while connectedness between brain areas is also decreased. Evidence indicates, that these dynamic connectivity changes are delivered by microstructural elements of sleep: short periods of environmental stimuli evaluation followed by sleep promoting procedures. The connectivity patterns of the latter, among other aspects of sleep microstructure, are still to be fully elucidated. We suggest here a methodology for the assessment and investigation of the connectivity patterns of EEG microstructural elements, such as sleep spindles. The methodology combines techniques in the preprocessing, estimation, error assessing and visualization of results levels in order to allow the detailed examination of the connectivity aspects (levels and directionality of information flow) over frequency and time with notable resolution, while dealing with the volume conduction and EEG reference assessment. The high temporal and frequency resolution of the methodology will allow the association between the microelements and the dynamically forming networks that characterize them, and consequently possibly reveal aspects of the EEG microstructure. The proposed methodology is initially tested on artificially generated signals for proof of concept and subsequently applied to real EEG recordings via a custom built MATLAB-based tool developed for such studies. Preliminary results from 843 fast sleep spindles recorded in whole night sleep of 5 healthy volunteers indicate a prevailing pattern of interactions between centroparietal and frontal regions. We demonstrate hereby, an opening to our knowledge attempt to estimate the scalp EEG connectivity that characterizes fast sleep spindles via an "EEG-element connectivity" methodology we propose. The application of the latter, via a computational tool we developed suggests it is able to investigate the connectivity patterns related to the occurrence of EEG microstructural elements. Network characterization of specified physiological or pathological EEG microstructural elements can potentially be of great importance in the understanding, identification, and prediction of health and disease.