SERAP AYDIN

The present study tests the hypothesis that emotions of fear and anger are associated with distinct psychophysiological and neural circuitry according to discrete emotion model due to contrasting neurotransmitter activities, despite being included in the same affective group in many studies due to similar arousal-valance scores of them in emotion models. EEG data is downloaded from OpenNeuro platform with access number of ds002721. Brain connectivity estimations are obtained by using both functional and effective connectivity estimators in analysis of short (2 sec) and long (6 sec) EEG segments across the cortex. In tests, discrete emotions and resting-states are identified by frequency band specific brain network measures and then contrasting emotional states are deep classified with 5-fold cross-validated Long Short Term Memory Networks. Logistic regression modeling has also been examined to provide robust performance criteria. Commonly, the best results are obtained by using Partial Directed Coherence in Gamma (31.5 − 60.5 Hz) sub-bands of short EEG segments. In particular, Fear and Anger have been classified with accuracy of 91.79%. Thus, our hypothesis is supported by overall results. In conclusion, Anger is found to be characterized by increased transitivity and decreased local efficiency in addition to lower modularity in Gamma-band in comparison to fear. Local efficiency refers functional brain segregation originated from the ability of the brain to exchange information locally. Transitivity refer the overall probability for the brain having adjacent neural populations interconnected, thus revealing the existence of tightly connected cortical regions. Modularity quantifies how well the brain can be partitioned into functional cortical regions. In conclusion, PDC is proposed to

The research related to brain oscillations and their connectivity is in a new take-off trend including the applications in neuropsychiatric diseases. What is the best strategy to learn about functional correlation of oscillations? In this report, we emphasize combined application of several analytical methods as power spectra, adaptive filtering of Event Related Potentials, inter-trial coherence and spatial coherence. These combined analysis procedure gives the most profound approach to understanding of EEG responses. Examples from healthy subjects, Alzheimer's Diseases, schizophrenia, and Bipolar Disorder are described.

In the present study, both single channel electroencephalography (EEG) complexity and two channel interhemispheric dependency measurements have newly been examined for classification of patients with obsessive–compulsive disorder (OCD) and controls by using support vector machine classifiers. Three embedding entropy measurements (approximate entropy, sample entropy, permutation entropy (PermEn)) are used to estimate single channel EEG complexity for 19-channel eyes closed cortical measurements. Mean coherence and mutual information are examined to measure the level of interhemispheric dependency in frequency and statistical domain, respectively for eight distinct electrode pairs placed on the scalp with respect to the international 10–20 electrode placement system. All methods are applied to short EEG segments of 2 s. The classification performance is measured 20 times with different 2-fold cross-validation data for both single channel complexity features (19 features) and interhemispheric dependency features (eight features). The highest classification accuracy of 85 ±5.2% is provided by PermEn at prefrontal regions of the brain. Even if the classification success do not provided by other methods as high as PermEn, the clear differences between patients and controls at prefrontal regions can also be obtained by using other methods except coherence. In conclusion, OCD, defined as illness of orbitofronto-striatal structures [Beucke et al., JAMA Psychiatry70 (2013) 619–629; Cavedini et al., Psychiatry Res.78 (1998) 21–28; Menzies et al., Neurosci. Biobehav. Rev.32(3) (2008) 525–549], is caused by functional abnormalities in the pre-frontal regions. Particularly, patients are characterized by lower EEG complexity at both pre-frontal regions and right fronto-temporal locations. Our results are compatible with imaging studies that define OCD as a sub group of anxiety disorders exhibited a decreased complexity (such as anorexia nervosa [Toth et al., Int. J. Psychophysiol.51(3) (2004) 253–260] and panic disorder [Bob et al., Physiol. Res.55 (2006) S113–S119]).

Four asymmetry measurements (conventional coherence function (CCF), cross wavelet correlation (CWC), phase lag index (PLI), and mean phase coherence (MPC)) have been compared to each other for the first time in order to recognize emotional states (pleasant (P), neutral (N), unpleasant (UP)) from controls in EEG sub-bands (delta (0–4 Hz), theta (4–8 Hz), alpha (8–16 Hz), beta (16–32 Hz), gamma (32–64 Hz)) mediated by affective pictures from the International Affective Picture Archiving System (IAPS). Eight emotional features, computed as hemispheric asymmetry between eight electrode pairs (Fp1 − Fp2, F7 − F8, F3 − F4, C3 − C4, T7 − T8, P7 − P8, P3 − P4, and O1 − O2), have been classified by using data mining methods. Results show that inter-hemispheric emotional functions are mostly mediated by gamma. The best classification is provided by a neural network classifier, while the best features are provided by CWC in time-scale domain due to non-stationary nature of electroencephalographic (EEG) series. The highest asymmetry levels are provided by pleasant pictures at mostly anterio-frontal (F3 − F4) and central (C3 − C4) electrode pairs in gamma. Inter-hemispheric asymmetry levels are changed by each emotional state at all lobes. In conclusion, we can state the followings: (1) Nonlinear and wavelet transform-based methods are more suitable for characterization of EEG; (2) The highest difference in hemispheric asymmetry was observed among emotional states in gamma; (3) Cortical emotional functions are not region-specific, since all lobes are effected by emotional stimuli at different levels; and (4) Pleasant stimuli can strongly mediate the brain in comparison to unpleasant and neutral stimuli.

In this study, singular spectrum analysis (SSA) has been used for the first time in order to extract emotional features from well-defined electroencephalography (EEG) frequency band activities (BAs) so-called delta (0.5–4Hz), theta (4–8Hz), alpha (8–16Hz), beta (16–32Hz), gamma (32–64Hz). These five BAs were estimated by applying sixth-level multi-resolution wavelet decomposition (MRWD) with Daubechies wavelets (db-8) to single channel nonaveraged emotional EEG oscillations of 6 s for each scalp location over 16 recording sites (Fp1, Fp2, F3, F4, F7, F8, C3, C4, P3, P4, T3, T4, T5, T6, O1, O2). Every trial was mediated by different emotional stimuli which were selected from international affective picture system (IAPS) to induce emotional states such as pleasant (P), neutral (N), and unpleasant (UP). Largest principal components (PCs) of BAs were considered as emotional features and data mining approaches were used for the first time in order to classify both three different (P, N, UP) and two contrasting (P and UP) emotional states for 30 healthy controls. Emotional features extracted from gamma BAs (GBAs) for 16 recording sites provided the high classification accuracies of 87.1% and 100% for classification of three emotional states and two contrasting emotional states, respectively. In conclusion, we found the followings: (1) Eigenspectra of high frequency BAs in EEG are highly sensitive to emotional hemispheric activations, (2) emotional states are mostly mediated by GBA, (3) pleasant pictures induce the higher cortical activation in contrast to unpleasant pictures, (4) contrasting emotions induce opposite cortical activations, (5) cognitive activities are necessary for an emotion to occur.

In estimating auditory Evoked Potentials (EPs) from ongoing EEG the number of sweeps should be reduced to decrease the experimental time and to increase the reliability of diagnosis. The first goal of this study is to demonstrate the use of basic estimation techniques in extracting auditory EPs (AEPs) from small number of sweeps relative to ensemble averaging (EA). For this purpose, three groups of basic estimation techniques are compared to the traditional EA with respect to the signal-to-noise ratio(SNR) improvements in extracting the template AEP. Group A includes the combinations of the Subspace Method (SM) with the Wiener Filtering (WF) approaches (the conventional WF and coherence weighted WF (CWWF). Group B consists of standard adaptive algorithms (Least Mean Square (LMS), Recursive Least Square (RLS), and one-step Kalman filtering (KF). The regularization techniques (the Standard Tikhonov Regularization (STR) and the Subspace Regularization (SR) methods) forms Group C. All methods are tested in simulations and pseudo-simulations which are performed with white noise and EEG measurements, respectively. The same methods are also tested with experimental AEPs. Comparisons based on the output signal-to-noise ratio (SNR) show that: 1) the KF and STR methods are the best methods among the algorithms tested in this study,2) the SM can reduce the large amount of the background EEG noise from the raw data, 3) the LMS and WF algorithms show poor performance compared to EA. The SM should be used as a pre-filter to increase their performance. 4) the CWWF works better than the WF when it is combined with the SM, 5) the STR method is better than the SR method. It is observed that, most of the basic estimation techniques show definitely better performance compared to EA in extracting the EPs.  The KF or the STR effectively reduce the experimental time (to one-fourth of that required by EA). The SM is a useful pre-filter to significantly reduce the noise on the raw data. The KF and STR are shown to be computationally inexpensive tools to extract the template AEPs and should be used instead of EA. They provide a clear template AEP for various analysis methods. To
reduce the noise level on single sweeps, the SM can be used as a pre-filter before various single sweep analysis methods. The second goal of this study is to to present a new approach to extract single sweep AEPs without using a template signal. The SM and a modified scale-space filter (MSSF) are applied consecutively. The SM is applied
to raw data to increase the SNR. The less-noisy sweeps are then individually filtered with the MSSF. This new approach is assessed in both pseudosimulations and experimental studies. The MSSF is also applied to actual auditory brainstem response (ABR) data to obtain a clear ABR from a relatively small number of sweeps. The wavelet transform coefficients (WTCs) corresponding to the signal and noise become distinguishable after the SM.  The MSSF is an effective filter in selecting the WTCs of the noise. The estimated single sweep EPs highly resemble the grand average EP although less number of sweeps are evaluated. Small amplitude variations are observed among the estimations. The MSSF applied to EA of 50 sweeps yields an
ABR that best fits to the grand average of 250 sweeps. We concluded that the combination of SM and MSSF is an efficient tool to obtain clear single sweep AEPs. The MSSF reduces the recording time to one-fifth of that required by EA in template ABR estimation. The proposed approach does not use a template signal (which is generally obtained using the average of small number of sweeps). It provides unprecedented results that support the basic assumptions in the additive signal model.

Keywords: Auditory Evoked Potential, Adaptive filtering, Tikhonov regularization,Wavelet Transform

In this study, 64-channel single trial auditory brain oscillations (STABO) have been firstly analyzed by using complexity metrics to observe the effect of musical experience on brain functions. Experimental data was recorded from eyes-opened volunteers during listening the musical chords by piano. Complexity estimation methods were compared to each other for classification of groups (professional musicians and non-musicians) by using both classifiers (support vector machine (SVM), Naive Bayes (NB)) and statistical tests (one-way ANOVA) with respect to electrode locations. Permutation entropy (PermEn) is found to be the best metric (p << 0.0001, 98.37% and 98.41% accuracies for tonal and atonal ensembles) at fronto-temporal regions which are responsible for cognitive task evaluation and perception of sound. PermEn also provides the meaningful results at the whole cortex (p << 0.0001, 99.81% accuracy for both tonal and atonal ensembles) through SVM with Radial Basis kernels superior to Gaussians. Almost the similar performance is also obtained for temporal features. Although, performance improvements are observed for spectral methods with NB, the considerable better results are obtained with SVM. The results indicate that musical stimuli cause pattern variations instead of spectral variations on STABO due to relatively higher neuronal activities around auditory cortex. In conclusion, temporal regions produce response to auditory stimuli, while frontal area integrates the auditory task at the same time. As well, the parietal cortex produces neural information according to the degree of attention generated by environmental changes such as atonal stimuli. It can be clearly stated that musical experience enhances the neural encoding performance of sound tonality at mostly fronto-temporal regions.

In the present study, new findings reveal the close association between graph theoretic global brain connectivity measures and cognitive abilities the ability to manage and regulate negative emotions in healthy adults. Functional brain connectivity measures have been estimated from both eyes-opened and eyes-closed resting-state EEG recordings in four groups including individuals who use opposite Emotion Regulation Strategies (ERS) as follow: While 20 individuals who frequently use two opposing strategies, such as rumination and cognitive distraction, are included in 1st group, 20 individuals who don't use these cognitive strategies are included in 2nd group. In 3rd and 4th groups, there are matched individuals who use both Expressive Suppression and Cognitive Reappraisal strategies together frequently and never use them, respectively. EEG measurements and psychometric scores of individuals were both downloaded from a public dataset LEMON. Since it is not sensitive to volume conduction, Directed Transfer Function has been applied to 62-channel recordings to obtain cortical connectivity estimations across the whole cortex. Regarding well defined threshold, connectivity estimations have been transformed into binary numbers for implementation of Brain Connectivity Toolbox. The groups are compared to each other through both statistical logistic regression models and deep learning models driven by frequency band specific network measures referring segregation, integration and modularity of the brain. Overall results show that high classification accuracies of 96.05% (1st vs 2nd) and 89.66% (3rd vs 4th) are obtained in analyzing full-band (0.5 − 45 Hz) EEG. In conclusion, negative strategies may upset the balance between segregation and integration. In particular, graphical results show that frequent use of rumination induces the decrease in assortativity referring network resilience. The psychometric scores are found to be highly correlated with brain network measures of global efficiency, local efficiency, clustering coefficient, transitivity and assortativity in even resting-state.


Discussion and conclusion
In the present study, healthy adults having different cognitive abilities in management of negative emotions in daily life were identified by resting-state Graph Theoretic network measures in both EO and EC states. The individuals were grouped according to their use of positive or negative cognitive/behavioral ERS. For each group, connectivity matrices were estimated by examining DTF based on Granger causality insensitive to volume conduction. BCT was used to compute the network measures from adjacency matrices, i.e. binary transformation of connectivity matrices according to non-overlapped short EEG segments across 61-channel recordings (VEOG recordings were not included in connectivity estimations). The groups were firstly classified by using LSTMNs driven by six different network measures together (CC, LE, GE, T, Q, r) with respect to both states (EO, EC) and frequency band intervals. In comparing both cognitive and behavioral opposing ERS, the highest classification performance was provided by full-band specific measures in EC state that refer the default mode network (DMN) of the brain. Eyes-opening can induce significant neural activities due to many external stimuli (Gorantla 2020). Therefore, eyes-closed resting-state can be conducive to understanding the dynamic characteristics of the brain (Liu and Wu 2020). The current results are compatible with these DMN research.
Regarding EC state, the main full-band specific findings are discussed in following items:
Frequent use of rumination is found to be characterized by high modularity due to maladaptive and repetitive negative thoughts that trigger re-experiencing negative emotions. In recent neuro-imaging studies reveal that depression causes the increase in network modularity in resting-state (, Li, BJ, Friston, K, 2018). It is well known that ruminative thoughts that result in failure to manage negative emotions lead to depression.Therefore, the present electrophysiological findings are clearly consistent with neuro-imaging results.
The frequent use expressive suppression is found to be characterized by high network resilience. Conceptually, functional network resilience has been linked with cognitive skills in both healthy (Stern 2018), and neuro-degenerative disorders (Cabeza 2018). Therefore, neuro-imaging discussions about network resilience supports the present results including the lower resilience originated from rumination and the higher resilience originated from behavioural ERs. Behavioural ERs provide the increase in network integration in comparison to cognitive ERs.
The large number of outliers were commonly observed in LE estimations in each group. These outliers might be originated from age differences among the individuals with varying ages lied between 20 and 65 because of the fact that LE was reported as incremental until adulthood in healthy subjects and then dropped with aging, while GE was found to be almost unchanged over the lifetime (Gao and Gilmore 2011).
In conclusion, Graph Theoretical global connectivity measures are found to be useful in discriminating opposing ERS in resting-state. In other words, the scores of the psychological metrics can be correlated with full-band network measures by means of segregation, integration and modularity of the brain. In particular, both segregation and integration are found to be highly sensitive to not only frequency band interval but also cognitive abilities, while the resilience represented as network assortativity is found to be almost insensitive to frequency interval. Since the brain is composed of spatially embedded complex sub-networks, there must be a balance between integration and segregation of neural information flow result in cognition and behavior in healthy brains (Bullmore and Sporns 2009). The later studies show that the number and strength of neural connections can change with aging, but the optimal balance occurred between neuronal wiring costs and communication efficiency (Bullmore and Sporns 2012; Cao 2014). Thus, the current overall findings can be concluded that cortico-functional balance is impaired by the presence of ruminative and negative thoughts. The present new findings are also compatible with the more recent neuro-imaging studies including structural connectivity analysis based on fMRI (Wang et al. 2021).
Data Availability and Information Sharing Statement
Both EEG data and behavioural test scores are openly available and are distributed along with the a data repository so called LEMON described in reference (Babayan 2019). http://fcon_1000.projects.nitrc.org/indi/retro/MPI_LEMON.html

In the present study, quantitative relations between Cognitive Emotion Regulation strategies (CERs) and EEG synchronization levels has been investigated for the first time. For this purpose, spectral coherence (COH), phase locking value and mutual information have been applied to short segments of 62-channel resting state eyes-opened EEG data collected from healthy adults who use contrasting emotion regulation strategies (frequently and rarely use of rumination&distraction, frequently and rarely use of suppression&reappraisal). In tests, the individuals are grouped depending on their self-responses to both emotion regulation questionnaire (ERQ) and cognitive ERQ. Experimental data are downloaded from publicly available database , LEMON. Regarding EEG electrode pairs that placed on right and left cortical regions, interhemispheric dependency measures are computed for non-overlapped short segments of 2 sec at 2 min duration trials. In addition to full-band EEG analysis, dependency metrics are also obtained for both alpha and beta sub-bands. The contrasting groups are discriminated from each other with respect to the corresponding features using cross-validated adaboost classifiers. High classification accuracies (CA) of 99.44% and 98.33% have been obtained through instant classification driven by full-band COH estimations. Considering regional features that provide the high CA, CERs are found to be highly relevant with associative memory functions and cognition. The new findings may indicate the close relation between neuroplasticity and cognitive skills.

Discussion:
In the present study, eyes-opened resting state surface EEG measurements were analyzed in order to investigate the possible cross-relation between inter-hemispheric neuronal coherence levels and contrasting cognitive emotion management skills. For this purpose, the features were computed by applying four different functional connectivity metrics to full-band, Alpha-band and Beta-band intervals of EEG series.

Regarding the use of spectral COH in comparison to PLV in discrimination of diversity between individuals having diverse cognitive skills or mental well-being, the application parameters as well as dta collection procedures differ from each other as listed in Table 4. The highest number of EEG recording channels are considered in the present study for classification of contrasting cognitive abilities in healthy individuals. Raw data is primarily filtered by an IIR-Notch filter and short non-overlapped EEG segments are filtered by three FIR filters in extracting full-band (0.5-40.5 Hz), alpha-band and beta-band in EO state, while BP is the mostly used pre-filter in other studies in Table 4. The largest recording interval (120h) and the largest EEG segments are examined in reference study to estimate the outcome of postanoxic coma (Gomez et al., 2021). In this study, the dependency metrics (COH, PLV, MI) are combined to determine a huge feature set, then, classifiers are trained with non-averaged estimations from segments of 5min over 120h. In both Handayani et al., 2018 and Zhang et al., 2014, the individuals are identified by the grand averaged estimations from 50% overlapped long segments (8sec and 10sec). As well, inter-hemispheric functional indicator is defined as he grand averaged dependency estimations from 75% overlapped shorter segments (4sec) in (Dell’Acqua et al., 2021). In the present study, cognitive abilities are identified by non-averaged dependency estimations from non-overlapped short segments (2sec) over 2min rather than the identification of individuals with averaged estimations from overlapped longer segments. In computing spectral COH values, most the studies commonly use WM in combination with PM (Handayani et al., 2018; Dell’Acqua et al., 2021; Zhang et al., 2014), while the others use either FT (Gomez et al., 2021; Mezeiova & Palus, 2012) or WT (Bob & Palus, 2008; Hussain et al., 2018). In the present study, BM is used to estimate spectral COH values based on the assumption that short EEG segments can be modeled by an Auto-Regressive model. So, the COH estimations provided the best results in classification of contrasting cognitive skills.

In analysis of resting-state surface EEG measurements, the following specifications can be proposed:
1) the length of EEG segment should be short as 2sec due to nerve action potential generation and propagation mechanism in addition to post-synaptic neurotransmitter transition during rest without any stimulus.
2) FIR filter should be used to extract specific frequency interval with sensitive and realistic filtering parameters
3) Short EEG segments can be assumed to be modeled by Auto-Regressive model
Averaging process cause loss of information in long EEG measurements due to time-varying post-synaptic potential variations across the cortex. Therefore, non-averaged inter-hemispheric dependency estimations should be used as indicators in detecting specific cognitive or mental states.

In the present study, the performances of two well-known linear filtering techniques are compared for extraction of auditory Evoked Potential (EP) from a relatively small number of sweeps. Both experimental and simulated data are filtered by the two algorithms into two groups. Group A consists of Wiener filtering (WF) applications, where conventional WF and Coherence Weighted WF (CWWF)) have been assessed in combination with the Subspace Method (SM). Group B consists of the well-known adaptive filtering algorithms Least Mean Square (LMS), Recursive Least Square (RLS), and one-step Kalman filtering (KF). Both groups are tested with respect to signal-to-noise ratio (SNR) enhancement by comparing to the traditional ensemble averaging (EA). We observed that KF is the best method among them. The application of the SM before filtering improves the performance of the LMS and the assessments of WF where the CWWF works better than the conventional WF in that case. In conclusion, most of the linear filters show definitely better performance compared to EA. KF effectively reduces the experimental time (to one-fourth of that required by EA). The SM that has recently been revealed in EP estimation is found to be a meaningful pre-filter as it significantly reduces the noise level of EEG raw data.

In the present study, the performances of two well-known linear filtering techniques are compared for extraction of auditory Evoked Potential (EP) from a relatively small number of sweeps. Both experimental and simulated data are filtered by the two algorithms into two groups. Group A consists of Wiener filtering (WF) applications, where conventional WF and Coherence Weighted WF (CWWF)) have been assessed in combination with the Subspace Method (SM). Group B consists of the well-known adaptive filtering algorithms Least Mean Square (LMS), Recursive Least Square (RLS), and one-step Kalman filtering (KF). Both groups are tested with respect to signal-to-noise ratio (SNR) enhancement by comparing to the traditional ensemble averaging (EA). We observed that KF is the best method among them. The application of the SM before filtering improves the performance of the LMS and the assessments of WF where the CWWF works better than the conventional WF in that case. In conclusion, most of the linear filters show definitely better performance compared to EA. KF effectively reduces the experimental time (to one-fourth of that required by EA). The SM that has recently been revealed in EP estimation is found to be a meaningful pre-filter as it significantly reduces the noise level of EEG raw data.
WOS:000257692900002

ABSTRACT
Electroencephalographic complexity and decreased randomness in drug-naive obsessive-compulsive patients
Objective: Studies investigating the complexity in electroencephalography (EEG) in various neuropsychiatric disorders have yielded abnormal results. However, few studies have examined EEG complexity in obsessive-compulsive disorder (OCD).
Methods: An eyes-closed scalp EEG series of 3 minutes was recorded in drug-naive patients with OCD and in healthy controls. Each single trial was segmented into multiple identical epochs using two windows of 10 and 30 seconds. Both Kolmogorov Complexity (KC) values and autoregressive (AR) model orders were estimated to quantify the EEG complexity for segmented EEG epochs.
Results: The EEG complexity, measured by both KC and AR model orders and in estimations using window lengths of 10 and 30 seconds, was lower in the patients than in the controls. In the AR model orders, the 10-second window differentiated the patients and controls better than the 30-second window. Conclusion: OCD is characterized by low EEG complexity, increased regularity, or decreased randomness. Segmentation of EEG signals is useful for their quantitative identification, a smaller window providing a more sensitive characterization of EEG.

Keywords: Autoregressive model order, EEG complexity, Kolmogorov complexity, obsessive compulsive disorder

In conclusion, the brains of drug-naive OCD patients are electrophysiologically less complex, more regular, and more random than the brains of controls. Investigating EEG trace in smaller window lengths may be more successful in differentiating patients and controls. These findings may contribute to the discussions of increased or decreased brain connectivity in the pathologies of the central nervous system when evaluated together with the former and future studies in this area.

The primary goal of this study is to state the clear changes in functional brain connectivity during all night sleep in psycho-physiological insomnia (PPI). The secondary goal is to investigate the usefulness of Mutual Information (MI) analysis in estimating cortical sleep EEG arousals for detection of PPI. For these purposes, healthy controls and patients were compared to each other with respect to both linear (Pearson correlation coefficient and coherence) and nonlinear quantifiers (MI) in addition to phase locking quantification for six sleep stages (stage.1-4, rem, wake) by means of interhemispheric dependency between two central sleep EEG derivations. In test, each connectivity estimation calculated for each couple of epoches (C3-A2 and C4-A1) was identified by the vector norm of estimation. Then, patients and controls were classified by using 10 different types of data mining classifiers for five error criteria such as accuracy, root mean squared error, sensitivity, specificity and precision. High performance in a classification through a measure will validate high contribution of that measure to detecting PPI. The MI was found to be the best method in detecting PPI. In particular, the patients had lower MI, higher PCC for all sleep stages. In other words, the lower sleep EEG synchronization suffering from PPI was observed. These results probably stand for the loss of neurons that then contribute to less complex dynamical processing within the neural networks in sleep disorders an the functional central brain connectivity is nonlinear during night sleep. In conclusion, the level of cortical hemispheric connectivity is strongly associated with sleep disorder. Thus, cortical communication quantified in all existence sleep stages might be a potential marker for sleep disorder induced by PPI.
Discussion and conclusion
In the present study, four hemispheric connectivity measurements were examined to obtain the electrophysiological arousals on sleep EEG epoches recorded from healthy controls and patients with PPI. All individuals can be classified correctly by using any data mining classifier for both entropy based MI estimations and spectral connectivity measurement so called coherence created by Welch’ method. When the Burg method was performed to compute the power spectral density estimation of sleep EEG epoch in estimating coherence, error free classification can be obtained by using only three classifiers (RBFNetwork, NNge, SMO). Regarding as PCC estimations, one person was misclassified in all classifiers. Concerning phase coherence estimations, one or two individuals were always misclassified.

In particular, lower interhemispheric coherence and lower MI estimations as well as higher PCC values were provided by patients in comparison to controls. In fact, only linear relations between particular hemispheric locations could be observed by using coherence, whereas the MI can measure both linear and nonlinear statistical dependencies of hemispheres in time domain. The results support that the cortex becomes more inactive as the sleep stage goes through from one stage to the next one in non REM sleep periods (stage.1–4), however, the cortex becomes much more active. It means that more neurons will be active in processing the information transmission during REM sleep in REM sleep periods. The higher order statistics of time series can be represented by nonlinear approaches, regarding as the information theory [7]. Therefore, the MI provided the most useful estimations.

The MI can give information in the context of functional connectivity such that its value highly depends on the accuracy of estimated JE derived from probability distribution. The results revealed that temporal dependency of cerebral hemispheres by means of MI can provide a very efficient tool for detection of PPI from sleep EEG recordings. The MI is a measure of statistical dependence between two random time series without making any assumption on the nature of these signals. Since, the duration of each single epoch was long enough, MI estimations gave stable estimates. Another factor making the MI be successful in detecting hemispheric functional changes between controls and PPI is that sleep EEG series are narrow band signals as stated in reference [4].

In the further study, the relationship between sleep stages and information transmission of multi-channel EEG measurements in controls will be investigated. Additionally, MI will be used to analyze sleep EEG series in detecting the effects of mood disorder depending on functional disorganization of the brain.

In the present study, a step-wise least square estimation algorithm (SLSA), implemented in a Matlab package called as ARfit, has been newly applied to clinical data for estimation of the accurate Auto-Regressive (AR) model orders of both normal and ictal EEG series where the power spectral density (PSD) estimations are provided by the Burg Method. The ARfit module is found to be usefull in comparison to a large variety of traditional methods such as Forward Prediction Error (FPE), Akaike’s Information Criteria (AIC), Minimum Description Lenght (MDL), and Criterion of Autoregressive Transfer function (CAT) for EEG discrimination. According to tests, the FPE, AIC and CAT give the identical orders for both normal and epileptic series whereas the MDL produces lower orders. Considering the resulting PSD estimations, it can be said that the most descriptive orders are provided by the SLSA. In conclusion, the SLSA can mark the seizure, since the estimated AR model orders meet the EEG complexity/regularity such that the low orders indicate an increase of EEG regularity in seizure. Then, the SLSA is proposed to select the accurate AR orders of long EEG series in diagnose for many possible future applications. The SLSA implemented by ARfit module is found to be superior to traditional methods since it is not heuristic and it is less computational complex. In addition, the more reasonable orders can be provided by the SLSA. Key Words: EEG, seizure, AR model, stepwise least square algorithm.

Discussion and conclusion:
Both cortical normal EEG measurements and intracortical epileptic EEG series, in addition to intracortical ictal records, are analyzed in the present study. Several traditional methods and the ARfit are implemented in Matlab to estimate the optimum AR model orders of these diagnostic records. Among them, the ARfit algorithm is found to be reliable and superior.
In literature, it was stated that the changes on the time series components such as oscillation periods and damping times can be characterized by MVAR models with respect to their SVD pairs [14]. Then, the ARfit module is developed to detect these changes. The current results show that the electrophysiological variations on EEG series can also be identified by using the ARfit. In other words, the meaningful sharp oscillations in EEG can be detected owing to the implementation of the ARfit module. Moreover, neither spurious peaks in the spectrum (in case of too high order), nor loss of spectral detail (in case of excessively low order) are encountered in the assessment of the ARfit.
Also, regarding as the PSD estimations, it can be said that the useful AR model orders can also be estimated by using the algorithms of FPE, AIC and CAT. Nevertheless, FPE, AIC and CAT are known to be heuristic and more subjective choices in many applications [21]. However, the ARfit is not heuristic and it is considerable less computational complex such that the optimum model can be estimated about pmax − pmin + 1 times faster than with those traditional algorithms that require pmax − pmin + 1 separate QR factorizations. The other criterion so called the MDL can not produce the adequate orders. In selecting of AR model order, the methods of FPE and AIC minimize the average error variance for a one-step and an information theoretical function, respectively [21]. The methods of FPE and AIC do not yield consistent estimates of the model order as the length of the time series increases whereas both are asymptotically equivalent [21]. The MDL criterion, also called the Bayesian information criterion, uses a penalty function which provides consistent
estimation of the model order [22].
In the ARfit module, the both the effect of rounding errors and data errors are minimized in the SLSA in association with determined approximate confidence interval [16]. The SLSA is stated as a numerically stable procedure in reference [23]. The using of the SLSA provides to obtain a more reliable residual noise variance. In fact, ARfit solves a regularized estimation problem with respect to an ill-conditioned moment matrix weighted with a regularization parameter. In summary, ARfit module is proposed as very useful, fast and efficient tool in brain activities to estimate a reliable AR model order. The results show that the estimated AR model orders can be used as markers to support the clinical findings in diagnose when ARfit is used.

In the present study, standard Tikhonov regularization (STR) Technique and the subspace regularization (SR) method have been applied to remove the additive EEG noise on average auditory-evoked potential (EP) signals. In methodological manner, the difference between these methods is the formation of regularization matrices which are used to solve the weighted problem of EP estimation. Those methods are compared to ensemble averaging (EA) with respect to signal-to-noise-ratio (SNR) improvement in experimental studies, simulations and pseudo-simulations. The results of tests no superiority of the SR in comparison to STR has been observed. In addition, the STR is found to be less computational complex. Moreover, results support the theoretical fact that the STR was introduced to be optimum for smooth solutions whereas the SR allows sharp variations in solutions. Thus, the STR is found to be more useful in removing the noise with the average signal remaining.

Conclusion: The regularization methods showed better performance compared to EA. It was observed that, the STR is marginally better than the SR in all cases. Note that the STR method is optimum for smooth solutions whereas the SR allows sharp variations in the solutions. The basis vectors are chosen from the dilated and shifted forms of a mother wavelet which resemble the waveform of the auditory EP. The linear combination of these smooth vectors models the EP. In line with the fact that a sharp variation in the coefficients of this combination is not expected, we have not observed the superiority of SR compared to the STR. In addition, the STR method has less computational complexity than the SR method. Thus, the use of the STR method is proposed instead of the SR for template auditory EP estimation. In conclusion, the STR effectively reduces the experimental time (to one-fourth of that required by EA). Both methods are closely related to Bayesian estimation but there is a distinct property between them: the SR solves a linear system where sharp variations are allowed besides; the STR provides the optimum smooth solution for the same system. Since the waveform of the EP signal is similar to a smooth wave having a fast positive peak and following a slower negative peak, the nature of the STR is more suitable in case of EP estimation. The present experimental and simulation based results support this theoretical suggestion such that the STR provides more SNR enhancement. In both simulations and pseudosimulations, the improvements were 20 dB. Besides, 5 dB improvement was obtained in experimental studies. These data dependent different achievements are originated by their autocorrelation functions which directly form the regularization matrix (L2) of interest such that there were no ripples in both pre and post-stimulus intervals in simulations in contrast to experimental data. In addition, actual background EEG noise is different from a white noise.

Inter-hemispheric sleep EEG coherence is studied in 10 subjects with psycho physiological insomnia, in 10 with paradoxical insomnia, and in 10 matched controls through different states of the sleep/wakefulness cycle. Inter hemispheric EEG coherence between central electrode pairs are compared to each other within these groups. A linear measure called as Coherence Function (CF) and a nonlinear measure called as Mutual Information (MI) are performed by using the Information Theory Toolbox in the present sleep EEG synchronization study. Regarding as tests, for all-night EEG recordings of participants, both measures indicate higher degree of EEG coherence for insomnia than for controls. The results further validate inter-hemispheric CF as a sign of activity in insomnia where the EEG series from stage2, REM sleep and the eyes closed waking state. In particular, the CF is found to be more useful tool than the MI for detection of insomnia when the power spectral density estimations of sleep stages are provided by the Burg Method. In conclusion, the CF provides insights into functional connectivity of brain regions during sleep. Since the CF has a characteristic shape for sleep states, it can be proposed to identify the degree of EEG complexity depending on sleep disorders.

Discussion and conclusion:
In the present study, insomnia is analyzed in frequency domain by using both linear (i.e., CF) and nonlinear (i.e., MI) EEG synchronization measures. Both intervals of CF and MI results give that the higher degree of EEG synchronization is observed when the brain can not asleep well. The CF is more useful tool than MI in sleep EEG analysis. Besides, it can be said that the clearest difference between ordered and disordered EEG series can be obtained via observation of frequency domain EEG synchronization for REM stages with respect to the other sleep stages.
In conclusion, the degree of EEG synchronization depends on healthy conditions in insomnia. Much lower CF values are obtained if there is no sleep disorder. Therefore, if people have a sleep disorder, it can be detected by consulting the CF curves of sleep EEG. In other words, if one has no sleep disorder, no high EEG synchronization is observed in association with any sleep stages. So, the CF can be proposed to observe the sleep EEG synchronization for detection of insomnia where the PSD estimations should be computed by using the BM. The basic idea of this proposition is that all the sleep stages are assumed to be modeled by low order AR model. As further works, other synchronization measure based on multichannel applications such as Omega Complexity as a global synchronization [14] and phase synchronization [15] will be addressed for sleep EEG analysis in insomnia in future work.

In this study, normal EEG series recorded from healthy volunteers and epileptic EEG series recorded from patients within and without seizure are classified by using Multilayer Neural Network (MLNN) architectures with respect to several time domain entropy measures such as Shannon Entropy (ShanEn), Log Energy Entropy (LogEn), and Sample Entropy (Sampen). In tests, the MLNN is performed with several numbers of neurons for both one hidden layer and two hidden layers. The results show that segments in seizure have significantly lower entropy values than normal EEG series. This result indicates an important increase of EEG regularity in epilepsy patients. The LogEn approach, which has not been experienced in EEG classification yet, provides the most reliable features into the EEG classification with very low absolute error as 0.01. In particular, the MLNN can be proposed to distinguish the seizure activity from the seizure-free epileptic series where the LogEn values are considered as signal features that characterize the degree of EEG complexity. The highest classification accuracy is obtained for one hidden layer architecture.

Discussion and Conclusion: In this study, three datasets consisting of normal and epileptic records in addition to ictal series are classified by using several MLNN architectures. The inputs of the NNs, i.e., the signal features, are computed by addressing the ShanEn, LogEn, and SamEn.

The results show that epileptic records (Set-D and Set-E) show lower entropies in comparison to healthy records (Set-A). In particular, seizure activity produces significantly lower entropies. It means that electro- physiological behavior of epileptogenic regions is less complex than behavior of healthy brain. It can be said that lower entropy indicates the severity of epilepsy. Since the LogEn values meet the most reliable features to analyze the nonlinear dynamics of both cortical and intracortical neuronal interactions, we propose the LogEn values as inputs of the MLNN in discriminat- ing the seizure.

The present study shows new findings that reveal the high association between emotional arousal and neuro-functional brain connectivity measures. For this purpose, contrasting discrete emotional states (happiness vs sadness, amusement vs disgust, calmness vs excitement, calmness vs anger, fear vs anger) are classified by using Support Vector Machines (SVMs) driven by Graph Theoretical segregation (clustering coefficients, transitivity, modularity) and integration (global efficiency, local efficiency) measures of the brain network. Emotional EEG data mediated by short duration video film clips is downloaded from publicly available database called DREAMER. Pearson Correlation (PC) and Spearman Correlation have been examined to estimate statistical dependencies between relatively shorter (6 sec) and longer (12 sec) non-overlapped EEG segments across the cortex. Then the corresponding brain connectivity encoded as a graph is transformed into binary numbers with respect to two different thresholds (60%max and mean). Statistical differences between contrasting emotions are obtained by using both one-way Anova tests and step-wise logistic regression modelling in accordance with variables (dependency estimation, segment length, threshold, network measure). Combined integration measures provided the highest classification accuracies (CAs) (75.00% 80.65%) when PC is applied to longer segments in accordance with particular threshold as the mean. The segregation measures also provided useful CAs (74.13% 80.00%), while the combination of both measures did not. The results reveal that discrete emotional states are characterized by balanced network measures even if both segregation and integration measures vary depending on arousal scores of audio-visual stimuli due to neurotransmitter release during video watching.

Discussion  and Conclusion:
In the present study, contrasting discrete emotional states have been classified with SVMs in accordance with two different kernels (RBFs vs GKs) with respect to methodological variables such as dependency methods (PC vs SC), EEG segmentation largeness (6 sec vs 12 sec), threshold definition in transforming the dependency levels into binary adjacency data (60%max and the mean). The group differences are also obtained using statistical one-way Anova tests and logistic regression modeling. The most useful results are provided by PC in larger segments in accordance with the specified threshold as the mean when RBFs are used as the kernels in SVMs. The main result of the study is to show the close correlations between emotional arousal of affective video film clips and Graph Theoretical complex network measures in terms of both segregation and integration.

Methodologically, superior performance of PC can be explained by following items: (1) PC is based on the linear relationship, while SC is based on the monotonic relationships between two variables, (2) PC can work with un-processed variables, while SC can work with rank-ordered variables. Thus, PC can be proposed for estimation of statistical dependency between EEG segments across the cortex in classifying emotions based on Graph Theoretical network analysis.

SVMs have been frequently used to classify EEG based emotional groups Torres-Valencia (2017); Doma and Pirouz (2020); Saeidi and Karwowski (2021). Since, architecture of SVMs is based on statistical learning theory that provides finding the best decision function, kernel specification affects its performance Debnath and Takahashi (2002). In the present study, RBFs are found to be superior to GKs as reported in references Bajaj and Pachori (2013); Kai (2014); Aydin (2019). In conclusion, classification performance of SVM depends on both feature space, i.e. spectral distribution of the features and kernels.

Regarding the identical dataset DREAMER, the best classification performance was obtained in analysis of shorter EEG segments (6 sec) identified by single-channel phase domain local complexity estimations in Aydin (2020), however, the recent useful results are provided in larger EEG segments (12 sec) represented by global brain network indices across the whole cortex. In Table 4, large variety of window length is given in emotion research based on EEG analysis: Due to decisive differences in both emotional stimulus types (visual, auditory, audio-visual) and experimental paradigms (inter-stimulus-inter duration, stimulus display/presentation time, recording equipment (number of recording electrodes)) as well as state definition (pleasantness in accordance with valance scores, negativity in accordance with arousal scores, discrete emotional state), the proposed segmentation length differs from each other. Moreover, emotional features have been extracted from EEG measurements by examining several methods depending on the goal such as observing neural activities at EEG recording placements (local analysis), quantifying inter-hemispheric neural communications (regional EEG analysis) and understanding the brain network mechanism across the whole cortex (global EEG analysis).

While defining this state can be based on the valence scores of the stimuli (pleasant-unpleasant), researchers have also combine the emotional states, placed in the identical quarter of circumplex emotion model (Fig. 4), in a single category (negative-positive). However, each discrete emotional states have been considered as a different state in accordance with discrete emotion model (Fig. 4) and then each discrete emotional state is identified by EEG based Graph Theoretical network measures in the present study, since both nerve action potentials superimposed by post-synaptic potentials including excitatory and inhibitory neurotransmitter activities are embedded in EEG series. Apart from brain-computer interfaces, it is crucial to assign each emotion as a separate discrete state in accordance with discrete emotion model (see Fig. 4), even if they have similar arousal-valance scores, in not only understanding the functional brain mechanism but also recognizing particular neuropsychiatric diseases characterized by perceptional deficit in computational and behavioural neuroscience. Several emotions (such as fear and anger) are considered as members of a single group in accordance with the identical quarter of arousal-valance dimensions in accordance with circumplex emotion model (see Fig. 4), although neurotransmitter activities embedded in EEG series are quite different in every emotional state.

Therefore, emotions are categorized into basic emotions and their derivatives mentioned as mixed emotions. It’s known that neurotransmitters have a great impact on emotion forming, behaviour and psychiatric disorders Ruhé et al. (2007); Liu et al. (2018); Wang et al. (2020). Three neurotransmitters of serotonin, dopamine and norephinephrine are presented as the most important neurotransmitters in psychopharmacology Schatzberg and Nemeroff (2017). The brief role of them is to change (increase/decrease) post-synaptic potentials of pyramidal nerve cells in the brain. Once temporal and spatial summation of synchronized post-synaptic potentials exceeds the threshold level, nerve action potential is generated and then propagated along with secondary neurons. Both generation and propagation of action potentials provide neural information flow across the cortex. Therefore, external stimulus type (auditory, acoustic, visual, audio-visual, somatosensorial, etc.), duration (time-locked, short duration, long duration), intensity (low-moderate-high) and content (emotional/affective, attentional, working memory, recalling memory, etc.) are all sources in releasing the specified neurotransmitters. Therefore, time-varying post-synaptic potentials including both excitatory and inhibitory actions driven by particular neurotransmitters are embedded in EEG segments. The current findings reveal the close association between emotional arousal score of external video stimuli and functional brain connectivity mechanism due to varying neurotransmitter release at pyramidal nerves

In the present study, a novel emotional complexity marker is proposed for classification of discrete emotions induced by affective video film clips. Principal Component Analysis (PCA) is applied to full-band specific phase space trajectory matrix (PSTM) extracted from short emotional EEG segment of 6 s, then the first principal component is used to measure the level of local neuronal complexity. As well, Phase Locking Value (PLV) between right and left hemispheres is estimated for in order to observe the superiority of local neuronal complexity estimation to regional neuro-cortical connectivity measurements in clustering nine discrete emotions (fear, anger, happiness, sadness, amusement, surprise, excitement, calmness, disgust) by using Long-Short-Term-Memory Networks as deep learning applications. In tests, two groups (healthy females and males aged between 22 and 33 years old) are classified with the accuracy levels of 68.52% and 79.36% through the proposed emotional complexity markers and and connectivity levels in terms of PLV in amusement. The groups are found to be statistically different (p ≪ 0.5) in amusement with respect to both metrics, even if gender difference does not lead to different neuro-cortical functions in any of the other discrete emotional states. The high deep learning classification accuracy of 98.00% is commonly obtained for discrimination of positive emotions from negative emotions through the proposed new complexity markers. Besides, considerable useful classification performance is obtained in discriminating mixed emotions from each other through full-band connectivity features. The results reveal that emotion formation is mostly influenced by individual experiences rather than gender. In detail, local neuronal complexity is mostly sensitive to the affective valance rating, while regional neurocortical connectivity levels are mostly sensitive to the affective arousal ratings.

Conclusion: In the present study, a new emotional recognition methodology has been presented. The close relationship between affective stimulus parameters and neuro-cortical activities in young females and males in nine discrete emotional states. For this purpose, PCA is applied to PSTM of short EEG segments.

The primary concept was to observe the gender effect on emotional neuro-complexity levels, and then, the second concept was to observe the usefulness of the proposed complexity markers for emotion recognition. In all applications, EEG measurements were segmented into short epochs of 6 s and 12 s in order to investigate the influence of analysis interval for emotion recognition. Conventional and deep learning networks were trained in not only instant classification but also subject classification manners for both segmentation steps. Regarding those main concepts, the results were compatible with to each other: Females were differed from males in E4 (amusement) through CL-4 for both shorter and larger segments. Thus, CL-4 provided the relatively lower performance for discrimination of E4 from baseline in comparison to recognition of the other emotional states for both segment statements. For recognition of E4 (amusement) and E8 (calmness), classification performances were increased when the subjects were classified through CL-4 instead of instant classification. When the analysis interval was largened from 6 s to 12 s, emotion recognition performances were decreased through all classifiers.

Considering emotional EEG complexity levels mediated by audio-visual affective video films, gender differences became slightly un-avoidable when subjects were classified instead of instant classification. Similarly, each emotional states were differed from baseline with very high CA levels when subjects were classified instead of instant classification. Among conventional and deep learning networks, the most useful classifier was CL-4, i.e., CNN. Although, SVM classifiers provided slightly better performance when the number of features was low, deep learning algorithms, CNNs and LSTMNs were found to be better when the large number of features were examined.

In conclusion, mixed emotions highly modulate the functional connectivity of the amygdala with the other regions of the brain. In particular, regional PCPSTM estimations characterize the dynamic signature of emotion formation depending on individual experiences driven by ongoing perception and cognition processes. From EEG signal processing point of view, primary extraction of PSTM reduces the background EEG, i.e., increases the signal-to-noise ratio. Then, application of PCA on PSTM highlights the main harmonics in association with audio-visual evoked potentials embedded in short epochs. Due to ongoing combination of excitatory and inhibitory post-synaptic potentials as well as Action Potentials (APs), EEG series are time-varying psychophysiological signals. In particular, time-varying audio-visual affective stimuli continuously cause both generation and propagation of nerve APs at auditory, visual and cognitive cortices simultaneously. Therefore, EEG segmentation and analysis of short non-overlapped epochs are crucial pre-processes for emotion recognition. As well, the most useful length of short epoch is found to be 6 s due to AP phases lasting about 2 s.

In recent neuroscience studies, it is highlighted that emotional responsiveness of individuals can be clinical support in not only rare disease so called pre-symptomatic Huntington’s disease [101] but also several important and widespread psychiatric conditions such as unipolar depression [102], Parkinson’s disease with lack of dementia [103] and autism spectrum [104] as well as amygdalar lesions [105]. Full-band PSTMCs can be proposed as single-channel emotion recognition system.

https://ieeexplore.ieee.org/document/8933102

Objective: Studies investigating the complexity in electroencephalography (EEG) in various neuropsychiatric disorders have yielded abnormal results. However, few studies have examined EEG complexity in obsessive-compulsive disorder (OCD). Methods: An eyes-closed scalp EEG series of 3 minutes was recorded in drug-naive patients with OCD and in healthy controls. Each single trial was segmented into multiple identical epochs using two windows of 10 and 30 seconds. Both Kolmogorov complexity (KC) values and auto regressive (AR) model orders were estimated to quantify the EEG complexity for segmented EEG epochs. Results: The EEG complexity, measured by both KC and AR model orders and in estimations using window lengths of 10 and 30 seconds, was lower in the patients than in the controls. In the AR model orders, the 10 second-window differentiated the patients and controls better than the 30 second-window did. Conclusion: OCD is characterized by low EEG complexity, increased regularity or decreased randomness. Segmentation of EEG signals is useful for their quantitative identification, a smaller window providing a more sensitive characterization of EEG.

Inter-hemispheric sleep EEG coherence is studied in 10 subjects with psycho physiological insomnia, in 10 with paradoxical insomnia, and in 10 matched controls through different states of the sleep/wakefulness cycle. Inter hemispheric EEG coherence between central electrode pairs are compared to each other within these groups. A linear measure called as Coherence Function (CF) and a nonlinear measure called as Mutual Information (MI) are performed by using the Information Theory Toolbox in the present sleep EEG synchronization study. Regarding as tests, for all-night EEG recordings of participants, both measures indicate higher degree of EEG coherence for insomnia than for controls. The results further validate inter-hemispheric CF as a sign of activity in insomnia where the EEG series from stage2, REM sleep and the eyes closed waking state. In particular, the CF is found to be more useful tool than the MI for detection of insomnia when the power spectral density estimations of sleep stages are provided by the Burg Method. In conclusion, the CF provides insights into functional connectivity of brain regions during sleep. Since the CF has a characteristic shape for sleep states, it can be proposed to identify the degree of EEG complexity depending on sleep disorders.

In the present study, the Singular Spectrum Analysis (SSA) is applied to sleep EEG segments collected from healthy volunteers and patients diagnosed by either psycho physiological insomnia or paradoxical insomnia. Then, the resulting singular spectra computed for both C3 and C4 recordings are assigned as the features to the Artificial Neural Network (ANN) architectures for EEG classification in diagnose. In tests, singular spectrum of particular sleep stages such as awake, REM, stage1 and stage2, are considered. Three clinical groups are successfully classified by using one hidden layer ANN architecture with respect to their singular spectra. The results show that the SSA can be applied to sleep EEG series to support the clinical findings in insomnia if ten trials are available for the specific sleep stages. In conclusion, the SSA can detect the oscillatory variations on sleep EEG. Therefore, different sleep stages meet different singular spectra. In addition, different healthy conditions generate different singular spectra for each sleep stage. In summary, the SSA can be proposed for EEG discrimination to support the clinical findings for psycho-psychological disorders.

In the present study, both linear and nonlinear EEG synchronization methods so called Coherence Function (CF) and Mutual Information (MI) are performed to obtain high quality signal features in discriminating the Central Sleep Apnea (CSA) and Obstructive Sleep Apnea (OSA) from controls. For this purpose, sleep EEG series recorded from patients and healthy volunteers are classified by using several Feed Forward Neural Network (FFNN) architectures with respect to synchronic activities between C3 and C4 recordings. Among the sleep stages, stage2 is considered in tests. The NN approaches are trained with several numbers of neurons and hidden layers. The results show that the degree of central EEG synchronization during night sleep is closely related to sleep disorders like CSA and OSA. The MI and CF give us cooperatively meaningful information to support clinical findings. Those three groups determined with an expert physician can be classified by addressing two hidden layers with very low absolute error where the average area of CF curves ranged form 0 to 10 Hz and the average MI values are assigned as two features. In a future work, these two features can be combined to create an integrated single feature for error free apnea classification.

Discussion and conclusion:
The MI and CF have recently been applied to sleep EEG series in discriminating apnea disorders so called CSA and OSA and apnea free healthy controls. To show the usefulness of EEG synchronization concept, the FFNN architectures have been successfully performed to classify these three groups where the average MI values and the average area of CF curves are assigned as signal features. The results support that, the degree of EEG synchronization between the central regions of right and left hemispheres are depend on brain dysfunctions closely related to apnea disorders. In conclusion, the CF and MI can cooperatively characterise apnea disorders. We can also conclude that these two synchronization measures can detect the difference of two synchronic sleep EEG series. In summary, we propose both linear and nonlinear EEG synchronization quantities for sleep EEG analysis in a large variety of sleep disorders from insomnia to apnea.

In a future work, we will combine the average MI values and the average area of CF curves to obtain a single feature for classification of apnea with error free. Additionally, Global Field Synchronisation [20] and Omega Complexity [21] can also be performed cooperatively to characterise the functional sleep quality and degree of EEG complexity by means of EEG synchronisation in a future work.

In this study, normal EEG series recorded from healthy volunteers and epileptic EEG series recorded from patients within and without seizure are classified by using Multilayer Neural Network (MLNN) architectures with respect to several time domain entropy measures such as Shannon Entropy (ShanEn), Log Energy Entropy (LogEn), and Sample Entropy (Sampen). In tests, the MLNN is performed with several numbers of neurons for both one hidden layer and two hidden layers. The results show that segments in seizure have significantly lower entropy values than normal EEG series. This result indicates an important increase of EEG regularity in epilepsy patients. The LogEn approach, which has not been experienced in EEG classification yet, provides the most reliable features into the EEG classification with very low absolute error as 0.01. In particular, the MLNN can be proposed to distinguish the seizure activity from the seizure-free epileptic series where the LogEn values are considered as signal features that characterize the degree of EEG complexity. The highest classification accuracy is obtained for one hidden layer architecture.

Four asymmetry measurements (conventional coherence function (CCF), cross wavelet correlation (CWC), phase lag index (PLI), and mean phase coherence (MPC)) have been compared to each other for the first time in order to recognize emotional states (pleasant (P), neutral (N), unpleasant (UP)) from controls in EEG sub-bands (delta (0-4 Hz), theta (4-8 Hz), alpha (8-16 Hz), beta (16-32 Hz), gamma (32-64 Hz)) mediated by affective pictures from the International Affective Picture Archiving System (IAPS). Eight emotional features, computed as hemispheric asymmetry between eight electrode pairs (Fp1 - Fp2, F7 - F8, F3 - F4, C3 - C4, T7 - T8, P7 - P8, P3 - P4, and O1 - O2), have been classified by using data mining methods. Results show that inter-hemispheric emotional functions are mostly mediated by gamma. The best classification is provided by a neural network classifier, while the best features are provided by CWC in time-scale domain due to non-stationary nature of electroencephalographic (EEG) series. The highest asymmetry levels are provided by pleasant pictures at mostly anterio-frontal (F3 - F4) and central (C3 - C4) electrode pairs in gamma. Inter-hemispheric asymmetry levels are changed by each emotional state at all lobes. In conclusion, we can state the followings: (1) Nonlinear and wavelet transform-based methods are more suitable for characterization of EEG; (2) The highest difference in hemispheric asymmetry was observed among emotional states in gamma; (3) Cortical emotional functions are not region-specific, since all lobes are effected by emotional stimuli at different levels; and (4) Pleasant stimuli can strongly mediate the brain in comparison to unpleasant and neutral stimuli.

In conclusion, inter-hemispheric neuronal functions, generated by long-duration affective stimuli, can be analyzed by examining CWC in gamma. The results support that there is a strong relationship between gamma-specific inter-neuronal functions and emotional states at almost each brain region, while anterio-frontal and central regions show relatively higher hemispheric asymmetry levels in gamma.

In the present study, the level of nonlinear interhemispheric synchronization has been estimated by using wavelet correlation (WC) method for detection of emotional dysfunctions. Due to non-stationary nature of EEG series in addition to the assumption that the high-frequency band is possibly associated with emotional activation, WC has been applied to five distinct frequency band activities (fba) (Delta: 0:5-4 Hz, Theta: 4-8 Hz, Alpha: 8-16 Hz, Beta: 16-32 Hz, Gamma: 32-64 Hz) embedded in nonaveraged single-trial EEG series mediated by convenient affective pictures from International Affective Picture System. Experimental data were collected from both healthy controls and patients, diagnosed with first-episode psychosis, through a 16-channel EEG cap. WC estimations, which are computed for eight electrode pairs (pre-frontal, anterio-frontal, central, parietal, occipital, posterio-frontal, anterio-temporal, posterio-temporal), in accordance with each specified fba and emotional state (pleasant, unpleasant, neutral) have been classified by using Least Squares Support Vector Machines with tenfold cross-validation to distinguish controls from patients. Results show that the highest classification accuracies of 88.06, 86.39, 83.89% are obtained in Gamma with respect to neutral, unpleasant, and pleasant stimuli, respectively. In each group (controls and patients), the largest WCs are observed at anterio-frontal and central lobes; however, controls generate the high WC in response to pleasant stimuli, whereas the patients generate the high WC in response to neutral stimuli in Gamma. In conclusion, fronto-central lobes are the most activated brain regions during emotional stimulation by means of inter-hemispheric correlation. Gamma is the most sensitive fba to visual affective pictures. Emotional dysfunctions are found to be characterized by decreased WC in pleasant state, increased WC in neutral state in Gamma.

In the present study, the performances of two well-known linear filtering techniques are compared for extraction of auditory Evoked Potential (EP) from a relatively small number of sweeps. Both experimental and simulated data are filtered by the two algorithms into two groups. Group A consists of Wiener filtering (WF) applications, where conventional WF and Coherence Weighted WF (CWWF)) have been assessed in combination with the Subspace Method (SM). Group B consists of the well-known adaptive filtering algorithms Least Mean Square (LMS), Recursive Least Square (RLS), and one-step Kalman filtering (KF). Both groups are tested with respect to signal-to-noise ratio (SNR) enhancement by comparing to the traditional ensemble averaging (EA). We observed that KF is the best method among them. The application of the SM before filtering improves the performance of the LMS and the assessments of WF where the CWWF works better than the conventional WF in that case.

In the present study, standard Tikhonov regularization (STR) Technique and the subspace regularization (SR) method have been applied to remove the additive EEG noise on average auditory-evoked potential (EP) signals. In methodological manner, the difference between these methods is the formation of regularization matrices which are used to solve the weighted problem of EP estimation. Those methods are compared to ensemble averaging (EA) with respect to signal-to-noise-ratio (SNR) improvement in experimental studies, simulations and pseudo-simulations. The results of tests no superiority of the SR in comparison to STR has been observed. In addition, the STR is found to be less computational complex. Moreover, results support the theoretical fact that the STR was introduced to be optimum for smooth solutions whereas the SR allows sharp variations in solutions. Thus, the STR is found to be more useful in removing the noise with the average signal remaining.

In the present study, well known scale-space filtering (SSF) algorithm is used in combination with a linear mapping approach (LMA) to obtain clear auditory evoked potential (EP) waveform. The proposed combination involves two sequential steps: At first, the EEG noise level is reduced from-5 to 0 dB owing to the LMA based on the singular-value-decomposition. In the secondary process, the EEG noise remaining on the projected data is removed by using the SSF. A small number of sweeps are composed as a raw matrix to project the data without using the ensemble averaging at the beginning of the proposed method. Then, single sweeps are individually filtered in wavelet domain by using the SSF in the secondary step. The experimental results show that the SSF can extract the clear single-sweep auditory EP waveform where the LMA is used as a primary filtering. As well, the results indicate that the EP signal and background EEG noise create different wavelet coefficients due to their different characteristics. However, this characteristic difference can be considered to distinguish the EP signal and the EEG noise when the Signal-to-Noise-Ratio is higher than 0 dB.

In the present study, a step-wise least square estimation algorithm (SLSA), implemented in a Matlab package called as ARfit, has been newly applied to clinical data for estimation of the accurate Auto-Regressive (AR) model orders of both normal and ictal EEG series where the power spectral density (PSD) estimations are provided by the Burg Method. The ARfit module is found to be usefull in comparison to a large variety of traditional methods such as Forward Prediction Error (FPE), Akaike's Information Criteria (AIC), Minimum Description Lenght (MDL), and Criterion of Autoregressive Transfer function (CAT) for EEG discrimination. According to tests, the FPE, AIC and CAT give the identical orders for both normal and epileptic series whereas the MDL produces lower orders. Considering the resulting PSD estimations, it can be said that the most descriptive orders are provided by the SLSA. In conclusion, the SLSA can mark the seizure, since the estimated AR model orders meet the EEG complexity/regularity such that the low orders indicate an increase of EEG regularity in seizure. Then, the SLSA is proposed to select the accurate AR orders of long EEG series in diagnose for many possible future applications. The SLSA implemented by ARfit module is found to be superior to traditional methods since it is not heuristic and it is less computational complex. In addition, the more reasonable orders can be provided by the SLSA.

Abstract

The primary goal of this study is to state the clear changes in functional brain connectivity during all night sleep in psycho-physiological insomnia (PPI). The secondary goal is to investigate the usefulness of Mutual Information (MI) analysis in estimating cortical sleep EEG arousals for detection of PPI. For these purposes, healthy controls and patients were compared to each other with respect to both linear (Pearson correlation coefficient and coherence) and nonlinear quantifiers (MI) in addition to phase locking quantification for six sleep stages (stage.1-4, rem, wake) by means of interhemispheric dependency between two central sleep EEG derivations. In test, each connectivity estimation calculated for each couple of epoches (C3-A2 and C4-A1) was identified by the vector norm of estimation. Then, patients and controls were classified by using 10 different types of data mining classifiers for five error criteria such as accuracy, root mean squared error, sensitivity, specificity and precision. High performance in a classification through a measure will validate high contribution of that measure to detecting PPI. The MI was found to be the best method in detecting PPI. In particular, the patients had lower MI, higher PCC for all sleep stages. In other words, the lower sleep EEG synchronization suffering from PPI was observed. These results probably stand for the loss of neurons that then contribute to less complex dynamical processing within the neural networks in sleep disorders an the functional central brain connectivity is nonlinear during night sleep. In conclusion, the level of cortical hemispheric connectivity is strongly associated with sleep disorder. Thus, cortical communication quantified in all existence sleep stages might be a potential marker for sleep disorder induced by PPI.
Discussion and conclusion
In the present study, four hemispheric connectivity measurements were examined to obtain the electrophysiological arousals on sleep EEG epoches recorded from healthy controls and patients with PPI. All individuals can be classified correctly by using any data mining classifier for both entropy based MI estimations and spectral connectivity measurement so called coherence created by Welch’ method. When the Burg method was performed to compute the power spectral density estimation of sleep EEG epoch in estimating coherence, error free classification can be obtained by using only three classifiers (RBFNetwork, NNge, SMO). Regarding as PCC estimations, one person was misclassified in all classifiers. Concerning phase coherence estimations, one or two individuals were always misclassified.

In particular, lower interhemispheric coherence and lower MI estimations as well as higher PCC values were provided by patients in comparison to controls. In fact, only linear relations between particular hemispheric locations could be observed by using coherence, whereas the MI can measure both linear and nonlinear statistical dependencies of hemispheres in time domain. The results support that the cortex becomes more inactive as the sleep stage goes through from one stage to the next one in non REM sleep periods (stage.1–4), however, the cortex becomes much more active. It means that more neurons will be active in processing the information transmission during REM sleep in REM sleep periods. The higher order statistics of time series can be represented by nonlinear approaches, regarding as the information theory [7]. Therefore, the MI provided the most useful estimations.

The MI can give information in the context of functional connectivity such that its value highly depends on the accuracy of estimated JE derived from probability distribution. The results revealed that temporal dependency of cerebral hemispheres by means of MI can provide a very efficient tool for detection of PPI from sleep EEG recordings. The MI is a measure of statistical dependence between two random time series without making any assumption on the nature of these signals. Since, the duration of each single epoch was long enough, MI estimations gave stable estimates. Another factor making the MI be successful in detecting hemispheric functional changes between controls and PPI is that sleep EEG series are narrow band signals as stated in reference [4].

In the further study, the relationship between sleep stages and information transmission of multi-channel EEG measurements in controls will be investigated. Additionally, MI will be used to analyze sleep EEG series in detecting the effects of mood disorder depending on functional disorganization of the brain.

The present study compares two Auto-Regressive (AR) model based (Burg Method (BM) and Yule Walker Method) and two subspace based (Eigen Method and Multiple Signal Classification Method) power spectral density predictors in computing the Coherence Function (CF) to observe EEG synchronization between right and left hemispheres. For this purpose, two channels intracortical EEG series recorded from WAG/Rij rats (a genetic model for human absence epilepsy) are analyzed. In tests, AR modelbased predictors result the close performance such that the CF estimations are sensitive to the AR model order. Dealing with the subspace-based predictors; certain peaks in CF estimations can also be detected in case of low noise subspace dimension. Besides, they are more computational complexity. In conclusion, high order BM is proposed in EEG synchronization. The results support that each EEG sequence probably meets a high order AR model where the dimension of the related noise subspace is relatively low in comparison to the model order.

Conclusions:  Overall results show that Set A consists of pre-lesion EEG series, whereas the others are related to epileptic lesions: There are more lesions in both Set D and Set E in comparison to Set B and Set C. All data sets show sleep-wave like oscillations.

Parametric PSD predictors (BM, YWM) are based on the AR model. The data is not windowed in performing these methods. Therefore, application of them eliminates the assumption that the autocorrelation sequence is zero outside the window. The BM and YWM are applied to short time EEG series at low SNR. It is stated that these predictors exhibit spectral line splitting in case of high SNR and sensitive to the initial phase in case of sinusoidal noisy signals.10 In the present application, the high order (p ≥ 20) parametric predictors provide us to useful CF estimations in the present EE synchronization where the number of FFT is 128.

Subspace-based PSD predictors (EM, MM) are based on an eigen-decomposition of the correlation matrix of the noisy data. These methods assume that the data consists of some real sinusoids and then the frequencies of sinusoidal components are estimated. For the data at low SNR, it is difficult to determine the number of principal eigenvectors.10 In the present case study, subspace-based predictors can also provide useful estimations when the empirically selected subspace dimension is low (K = 10) such that the EM is less sensitive to this parameter. Therefore, the low dimensional EM can also be proposed in EEG synchronization for short time EEG series at low SNR. However, adequate estimations are obtained when the number of FFT is 256 at least.

In conclusion, it can be said that high order AR model-based predictors are more suitable than the subspace-based predictors. The reasons of this superiority can be summarized as: When the sampling frequency is low, short epochs are advised in EEG analysis as best guarantee for wide sense stationary. Therefore, a stable AR model can be yielded by the parametric estimations. High frequency resolution can be obtained by using both BM and YWM without consuming a large memory.

Note that, the recent results support that the short time intracortical EEG series can be represented by a high order (p ≥ 20) AR model as indicated in past literature.2,6 In the former EEG study, p = 10 was provided for successful AR modeling with high accuracy. It was suggested in the latter study that p = 30 is the best order to verify normality in EEG series collected during anesthesia.

Objective: Some studies on schizophrenia showed an increased complexity in electroencephalography (EEG) whereas others detected a decreased complexity. Because this discrepancy might be due to the clinical features or complexity measures used, we employed two different complexity measures in a group of schizophrenics similar in illness duration (chronic) and symptom profile (residual). Methods: Right-handed chronic residual schizophrenic patients (10 male, 10 female) and age- and sex-matched 20 healthy controls were included in the study. Eyes-closed resting EEG series were measured through quantitative EEG band activities, the log energy entropy (LEE) values, and the Hurst exponents (HE) of EEG measurements were computed for each electrode site. Results: Significantly higher LEE values in the prefrontal, frontal, temporal and parietal locations were observed in schizophrenic patients compared with controls. HE values were significantly higher on the right frontal area in the schizophrenics. Patient group showed increased prefrontal, frontal and parietal delta activity, prefrontal, left temporal and right parietal theta activity and increased left temporal alpha activity. Discussion: In the present study, we found that chronic residual schizophrenia is associated with decreased complexity and increased smoothness in EEG. In addition, EEG of patients was characterized by obvious slowness at prefrontal and frontotemporal regions, dominantly. An integration of EEG complexity and frequency analysis can be proposed as an innovative tool in schizophrenia research.

In this study, cognitive and behavioral emotion regulation strategies (ERS) are classified by using machine learning models driven by a new local EEG complexity approach so called Frequency Specific Complexity (FSC) in restingstates (eyes-opened (EO), eyes-closed (EC)). According to international 10-20 electrode placement system, FSC is defined as entropy estimations in Alpha (8−12 Hz) and Beta (12.5−30 Hz) frequency band intervals of non-overlapped short EEG segments to observe local EEG complexity variations at 62 points on scalp surface. The healthy adults who use both rumination and cognitive distraction frequently are included in the 1st groups, while the others who use these strategies rarely are included in the 2nd group with respect to Cognitive Emotion Regulation Questionnaire (CERQ) scores of them. EEG data and CERQ scores are downloaded from publicly available database LEMON. In order to test the reliability of the proposed method, five different supervised machine learning methods in addition to two Extreme Learning Machine models are examined with 5-fold cross-validation for discrimination of the contrasting groups. The highest classification accuracy (CA) of 99.47% is provided by Class-specific Cost Regulation Extreme Learning Machines in EC state. Regarding cortical regions (anterio-frontal, central, temporal, parieto-occipital), the regional FSC estimations did not provide the higher performance, however, corresponding statistical distribution shows the decrease in EEG complexity at mostly anterior cortex in the 1st group characterized by maladaptive rumination. In conclusion, FSC can be proposed to investigate cognitive dysfunctions often caused by the use of rumination.

Highlights:
·        In the present study, a new quantitative single-channel EEG marker called as Frequency Specific Complexity for classiffication of maladaptive rumination at resting-state.
·        The reliability of the proposed method has been provided by using seven different 5-fold cross-validated classiffiers with respect to both states (eyes-opened vs eyes-closed) and cortical regions.
·        The new findings show that maladaptive rumination cause decrease neuronal complexity at mostly anterior regions.

The goal of the present study is to propose the use of global connectivity measures as quantitative indicators of long-term medication in pediatric patients with Attention-Deficit-Hyperactivity Disorder, combined type (ADHD-C). For this purpose, graph theoretical brain connectivity indices are computed from connectivity estimations across eyes-opened resting-state EEG recordings measured before and after the treatment with osmotic release oral system-methylphenidate for a month in 18 boys (aged between 7–12 years). In order to present the reliable results, neurofunctional correlations are firstly estimated in time (Pearson Correlation (PC), Spearman Correlation), frequency (Directed Transfer Function, Partial Directed Coherence) and phase (Phase Locking Value, Phase Lag Index) domains in between short segments of
2 sec over single trials of 1 min. Later, transitivity, clustering coefficients, assortativity, global efficiency and modularity are computed from EEG based connectivity matrices produced by each approach. Since the highest classification accuracy of 83.79%
is provided by PC, statistical tests (one-way Anova, pair-wise multiple comparison) and step-wise logistic regression modelling are all examined to detect significant differences between pre- and post- treatment relevant connectivity measures. Statistical boxplots are also shown, as well. Overall results reveal that global brain connectivity can be increased by long-term medication in pediatric ADHD-C in terms of increased segregation & resilience. This is the first study to demonstrate that long-term medication can normalize the functional brain connectivity in ADHD, which is characterized by decreased connectivity compared to controls