Multivariate multiscale entropy (mMSE) as a tool for understanding the resting-state EEG signal dynamics: the spatial distribution and sex/gender-related differences.

BACKGROUND: The study aimed to determine how the resting-state EEG (rsEEG) complexity changes both over time and space (channels). The complexity of rsEEG and its sex/gender differences were examined using the multivariate Multiscale Entropy (mMSE) in 95 healthy adults. Following the probability maps (Giacometti et al. in J Neurosci Methods 229:84-96, 2014), channel sets have been identified that correspond to the functional networks. For each channel set the area under curve (AUC), which represents the total complexity, MaxSlope-the maximum complexity change of the EEG signal at thefine scales (1:4 timescales), and AvgEnt-to the average entropy level at coarse-grained scales (9:12 timescales), respectively, were extracted. To check dynamic changes between the entropy level at the fine and coarse-grained scales, the difference in mMSE between the #9 and #4 timescale (DiffEnt) was also calculated. RESULTS: We found the highest AUC for the channel sets corresponding to the somatomotor (SMN), dorsolateral network (DAN) and default mode (DMN) whereas the visual network (VN), limbic (LN), and frontoparietal (FPN) network showed the lowest AUC. The largest MaxSlope were in the SMN, DMN, ventral attention network (VAN), LN and FPN, and the smallest in the VN. The SMN and DAN were characterized by the highest and the LN, FPN, and VN by the lowest AvgEnt. The most stable entropy were for the DAN and VN while the LN showed the greatest drop of entropy at the coarse scales. Women, compared to men, showed higher MaxSlope and DiffEnt but lower AvgEnt in all channel sets. CONCLUSIONS: Novel results of the present study are: (1) an identification of the mMSE features that capture entropy at the fine and coarse timescales in the channel sets corresponding to the main resting-state networks; (2) the sex/gender differences in these features.

Leveraging Multiple Distinct EEG Training Sessions for Improvement of Spectral-Based Biometric Verification Results.

Most studies on EEG-based biometry recognition report results based on signal databases, with a limited number of recorded EEG sessions using the same single EEG recording for both training and testing a proposed model. However, the EEG signal is highly vulnerable to interferences, electrode placement, and temporary conditions, which can lead to overestimated assessments of the considered methods. Our study examined how different numbers of distinct recording sessions used as training sessions would affect EEG-based verification. We analyzed the original data from 29 participants with 20 distinct recorded sessions each, as well as 23 additional impostors with only one session each. We applied raw coefficients of power spectral density estimate, and the coefficients of power spectral density estimate converted to the decibel scale, as the input to a shallow neural network. Our study showed that the variance introduced by multiple recording sessions affects sensitivity. We also showed that increasing the number of sessions above eight did not improve the results under our conditions. For 15 training sessions, the achieved accuracy was 96.7 ± 4.2%, and for eight training sessions and 12 test sessions, it was 94.9 ± 4.6%. For 15 training sessions, the rate of successful impostor attacks over all attack attempts was 3.1 ± 2.2%, but this number was not significantly different from using six recording sessions for training. Our findings indicate the need to include data from multiple recording sessions in EEG-based recognition for training, and that increasing the number of test sessions did not significantly affect the obtained results. Although the presented results are for the resting-state, they may serve as a baseline for other paradigms.

Impact of EEG Frequency Bands and Data Separation on the Performance of Person Verification Employing Neural Networks.

The paper is devoted to the study of EEG-based people verification. Analyzed solutions employed shallow artificial neural networks using spectral EEG features as input representation. We investigated the impact of the features derived from different frequency bands and their combination on verification results. Moreover, we studied the influence of a number of hidden neurons in a neural network. The datasets used in the analysis consisted of signals recorded during resting state from 29 healthy adult participants performed on different days, 20 EEG sessions for each of the participants. We presented two different scenarios of training and testing processes. In the first scenario, we used different parts of each recording session to create the training and testing datasets, and in the second one, training and testing datasets originated from different recording sessions. Among single frequency bands, the best outcomes were obtained for the beta frequency band (mean accuracy of 91 and 89% for the first and second scenarios, respectively). Adding the spectral features from more frequency bands to the beta band features improved results (95.7 and 93.1%). The findings showed that there is not enough evidence that the results are different between networks using different numbers of hidden neurons. Additionally, we included results for the attack of 23 external impostors whose recordings were not used earlier in training or testing the neural network in both scenarios. Another significant finding of our study shows worse sensitivity results in the second scenario. This outcome indicates that most of the studies presenting verification or identification results based on the first scenario (dominating in the current literature) are overestimated when it comes to practical applications.

Reciprocal inhibition and slow calcium decay in perigeniculate interneurons explain changes of spontaneous firing of thalamic cells caused by cortical inactivation.

The role of cortical feedback in the thalamocortical processing loop has been extensively investigated over the last decades. With an exception of several cases, these searches focused on the cortical feedback exerted onto thalamo-cortical relay (TC) cells of the dorsal lateral geniculate nucleus (LGN). In a previous, physiological study, we showed in the cat visual system that cessation of cortical input, despite decrease of spontaneous activity of TC cells, increased spontaneous firing of their recurrent inhibitory interneurons located in the perigeniculate nucleus (PGN). To identify mechanisms underlying such functional changes we conducted a modeling study in NEURON on several networks of point neurons with varied model parameters, such as membrane properties, synaptic weights and axonal delays. We considered six network topologies of the retino-geniculo-cortical system. All models were robust against changes of axonal delays except for the delay between the LGN feed-forward interneuron and the TC cell. The best representation of physiological results was obtained with models containing reciprocally connected PGN cells driven by the cortex and with relatively slow decay of intracellular calcium. This strongly indicates that the thalamic reticular nucleus plays an essential role in the cortical influence over thalamo-cortical relay cells while the thalamic feed-forward interneurons are not essential in this process. Further, we suggest that the dependence of the activity of PGN cells on the rate of calcium removal can be one of the key factors determining individual cell response to elimination of cortical input.

Enhancing autism spectrum disorder classification in children through the integration of traditional statistics and classical machine learning techniques in EEG analysis.

Autism Spectrum Disorder (ASD) is a neurodevelopmental disorder hallmarked by challenges in social communication, limited interests, and repetitive, stereotyped movements and behaviors. Numerous research efforts have indicated that individuals with ASD exhibit distinct brain connectivity patterns compared to control groups. However, these investigations, often constrained by small sample sizes, have led to inconsistent results, suggesting both heightened and diminished long-range connectivity within ASD populations. To bolster our analysis and enhance their reliability, we conducted a retrospective study using two different connectivity metrics and employed both traditional statistical methods and machine learning techniques. The concurrent use of statistical analysis and classical machine learning techniques advanced our understanding of model predictions derived from the spectral or connectivity attributes of a subject's EEG signal, while also verifying these predictions. Significantly, the utilization of machine learning methodologies empowered us to identify a unique subgroup of correctly classified children with ASD, defined by the analyzed EEG features. This improved approach is expected to contribute significantly to the existing body of knowledge on ASD and potentially guide personalized treatment strategies.

Decoding working memory-related information from repeated psychophysiological EEG experiments using convolutional and contrastive neural networks.

Objective.Extracting reliable information from electroencephalogram (EEG) is difficult because the low signal-to-noise ratio and significant intersubject variability seriously hinder statistical analyses. However, recent advances in explainable machine learning open a new strategy to address this problem.Approach.The current study evaluates this approach using results from the classification and decoding of electrical brain activity associated with information retention. We designed four neural network models differing in architecture, training strategies, and input representation to classify single experimental trials of a working memory task.Main results.Our best models achieved an accuracy (ACC) of 65.29 ± 0.76 and Matthews correlation coefficient of 0.288 ± 0.018, outperforming the reference model trained on the same data. The highest correlation between classification score and behavioral performance was 0.36 (p= 0.0007). Using analysis of input perturbation, we estimated the importance of EEG channels and frequency bands in the task at hand. The set of essential features identified for each network varies. We identified a subset of features common to all models that identified brain regions and frequency bands consistent with current neurophysiological knowledge of the processes critical to attention and working memory. Finally, we proposed sanity checks to examine further the robustness of each model's set of features.Significance.Our results indicate that explainable deep learning is a powerful tool for decoding information from EEG signals. It is crucial to train and analyze a range of models to identify stable and reliable features. Our results highlight the need for explainable modeling as the model with the highest ACC appeared to use residual artifactual activity.

Stronger connectivity and higher extraversion protect against stress-related deterioration of cognitive functions.

Here we attempted to define the relationship between: EEG activity, personality and coping during lockdown. We were in a unique situation since the COVID-19 outbreak interrupted our independent longitudinal study. We already collected a significant amount of data before lockdown. During lockdown, a subgroup of participants willingly continued their engagement in the study. These circumstances provided us with an opportunity to examine the relationship between personality/cognition and brain rhythms in individuals who continued their engagement during lockdown compared to control data collected well before pandemic. The testing consisted of a one-time assessment of personality dimensions and two sessions of EEG recording and deductive reasoning task. Participants were divided into groups based on the time they completed the second session: before or during the COVID-19 outbreak 'Pre-pandemic Controls' and 'Pandemics', respectively. The Pandemics were characterized by a higher extraversion and stronger connectivity, compared to Pre-pandemic Controls. Furthermore, the Pandemics improved their cognitive performance under long-term stress as compared to the Pre-Pandemic Controls matched for personality traits to the Pandemics. The Pandemics were also characterized by increased EEG connectivity during lockdown. We posit that stronger EEG connectivity and higher extraversion could act as a defense mechanism against stress-related deterioration of cognitive functions.

A hidden message: Decoding artistic intent.

Understanding how art makes impressions upon the perceiver has been a fundamental topic of philosophical interest since the time of ancient Greece. However, the extent of artistic perception and aesthetic appreciation has been the topic of empirical studies only recently, following the emergence of psychology as an independent field of science. The present study discusses the hypothesis that the impression created by artwork on the viewer can be predicted by examining activity of neuronal networks. In particular, we focus on neural activity evoked by abstract stimuli that matches elements of the viewers' previously learned conceptual dictionary. We show that artistic appreciation fundamentally depends on how easily the author's intent expressed in his or her artwork can be abstracted and decoded, on a neuronal level, into new or merged concept networks. More diverse intellectual and personal experiences-and their corollary neural networks-may facilitate the creation of new networks. These new networks, in turn, modulate the extent to which art can be apprehended and appreciated.

Resting-state EEG activity predicts frontoparietal network reconfiguration and improved attentional performance.

Mounting evidence indicates that resting-state EEG activity is related to various cognitive functions. To trace physiological underpinnings of this relationship, we investigated EEG and behavioral performance of 36 healthy adults recorded at rest and during visual attention tasks: visual search and gun shooting. All measures were repeated two months later to determine stability of the results. Correlation analyses revealed that within the range of 2-45 Hz, at rest, beta-2 band power correlated with the strength of frontoparietal connectivity and behavioral performance in both sessions. Participants with lower global beta-2 resting-state power (gB2rest) showed weaker frontoparietal connectivity and greater capacity for its modifications, as indicated by changes in phase correlations of the EEG signals. At the same time shorter reaction times and improved shooting accuracy were found, in both test and retest, in participants with low gB2rest compared to higher gB2rest values. We posit that weak frontoparietal connectivity permits flexible network reconfigurations required for improved performance in everyday tasks.

Spectral analysis versus signal complexity methods for assessing attention related activity in human EEG.

We aimed to find the most effective analytical method for assessment of attention related activity to be used in neurofeedback training. We compared commonly used spectral EEG methods with those measuring signal complexity - based on calculation of entropy and fractal dimension. The 14 subjects were examined with a modified delayed matching-to-sample task. All investigated methods revealed significant differences of EEG signals recorded in control and attentional trials, however the selection of signals with such differences varied between subjects and applied methods. The results indicated: (i) the importance of the individual analysis of signals from each subject and session, (ii) benefits of applying signal complexity methods to support spectral analysis in a further application and (iii) an advantage of the signal complexity method, carrying information of assembles of spectral components, over common spectral methods.

EEG-neurofeedback training of beta band (12-22Hz) affects alpha and beta frequencies - A controlled study of a healthy population.

The frequency-function relation of various EEG bands has inspired EEG-neurofeedback procedures intending to improve cognitive abilities in numerous clinical groups. In this study, we administered EEG-neurofeedback (EEG-NFB) to a healthy population to determine the efficacy of this procedure. We evaluated feedback manipulation in the beta band (12-22Hz), known to be involved in visual attention processing. Two groups of healthy adults were trained to either up- or down-regulate beta band activity, thus providing mutual control. Up-regulation training induced increases in beta and alpha band (8-12Hz) amplitudes during the first three sessions. Group-independent increases in the activity of both bands were observed in the later phase of training. EEG changes were not matched by measured behavioural indices of attention. Parallel changes in the two bands challenge the idea of frequency-specific EEG-NFB protocols and suggest their interdependence. Our study exposes the possibility (i) that the alpha band is more prone to manipulation, and (ii) that changes in the bands' amplitudes are independent from specified training. We therefore encourage a more comprehensive approach to EEG-neurofeedback training embracing physiological and/or operational relations among various EEG bands.

Beware: Recruitment of Muscle Activity by the EEG-Neurofeedback Trainings of High Frequencies.

EEG-neurofeedback (NFB) became a very popular method aimed at improving cognitive and behavioral performance. However, the EMG frequency spectrum overlies the higher EEG oscillations and the NFB trainings focusing on these frequencies is hindered by the problem of EMG load in the information fed back to the subjects. In such a complex signal, it is highly probable that the most controllable component will form the basis for operant conditioning. This might cause different effects in the case of various training protocols and therefore needs to be carefully assessed before designing training protocols and algorithms. In the current experiment a group of healthy adults (n = 14) was trained by professional trainers to up-regulate their beta1 (15-22 Hz) band for eight sessions. The control group (n = 18) underwent the same training regime but without rewards for increasing beta. In half of the participants trained to up-regulate beta1 band (n = 7) a systematic increase in tonic EMG activity was identified offline, implying that muscle activity became a foundation for reinforcement in the trainings. The remaining participants did not present any specific increase of the trained beta1 band amplitude. The training was perceived effective by both trainers and the trainees in all groups. These results indicate the necessity of proper control of muscle activity as a requirement for the genuine EEG-NFB training, especially in protocols that do not aim at the participants' relaxation. The specificity of the information fed back to the participants should be of highest interest to all therapists and researchers, as it might irreversibly alter the results of the training.

The Do's and Don'ts of Neurofeedback Training: A Review of the Controlled Studies Using Healthy Adults.

The goal of EEG neurofeedback (EEG-NFB) training is to induce changes in the power of targeted EEG bands to produce beneficial changes in cognitive or motor function. The effectiveness of different EEG-NFB protocols can be measured using two dependent variables: (1) changes in EEG activity and (2) behavioral changes of a targeted function (for therapeutic applications the desired changes should be long-lasting). To firmly establish a causal link between these variables and the selected protocol, similar changes should not be observed when appropriate control paradigms are used. The main objective of this review is to evaluate the evidence, reported in the scientific literature, which supports the validity of various EEG-NFB protocols. Our primary concern is to highlight the role that uncontrolled nonspecific factors can play in the results generated from EEG-NFB studies. Nonspecific factors are often ignored in EEG-NFB designs or the data are not presented, which means conclusions should be interpreted cautiously. As an outcome of this review we present a do's and don'ts list, which can be used to develop future EEG-NFB methodologies, based on the small set of experiments in which the proper control groups have excluded non-EEG-NFB related effects. We found two features which positively correlated with the expected changes in power of the trained EEG band(s): (1) protocols which focused on training a smaller number of frequency bands and (2) a bigger number of electrodes used for neurofeedback training. However, we did not find evidence in support of the positive relationship between power changes of a trained frequency band(s) and specific behavioral effects.