EEG Sub-bands Research Articles

Exploring Seafarers’ Workload Recognition Model with EEG, ECG and Task Scenarios’ Complexity: A Bridge Simulation Study

Seafarers are prone to reduce behavioral reliability under high workloads, resulting in human errors and accidents. To explore the changes in seafarers’ workload and physiological activities under complex task conditions, a bridge simulator experiment was conducted to collect the EEG and ECG data of 23 seafarers. The power in different EEG sub-bands was extracted from a one-channel EEG acquisition headset employed by Welch’s method and ratio processing. The features such as root mean square of RR interval difference (RMSSD) were extracted from ECG. Then, an improved seafarer workload recognition method based on EEG combined with ECG and complex task scenarios was proposed, and the performance of the machine learning algorithm was evaluated by cross-validation. Compared with the recognition model that only uses the task scenarios as the workload calibration, the EEG recognition model based on the workload level calibrated by the ECG and the task scenarios is more effective, with an accuracy rate of 92.5%, an increase of 25.9%. The results show that the improved workload recognition model can effectively identify seafarers’ workload, and the model trained with the bagging algorithm has the best performance. Furthermore, time domain features of EEG and ECG fluctuate regularly with the task scenarios’ complexity. The research results can develop online intelligent monitoring, and human–computer interaction active early warning technology and equipment.

EEG sub-bands based sleep stages classification using Fourier Synchrosqueezed transform features

Sleep is a state that coordinates all the human body's major functions. Irregularity of sleep or its deprivation disrupts the physical and mental state of a person. An automated method that can analyze sleep stages can help sleep physicians identify and treat various sleep-related ailments is the need of the hour. The property of the Synchrosqueezed transform to concentrate the time–frequency representation of a signal around the ridges is used to extract discriminatory information. EEG signal is divided into seven sub-bands, and features for six sleep stages are extracted. These features include Normalized Sub-band Power, Inter-quartile Range, Mean Absolute Deviation, Movement, and Fourier Synchrosqueezed Transform based features for the six sleep stages over the seven EEG sub-bands. Sleep-EDF, Sleep EDF Extended, and DREAMS databases have marked sleep stages following the Rechtschaffen and Kales (R&K) and the American Academy of Sleep Medicine (AASM) criterion, are used to extract the proposed features. The performance of the SVM classifier is evaluated based on 10-fold cross-validation on the three databases for 2 to 6 class problems. The comparative performance with the recently published results using the same database shows improved performance of the proposed work. Further, the effect of the gamma band is also studied, and its inclusion shows improvement in the proposed model performance.

Early Detection of Mild Cognitive Impairment, Dementia and Alzheimer’s Using qEEG

Abstract This article reports the biomarkers of the Mild Cognitive Impairment (MCI) among the elderly group aged around 60-75 years old by analyzing the EEG signals recorded by using the quantitative electroencephalograph (qEEG). There is growing evidence that EEG analysis in resting state condition are useful in early detection of neural signatures of Alzheimer’s and dementia. EEG findings and analysis shows potential of discriminating MCI, Alzheimer’s and dementia. In this research, the purpose is to 1) develop the indexes for each of the EEG bands and sub-bands such as delta (1.5 – 3.5 Hz), theta (3.3 – 7.5 Hz), alpha ( 8 – 12 Hz), beta 1 (15 – 18 Hz) and beta 2 (22 – 30 Hz); 2) provide reference for early diagnosis; 3) extract and analyze the brainwave pattern of MCI and cognitively healthy group. This study involved 19 channel resting state EEG from a total of 30 subjects, 18 diagnosed as having MCI and 12 cognitively healthy elderly with criteria for inclusion if the mini mental state examination (MMSE) score is more than or equal to 28, based on the age and educational level. Development of qEEG index started by decomposition of EEG by performing Fourier analysis, averaging and normalizing the value from the 19 channels to obtain the z-score. Findings showed promise of utility in early detection of Alzheimer’s. Notably, 1) Increases in delta/ theta posterior temporal and prefrontal area i.e., H(6.64) vs. MCI(26.29); beta temporal region, 2) Decreases in alpha at sensory motor region i.e., H(0.27) vs. MCI(0.23).

Cortical reorganization to improve dynamic balance control with error amplification feedback

BackgroundError amplification (EA), virtually magnify task errors in visual feedback, is a potential neurocognitive approach to facilitate motor performance. With regional activities and inter-regional connectivity of electroencephalography (EEG), this study investigated underlying cortical mechanisms associated with improvement of postural balance using EA.MethodsEighteen healthy young participants maintained postural stability on a stabilometer, guided by two visual feedbacks (error amplification (EA) vs. real error (RE)), while stabilometer plate movement and scalp EEG were recorded. Plate dynamics, including root mean square (RMS), sample entropy (SampEn), and mean frequency (MF) were used to characterize behavioral strategies. Regional cortical activity and inter-regional connectivity of EEG sub-bands were characterized to infer neural control with relative power and phase-lag index (PLI), respectively.ResultsIn contrast to RE, EA magnified the errors in the visual feedback to twice its size during stabilometer stance. The results showed that EA led to smaller RMS of postural fluctuations with greater SampEn and MF than RE did. Compared with RE, EA altered cortical organizations with greater regional powers in the mid-frontal cluster (theta, 4–7 Hz), occipital cluster (alpha, 8–12 Hz), and left temporal cluster (beta, 13–35 Hz). In terms of the phase-lag index of EEG between electrode pairs, EA significantly reduced long-range prefrontal-parietal and prefrontal-occipital connectivity of the alpha/beta bands, and the right tempo-parietal connectivity of the theta/alpha bands. Alternatively, EA augmented the fronto-centro-parietal connectivity of the theta/alpha bands, along with the right temporo-frontal and temporo-parietal connectivity of the beta band.ConclusionEA alters postural strategies to improve stance stability on a stabilometer with visual feedback, attributable to enhanced error processing and attentional release for target localization. This study provides supporting neural correlates for the use of virtual reality with EA during balance training.

Emotion Recognition by Discriminating EEG Segments With High Affective Content From Automatically Selected Relevant Channels

In EEG-based emotion recognition, subject’s affective responses are captured by several channels (scalp level) by presenting target stimuli. However, the emotional responses are inconsistent throughout the acquired signal, and rather, it arises at certain duration with high prominence on certain channels. However, existing studies ignored this vital issue and considered the entire acquired signal for processing leading to inaccurate results. Therefore, this work proposes emotion recognition by identifying highly affective EEG segments from automatically chosen relevant channels based on instantaneous phase synchronization measurement among all channels involving Hilbert transform and phase locking values followed by a majority voting algorithm. Next, a random matrix theory discriminates the appropriate segments whose power spectral density has been used in the <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$k$ </tex-math></inline-formula> -NN-based classification. Experimental validation has been performed using the EEG dataset from SEED (with positive, neutral, and negative emotions). Five EEG subbands ( <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\delta $ </tex-math></inline-formula> , <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\theta $ </tex-math></inline-formula> , <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\alpha $ </tex-math></inline-formula> , <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\beta $ </tex-math></inline-formula> , and <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\gamma$ </tex-math></inline-formula> ) are exclusively examined for their high correlation with emotions. The analyses include distinguishing proper segment lengths and their locations, minimum dominant channels while achieving the best classification for individual subband. The proposed method achieved the highest classification accuracy of up to 95% for higher frequency subbands ( <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\beta $ </tex-math></inline-formula> and <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\gamma$ </tex-math></inline-formula> ) with 15 channels. Furthermore, the segment locations for positive and neutral emotions lie from start till 75% of the entire experiment time, whereas, for negative emotion, it is 25%–75%. Moreover, channels from the left frontal, central, and temporal regions are found very active, which is steady. In a comparative study, the proposed idea demonstrates its superiority by displaying the highest efficiency considering less data from minimum channels.

Корреляции ЭЭГ с клиническими оценками у больных депрессией с суицидальными попытками в анамнезе

Study Objective: to clarify the role of neurophysiological mechanisms in the pathogenesis of suicidal behavior in young depressed patients with a history of suicidal attempts. Study Design: open study. Materials and methods. The study included 35 female patients aged 16-25 years. Relationships between pre-treatment quantitative clinical scores (according to the HDRS-17 scale) and values of the absolute spectral power of the background resting EEG in narrow frequency sub- bands were analyzed. Study Results. Significant negative correlations (p &lt; 0.05) between the EEG parameters and the sum of depression cluster scores of the HDRS-17 scale indicate reduced activation of the frontal-anterotemporal cortex areas of both hemispheres (according to the values of the alpha2 EEG sub-band (9–11 Hz) spectral power in F7, F3, F4 and F8 leads), and especially, of the left anterior temporal region (according to the values of the EEG beta2 sub-band (20–30 Hz) spectral power in F7 lead). In addition, these patients are characterized by a somewhat reduced functional state of the central temporal-parietal zones of the right hemisphere in the form of significant positive correlations (p &lt; 0.05–0.01) of the slow-wave delta (2–4 Hz), theta1 (4–6 Hz) and theta2 (6–8 Hz) EEG activity spectral power values in F4, C4, T4 and P4 leads with the sum of anxiety cluster scores of the HDRS-17 scale. Conclusion. The results obtained make it possible to clarify the neurophysiological mechanisms of the pathogenesis of suicidal behavior in young depressive patients. Keywords: depression, young age, suicide attempts, quantitative EEG.

Quantitative Analysis of Inter- and Intrahemispheric Coherence on Epileptic Electroencephalography Signal.

When an epileptic seizure occurs, the neuron's activity of the brain is dynamically changed, which affects the connectivity between brain regions. The connectivity of each brain region can be quantified by electroencephalography (EEG) coherence, which measures the statistical correlation between electrodes spatially separated on the scalp. Previous studies conducted a coherence analysis of all EEG electrodes covering all parts of the brain. However, in an epileptic condition, seizures occur in a specific region of the brain then spreading to other areas. Therefore, this study applies an energy-based channel selection process to determine the coherence analysis in the most active brain regions during the seizure. This paper presents a quantitative analysis of inter- and intrahemispheric coherence in epileptic EEG signals and the correlation with the channel activity to glean insights about brain area connectivity changes during epileptic seizures. The EEG signals are obtained from ten patients’ data from the CHB-MIT dataset. Pair-wise electrode spectral coherence is calculated in the full band and five sub-bands of EEG signals. The channel activity level is determined by calculating the energy of each channel in all patients. The EEG coherence observation in the preictal (Cohpre) and ictal (Cohictal) conditions showed a significant decrease of Cohictal in the most active channel, especially in the lower EEG sub-bands. This finding indicates that there is a strong correlation between the decrease of mean spectral coherence and channel activity. The decrease of coherence in epileptic conditions (Cohictal <Cohpre) indicates low neuronal connectivity. There are some exceptions in some channel pairs, but a constant pattern is found in the high activity channel. This shows a strong correlation between the decrease of coherence and the channel activity. The finding in this study demonstrates that the neuronal connectivity of epileptic EEG signals is suitable to be analyzed in the more active brain regions.

Hurst entropy profiles for Beta Low and Beta High EEG sub-bands Part I: Intragroup comparison.

The non-linear exploration of the entropic structure of the information contained in an EEG allows access to a new source of data that characterizes, in a more global way, the functioning of the brain as a whole. On the understanding that the brain is a hyper-complex system, describing it only through a linear statistical approach, mainly based on the comparison of differences in the magnitude of the power of the electroencephalographic signal between comparable conditions, leaves incomplete the part of the brain phenomenon that obeys non-linear laws. The present work explores the possibility of characterizing the entropic structure of a human brain, through the construction of order/chaos balance profiles. We proceeded to study the entropic content of the EEG signal, by estimating the Hurst exponent in segments of different sizes, for a set of data from 14 EEG electrodes (Emotiv Epoc Research Edition), taken in basal resting conditions, during 3-5 minutes with eyes closed. The EEG recordings of thirteen subjects (N=13) were studied for the low beta (13-21Hz) and high beta (22-30Hz) sub-bands, in four time-processing windows: 125ms, 250ms, 500ms, and 1s. In addition to obtaining the traditional Hurst estimator, the behavior of two derivatives of the Hurst exponent: the moving-Hurst (muHurst), and the mHurst, an estimator interpreted as a global entropic trend of the time series, were also evaluated. The results showed a clear differentiation between the entropic profiles obtained for low beta and high beta, for all individuals, as well as a different entropic structure of both sub-bands, when comparing Hurst exponents and their derivatives with different time processing windows. At the 1s scale (128 EEG frames), the average entropic content of the time series (muHurst) is relatively higher than the entropic load contained in the mid-term estimation of the whole signal (H(trad)). At the same time, estimates of mHurst, a derivative that assesses the overall entropic trend of the signal, indicated mostly lower entropic (more organized) content than estimates at shorter timescales. Our nonlinear exploration of the entropic structure of the brain revealed a wealth of details that can be used to complement the traditional characterization of the brain that linear approaches have provided so far. Our procedure allowed us to discern substantive differences between the entropic characteristics of the low beta and high beta EEG sub-bands. As well as individual differences within the group, potential substrates for the search of neural markers of psychological, pharmacological, or clinical utility.

Brain Functional Connectivity as Biomarker for Propofol-Induced Alterations of Consciousness.

Understanding neural correlates of consciousness and its alterations poses a grand challenge for modern neuroscience. Even though recent years of research have shown many conceptual and empirical advances, the evolution of a system that can track anesthesia-induced loss of consciousness is hindered by the lack of reliable markers. The work presented herein estimates the functional connectivity (FC) between 21 scalp electroencephalogram (EEG) recordings to evaluate its utility in characterizing changes in brain networks during propofol sedation. The sedation dataset in the University of Cambridge data repository was used for analyses. FC was estimated using the debiased estimator of the squared Weighted Phase Lag Index (dWPLI2). Spectral FC networks before, during, and after sedation was considered for 5 EEG sub-bands. Results demonstrated significantly higher alpha band FC during baseline, mild and moderate sedation, and recovery stages. A striking association between frontal brain activity and propofol-sedation was also noticed. Furthermore, inhibition of frontal to parietal and frontal to occipital connections were observed as characteristic features of propofol-induced alterations in consciousness. A random subspace ensemble framework using logistic model tree as the base classifier, and 18 functional connections as features, yielded a cross-validation accuracy of 98.75% in discriminating baseline, mild and moderate sedation, and recovery stages. These findings validate that EEG-based FC can reliably distinguish altered conscious states associated with anaesthesia.

Midline EEG Functional Connectivity As Biomarker for Conscious States in Sleep and Wakefulness.

Functional connectivity (FC) between different cortical regions of the brain has long been hypothesized to be necessary for conscious states in several modeling and empirical studies. The work presented herein estimates the FC between two bipolar midline electroencephalogram (EEG) recordings to evaluate its utility in discriminating consciousness levels across wakefulness and sleep. Consciousness levels were defined as Low, Medium, and High depending upon the ability of a subject to self-report their experiences at a later stage. The sleep EDF [expanded] dataset available in the Physionet data repository was used for analyses. FC was estimated using the debiased estimator of the squared Weighted Phase Lag Index (dWPLI2) metric. A total of 40 features extracted from the FC spectra for 10 EEG sub-bands were considered. FC trends demonstrated the highest alpha synchrony in the 'Low' conscious state. While the 'Medium' conscious state demonstrated superior phase synchronization in the low-gamma band, the 'High' conscious state was characterized by comparatively lower phase synchronization in all frequency bands. A Multi-Layer Perceptron (MLP) framework using a combination of 7 features yielded the highest cross-validation accuracy of 95.15% in distinguishing these conscious states. The study results provide a pertinent validation for the hypothesis that midline EEG FC is a reliable and robust signature of conscious states in sleep and wakefulness.

EEG personal recognition based on ‘qualified majority’ over signal patches

Electroencephalography (EEG)-based personal recognition in realistic contexts is still a matter of research, with the following issues to be clarified: (1) the duration of the signal length, called ‘epoch’, which must be very short for practical purposes and (2) the contribution of EEG sub-bands. These two aspects are connected because the shorter the epoch’s duration, the lower the contribution of the low-frequency sub-bands while enhancing the high-frequency sub-bands. However, it is well known that the former characterises the inner brain activity in resting or unconscious states. These sub-bands could be of no use in the wild, where the subject is conscious and not in the condition to put himself in a resting-state-like condition. Furthermore, the latter may concur much better in the process, characterising normal subject activity when awake. This study aims at clarifying the problems mentioned above by proposing a novel personal recognition architecture based on extremely short signal fragments called ‘patches’, subdividing each epoch. Patches are individually classified. A ‘qualified majority’ of classified patches allows taking the final decision. It is shown by experiments that this approach (1) can be adopted for practical purposes and (2) clarifies the sub-bands’ role in contexts still implemented in vitro but very similar to that conceivable in the wild.

Emotion recognition by distinguishing appropriate EEG segments based on random matrix theory

This work proposes an emotion recognition technique by distinguishing appropriate electroencephalogram (EEG) segments from acquired signal for target emotions. Generally, EEG based emotion recognition system works considering entire acquired signal from different channels, which is improper as emotional states do not remain steady for entire duration of signal acquisition. Therefore, in this work, appropriate EEG signal segments (L) from diverse channels have been distinguished using random matrix theory (RMT). Next, classification tasks have been performed considering two features of the selected EEG segments — power spectral density (PSD) and continuous wavelet transform (CWT), in three classifiers — k-nearest neighbour (k-NN), support vector machine (SVM) and random forest (RF). Experimental validations have been performed on two EEG datasets for emotion analysis — SEED (distinct emotions) and DEAP (dimensional emotions). The proposed method achieved accuracies upto 95% (L = 60 s) and 86% (L = 2 s) for SEED (positive vs negative) and DEAP (high vs low arousal) respectively, when considering segments from entire EEG signals. Meanwhile, the EEG sub-bands, α, β and γ, displayed the highest accuracy upto 98% (L = 60–80 s) and 94% (L = 2–15 s) for SEED and DEAP respectively. For both cases, k-NN classifier with CWT feature and channels from frontal, temporal and occipital brain regions have been found very suitable. Additionally, the selected EEG segments having low coefficient-of-variation of entropy achieves the highest classification accuracy. In a comparative study, this method reveals its superiority by achieving the highest classification performance by considering automatically selected short length EEG signal segments from distinct channels.

Visual Effect on Brain Connectome That Scales Feedforward and Feedback Processes of Aged Postural System During Unstable Stance.

Older adults with degenerative declines in sensory systems depend strongly on visual input for postural control. By connecting advanced neural imaging and a postural control model, this study investigated the visual effect on the brain functional network that regulates feedback and feedforward processes of the postural system in older adults under somatosensory perturbations. Thirty-six older adults conducted bilateral stance on a foam surface in the eyes-open (EO) and eyes-closed (EC) conditions while their center of pressure (COP) and scalp EEG were recorded. The stochastic COP trajectory was modeled with non-linear stabilogram diffusion analysis (SDA) to characterize shifts in postural control in a continuum of feedback and feedforward processes. The EEG network was analyzed with the phase-lag index (PLI) and minimum spanning tree (MST). The results indicated that visual input rebalanced feedforward and feedback processes for postural sway, resulting in a greater critical point of displacement (CD), short-term effective diffusion coefficients (Ds) and short-term exponent (Hs), but the smaller critical point of time (CT) and long-term exponent (Hl) for the EC state. The EC network demonstrated stronger frontoparietal-occipital connectivity but weaker fronto-tempo-motor connectivity of the theta (4–7 Hz), alpha (8–12 Hz), and beta (13–35 Hz) bands than did the EO network. MST analysis revealed generally greater leaf fraction and maximal betweenness centrality (BCmax) and kappa of the EC network, as compared with those of the EO network. In contrast, the EC network exhibited a smaller diameter and average eccentricity than those of the EO network. The modulation of long-term negative feedback gain of the aged postural system with visual occlusion was positively correlated with leaf fraction, BCmax, and kappa, but negatively correlated with the diameter and average eccentricity for all EEG sub-bands. In conclusion, the aged brain functional network in older adults is tuned to visual information for modulating long-term negative feedback of the postural system under somatosensory perturbations.

Task specific Brain Synchronization in EEG based Speech and Speech Imagery Procedures

Speech being a complicated cognitive skill of the brain, retrieving verbal information from spoken and imagined speech using electrophysiological signal is widely under research to assist the impaired. Electroencephalography has been a successful medium in analyzing the temporal and behavioral changes happening in the brain during any task. In this work an endeavor has been made to distinguish such changes while speaking and envisioning speaking the phoneme of vowel ‘a’. The brain activity was recorded from volunteers while speaking and envisioning speaking the vowel ‘a’ and the basic signal processing routines were applied. The EEG sub-band frequencies were extracted using Fast Fourier Transform (fft). The synchronization indices were assessed for the segmented signals of various sub-band frequencies and the behavioral changes were enlisted. It was seen that the left hemisphere of the brain functioned synchronously during speech task and an inverse phenomenon was observed during the imagery task. Hence the possibility of reconstructing the broken or incomplete speech from the directly recorded brain activity has reached a step ahead towards the milestone. This work exhibits the analysis of brain connectivity parameters while speaking and imagining a vowel ‘a’. Speech imagery can be used to decode the thoughts for patients who are unable to express their thoughts, using non-invasive techniques.

Classifying Electroencephalogram (EEG) Signals Using BAT-SVM Classifier for Detecting Epilepsy

Electroencephalogram (EEG) signals are a valuable source of information for detecting epileptic seizures. However, monitoring EEG for long periods of time is very exhausting and time consuming. Thus, detecting epilepsy in EEG signals automatically is highly appreciated. In this study, three classes, namely normal, interictal (out of seizure time), and ictal (during seizure), are considered. Moreover, a comparative study is provided for the efficient features in literature resulting in a suggested combination of only three discriminative features, namely R'enyi entropy, line length, and energy. These features are calculated from each of the EEG sub-bands. Finally, support vector machines (SVM) classifier optimized using BAT algorithm (BAT-SVM) is introduced by this study for discriminating between the three classes. Experiments were conducted using Andrzejak database. The accomplished experiments and comparisons in this study emphasize the superiority of the proposed BAT-SVM along with the suggested feature set in achieving the best results.

Lacsogram: A New EEG Tool to Diagnose Alzheimer's Disease.

This work proposes the application of a new electroencephalogram (EEG) signal processing tool - the lacsogram - to characterize the Alzheimer's disease (AD) activity and to assist on its diagnosis at different stages: Mild Cognitive Impairment (MCI), Mild and Moderate AD (ADM) and Advanced AD (ADA). Statistical analyzes are performed to lacstral distances between conventional EEG subbands to find measures capable of discriminating AD in all stages and characterizing the AD activity in each electrode. Cepstral distances are used for comparison. Comparing all AD stages and Controls (C), the most important significances are the lacstral distances between subbands θ and α ( p = 0.0014 0.05). The topographic maps show significant differences in parietal, temporal and frontal regions as AD progresses. Machine learning models with a leave-one-out cross-validation process are applied to lacstral/cepstral distances to develop an automatic method for diagnosing AD. The following classification accuracies are obtained with an artificial neural network: 95.55% for All vs All, 98.06% for C vs MCI, 95.99% for C vs ADM, 93.85% for MCI vs ADM-ADA. In C vs MCI, C vs ADM and MCI vs ADM-ADA, the proposed method outperforms the state-of-art methods by 5%, 1%, and 2%, respectively. In All vs All, it outperforms the state-of-art EEG and non-EEG methods by 6% and 2%, respectively. These results indicate that the proposed method represents an improvement in diagnosing AD.

Classification of sleep apnea based on EEG sub-band signal characteristics

Sleep apnea syndrome (SAS) is a disorder in which respiratory airflow frequently stops during sleep. Alterations in electroencephalogram (EEG) signal are one of the physiological changes that occur during apnea, and can be used to diagnose and monitor sleep apnea events. Herein, we proposed a method to automatically distinguish sleep apnea events using characteristics of EEG signals in order to categorize obstructive sleep apnea (OSA) events, central sleep apnea (CSA) events and normal breathing events. Through the use of an Infinite Impulse Response Butterworth Band pass filter, we divided the EEG signals of C3-A2 and C4-A1 into five sub-bands. Next, we extracted sample entropy and variance of each sub-band. The neighbor composition analysis (NCA) method was utilized for feature selection, and the results are used as input coefficients for classification using random forest, K-nearest neighbor, and support vector machine classifiers. After a 10-fold cross-validation, we found that the average accuracy rate was 88.99%. Specifically, the accuracy of each category, including OSA, CSA and normal breathing were 80.43%, 84.85%, and 95.24%, respectively. The proposed method has great potential in the automatic classification of patients' respiratory events during clinical examinations, and provides a novel idea for the development of an automatic classification system for sleep apnea and normal events without the need for expert intervention.

The identification of fractional order systems by multiscale multivariate analysis

Abstract The recognition of complex systems plays a vital role in system dynamics. To distinguish chaotic systems and stochastic processes, the complex-entropy causality plane (CECP) was proposed and received considerable attention. However, few methods have been applied to the identification of fractional order chaotic systems. In this paper, we propose multiscale multivariate fractional dispersion entropy (MMFDE) and MMFDE plane to study the structural complexity of multivariate nonlinear systems. Unlike the ordinal pattern in CECP, the dispersion pattern is used in MMFDE plane to solve the computational burden in fractional order systems. Moreover, the fractional order α in MMFDE can be used to adjust the ability of MMFDE plane to distinguish between different systems. Through simulation data, we find that the MMFDE plane can classify not only chaotic systems and stochastic processes but also different fractional order chaotic systems even if adding noise to the signals. It is then applied to financial time series and epileptic EEG recordings. The results show that the MMFDE plane can identify developed and emerging markets. For epileptic EEG recordings, both the EEG subbands and the original data without filtering can be used to distinguish different states of healthy subjects and patients with epilepsy. The coarse-graining process is also discussed to demonstrate the consequences further.

Quantitative EEG in Prediction of Efficiency of Combined Antidepressive Treatment Using Transcranial Magnetic Stimulation

In order to search for possible EEG predictors of the effectiveness of combined antidepressive therapy, which includes a three-week course of high-frequency rhythmic transcranial magnetic stimulation (rTMS) of the dorsolateral prefrontal cortex (DLPFC) of the left hemisphere in 30 patients with drug-resistant depression the correlations between pre-treatment EEG spectral power values, and post-treatment quantitative outcome conditions of patients were analyzed. The “responders” (23 out of 30) exhibited significant negative correlations of the post-treatment HDRS-17 scale total score with the EEG spectral power values in θ2 (6–8 Hz) and α1 (8–9 Hz) frequency sub-bands in the parietal–occipital–posterior temporal leads. The “non-responders” (7 out of 30) exhibited similar correlations between the post-treatment HDRS-17 scale total score and θ2 (6–8 Hz) EEG sub-band spectral power in the frontal–central–temporal areas of the right hemisphere. Thus, possible EEG predictors of the effectiveness of combined antidepressive therapy, including rTMS of the left DLPFC, are the pre-treatment parameters of θ2 and α1 EEG frequency sub-bands.

A novel electroencephalography based approach for Alzheimer’s disease and mild cognitive impairment detection

Background and objectiveAlzheimer’s disease (AD) is characterized by cognitive, behavioral and intellectual deficits. The term mild cognitive impairment (MCI) is used to describe individuals whose cognitive impairment departing from their expectations for the age that does not interfere with daily activities. To diagnose these disorders, a combination of time-consuming, expensive tests that has difficulties for the target population are evaluated, moreover, the evaluation may yield subjective results. In the presented study, a novel methodology is developed for the automatic detection of AD and MCI using EEG signals. MethodsThis study analyzed the EEGs of 35 subjects (16 MCI, 8 AD, 11 healthy control) with the developed algorithm. The algorithm consists of 3 methods for analysis, discrete wavelet transform(DWT), power spectral density (PSD) and coherence. In the first approach, DWT is applied to the signals to obtain major EEG sub-bands, afterward, PSD of each sub-band is calculated using Burg’s method. In the second approach, interhemispheric coherence values are calculated. The variance and amplitude summation of each sub-bands’ PSD and the amplitude summation of the coherence values corresponding to the major sub-bands are determined as features. Bagged Trees is selected as a classifier among the other tested classification algorithms. Data set is used to train the classifier with 5-fold cross-validation. ResultsAs a result, accuracy, sensitivity, and specificity of 96.5%, 96.21%, 97.96% are achieved respectively. ConclusionIn this study, we have investigated whether EEG can provide efficient clues about the neuropathology of Alzheimer's Disease and mild cognitive impairment for early and accurate diagnosis. Accordingly, a decision support system that produces reproducible and objective results with high accuracy is developed.