Search

The review focuses on the most promising methods for classifying EEG signals for non-invasive BCIs and theoretical approaches for the successful classification of EEG patterns. The paper provides an overview of articles using Riemannian geometry, deep learning methods and various options for preprocessing and "clustering" EEG signals, for example, common-spatial pattern (CSP). Among other approaches, pre-processing of EEG signals using CSP is often used, both offline and online. The combination of CSP, linear discriminant analysis, support vector machine and neural network (BPNN) made it possible to achieve 91% accuracy for binary classification with exoskeleton control as a feedback. There is very little work on the use of Riemannian geometry online and the best accuracy achieved so far for a binary classification problem is 69.3% in the work. At the same time, in offline testing, the average percentage of correct classification in the considered articles for approaches with CSP – 77.5 ± 5.8%, deep learning networks – 81.7 ± 4.7%, Riemannian geometry – 90.2 ± 6.6%. Due to nonlinear transformations, Riemannian geometry-based approaches and complex deep neural networks provide higher accuracy and better extract of useful information from raw EEG recordings rather than linear CSP transformation. However, in real-time setup, not only accuracy is important, but also a minimum time delay. Therefore, approaches using the CSP transformation and Riemannian geometry with a time delay of less than 500 ms may be in the future advantage.

The study provides the data related to synchronization/desynchronization when performing verbal creative (to create the endings of proverbs) and noncreative (to find a synonym for the ending of the known proverb; to remember the known ending of the proverb) tasks, as well as the data on classification of EEG patterns during these tasks. Twenty-four volunteers (18–22 y.o., 20 women, 4 men) participated in the study. Creation of original endings vs. memory task was accompanied by the higher values of EEG power in the frontal region of the right hemisphere in the frequency range 8–9 Hz for the time interval of 400–730 ms. In the parietal region of the left hemisphere in the same frequency range, the higher EEG spectral power EEG values were obtained while creating both original and synonymous endings compared to the “recall/memory” control task. When creating original and synonymous endings, the power of EEG was higher in the central frontal regions for the 14–15 Hz frequency band, as well as in the right hemisphere F4 and P4 used for the tasks with synonyms compared to the memory task (850–950 ms). For the frequencies of 17–21 Hz, there were no differences between the creative and synonymous tasks; at the same time, the creative task as compared to the task for memory was characterized by differences in the parietal sites bilaterally and in the central frontal region only for the frequencies of 17–18 Hz, while the synonyms task as compared to the memory one differed in the specified regions for the frequency range of 17–21 Hz. The EEG signal classification was carried out using the classifier learning software package in the matlab environment. The results of classification were considered on the basis of linear discriminant analysis, the support vector machine, and the method that gave the best classification result. An EEG signal converted to current source density (CSD) from frontal (F3, Fz, F4) and parietal regions (P3, Pz, P4) located on the surface of the skull according to the 10–20 System was used for classification. The average accuracy of single trial classification for three types of tasks in all subjects was 48.7 ± 5% [SD] for the best classifier; the best result of the individual subject (58.5%) was achieved using linear discriminant analysis at the theoretical threshold of random classification of 33%.

Mathematical problem solving is an explicitly formed skill formed during school education. The goal of the work was to study cognitive event-related potentials (ERPs) in solving mathematical equations (addition, subtraction) in adolescents (12–16 years of age) living in different regions of northern Russia: Arkhangelsk (n = 14) and Murmansk (non-Sami, n = 17; Sami, n = 13) regions. The subjects were asked to solve delayed answer verification tasks: the trial consisted of two stimuli: the trial started with presenting a problem for 400 ms; then, after 700 ms, an answer was presented, either correct or incorrect one. The subjects had to press the left mouse button at correct answer. Differences in the ERP amplitude were observed at the interval of 220–270 ms after presentation of the first stimulus due to shorter P2–N3 complex latency in adolescents from the Circumpolar region as compared with adolescents from the Polar region. Differences in the amplitudes of the late positive component at the interval of 430–530 ms were also observed (greater values were observed in the Sami adolescents). Differences between the groups were observed at presentation of a correct solution and were characterized by the greater latency of the negative ERP component in frontal, central, and parietal areas in the adolescents living in the Polar region. Thus, the characteristics of the ERP components during mathematical problem solving show differences probably related to the regional aspects of development in children.

In the paper issues of brain-computer interface applications in assistive technologies are considered in particular for robotic devices control. Noninvasive brain-computer interfaces are built based on the classification of electroencephalographic signals, which show bioelectrical activity in different zones of the brain. Such brain-computer interfaces after training are able to decode electroencephalographic patterns corresponding to different imaginary movements and patterns corresponding to different audio-visual stimulus. The requirements which must be met by brain-computer interfaces operating in real time, so that biological feedback is effective and the user's brain can correctly associate responses with events are formulated. The process of electroencephalographic signal processing in noninvasive brain-computer interface is examined including spatial and temporal filtering, artifact removal, feature selection, and classification. Descriptions and comparison of classifiers based on support vector machines, artificial neural networks, and Riemann geometry are presented. It was shown that such classifiers can provide accuracy at the level of 60-80% for recognition of imaginary movements from two to four classes. Examples of application of the classifiers to control robotic devices were presented. The approach is intended both to help healthy users to perform daily functions better and to increase the quality of life of people with movement disabilities. Tasks to increase the efficiency of technology application are formulated.

The study aimed to explore EEG correlates of a person’s creative (non-trivial) and trivial decisions finding in conditions of associative search by means of event-related synchronization/event-related desynchronization (ERS/ERD) spectral power changes analysis. Sixty subjects had to fulfill RAT (Remote associative test) alike tasks while EEG registering. There were no differences in time responses while remote association and trivial decisions search (around 3 seconds in both case). An increase in ERS was observed for the central frequencies 4 Hz at F7 site (1500-1000 ms before note of decision), 3 Hz at the Fz site (500-1000 ms before note of decision), and 3.5 Hz at T3 site (in the last 500 ms before decision). Obtained findings stressing the role of low frequency bands (theta and delta) in wide associative field search leading for creative and original decisions.

The study aimed to reveal task-related differences in story creation with and without the mental effort of overcoming self-induced stereotypes. Eighteen right-handed subjects (19.3 &plusmn, 1.1 years old) created stories. The subjects reported the formation of story plot stereotypes (as we call them: self-induced) during self-regulated creative production, which had to be overcome with the instruction to continue the story. Creative task fulfillment (without formed stereotypes&mdash, first stage of creation) was characterized by a decrease in the wave percentages of 9&ndash, 10 Hz, 10&ndash, 11 Hz and 11&ndash, 12 Hz frequencies and EEG desynchronization (decreases in EEG spectral power) in the theta (4&ndash, 8 Hz), alpha1 (8&ndash, 10 Hz) and alpha2 (10&ndash, 13 Hz) frequency bands in comparison with the REST (random episodic silent thought) state. The effortful creation task (with overcoming of self-induced stereotypes-second stage of creation) was characterized by increases in waves with frequencies of 9&ndash, 11 Hz, 11&ndash, 12 Hz in temporal, occipital areas and pronounced EEG synchronization in alpha1,2 frequency bands in comparison with the free creation condition. It was also found, that the participants with the higher originality scores in psychological tests demonstrated increased percentage of high frequencies (11&ndash, 12 Hz in comparison with those who had lower originality scores. Obtained results support the role of alpha and theta frequency bands dynamics in creative cognition.

Components of event-related potentials (ERP) were studied using an active sound signal discrimination task in northern schoolchildren aged 12–18 years with impaired mental development and learning difficulties (IQ 65 ± 4) and without mental developmental impairments (IQ 90 ± 10). The peak latency of the P300 component in the central and parietal areas of the cortex on perception and discrimination of target sound stimuli was greater in northern children with impaired mental development (Cz 370 ± 36 msec, 8.8 ± 4.1 μV; Pz 375 ± 39 msec, 10.2 ± 3.3 μV) as compared with their normally developing peers (339 ± 28 msec, 10.4 ± 3.0 μV; 340 ± 35 msec, 12.2 ± 2.2 μV, respectively). Negative correlations were found between cognitive abilities, the accuracy and speed of information processing in an attention task, and the latency of the P300 component, with positive correlations with the amplitude of the P3 component for measures of attention. The spatial, temporal, and amplitude characteristics of ERP in adolescents with impaired mental development reflect the functional incompetence or immaturity of the hierarchical organization of the system processing and discriminating information arriving in the brain and requiring concentration of attention and decision-taking.

We studied neurophysiological characteristics of the age-related development in junior school students (7–8 and 10–11 years of age) living in the Arctic region of the Russian Federation. The background electroencephalograms (EEGs) were recorded during quiet wakefulness with the eyes closed and open, and event related potentials (ERP) were recorded during the passive perception of sound stimuli within the oddball paradigm in the group of children (33 subjects, 18 boys and 15 girls). A decrease of the latency period and the spatial rearrangement of mismatch negativity with an increase in the amplitude in the centrofrontal cortex have been revealed in the groups of children aged 10–11 years during the perception of a rare stimulus and a decrease of the latency period of the Р300 component in the central and parietal areas associated with the maturation of mechanisms for involuntary auditory attention. Age-specific differences in the components of auditory ERP (N1 and N2) have been shown during passive perception of rare and frequent sounds, which reflect the processes of the morphofunctional maturation of the brain cortex in healthy Arctic school students (an increase of the N1 component amplitude, a decrese of the amplitude and the latency period of the N2 component). The analysis of the background EEG characteristics has shown both the common features, such as a decrease with age of the EEG power in the Δ and θ bands in the eyes-open state, and the different direction and topographic specificities in the age-dependent reorganization of bioelectrical activity in boys and girls in the α1 and α2 EEG bands.

The aim of the present study was to investigate the reading mechanisms in adults (27 subjects; mean age, 19.5 ± 0.8 [SD] years) with different levels of written text comprehension using fMRI. The main objective was to analyze the basic brain mechanisms of verbal stimuli perception with and without semantic component during reading discrimination tasks. The BOLD signal changes during WORD and PSEUDOWORD reading comparing to GAZE FIXATION state were estimated using both analysis of whole brain activation and ROIs (structures connected with the brain system providing reading) in two groups of subjects, “good” and “poor” readers. It was revealed that activations were higher in “poor” readers in lingual gyrus, SMG, STG compared to “good” readers during PSEUDOWORD reading. It was supposed, that the strategies of words and pseudowords recognition differed in two groups of readers: “good” readers identified words or pseudowords already at the stage of visual analysis of “word” structure and demonstrated attempts to decode pseudowords (i.e., language lexical zones were not activated); “poor” readers, apparently, tried to read pseudowords using the same strategy as for the words reading referring to the lexicon, and after failure identified pseudowords as meaningless concepts. In that case, activations of both lexical “language” zones and visual word form area (VWFA) were observed.

Event-related potentials (ERPs) during passive perception of auditory stimuli were studied using the oddball paradigm in healthy and mild mentally retarded adolescents. The study involved 25 subjects aged 11–15 (13.1 ± 1.4) years from Northern Russia, including Arctic regions. The peak latency of the difference wave for deviant and standard stimuli in frontal central derivation was 129 ± 21 ms in the healthy children, and the mean amplitude was–2.6 ± 1.3 μV. In the mentally retarded group, a negative peak of the difference wave was observed only in 9 out of 13 adolescents, its latency was more than in the healthy adolescents (156 ± 29 ms), and the mean amplitude was–2.1 ± 1.4 μV. Differences in perception of deviant and standard stimuli were observed in the healthy adolescents, in particular, along the central line. In the adolescents with mental disorders, there was no significant difference in the fronto-central and central derivations. A discriminant analysis of the amplitudes of ERP components observed in the fronto-central derivations in response to deviant stimuli and the difference in amplitude between ERPs evoked in the fronto-central derivation by standard and deviant stimuli differentiated the adolescents with and without mental disorders. Based on the findings, ERP components in the oddball paradigm were assumed to provide potential markers of disorders in mental development.

The results of four-class classification of the motor imagery EEG patterns corresponding to the right hand finger movements (little finger, thumb, index and middle fingers) of eight healthy subjects are presented in this study. The motor imagery of individual right-hand finger movements was executed by the subjects in a prescribed rhythm and the trials contained no external stimuli. Classification was performed by means of a specially developed two-level committee of classifiers on the basis of support vector machine and artificial neural networks at the first level and by generalizing an artificial neural network at the second level. The area under the EEG signal curve and the curve length calculated in a sliding time window for sites F 3, C 3, and Cz of the International 10?20 System were selected as the key features of signals from the sensorimotor and adjoining frontal cortical areas contralateral to the movements. The average accuracy of four-class singletrial classification for all subjects was 50 ± 7 [SD] (maximum, 58%) for the pair of sites F 3–C 3 and 46 ± 11% [SD] (maximum 62%) for the pair of sites C 3–Cz with a theoretical guessing level 25%.

The article, based on the investigations carried out in 91 school students aged 7–18 years who were exposed to severe weather conditions as residents of the European North of Russia, analyzes the processes underlying the development of structures for providing the spatial synchronization of cerebral bioelectric potentials. The coherence values were evaluated for 19 channels (171 pairs of leads) in five basic EEG frequency bands (Δ, θ, α1, α2, and β). The article describes the age-related dynamics, gender-specific features, and topical specificity in the development of coherent connections in the left and right cerebral hemispheres, as well as in the formation of interhemispheric and interregional connections. We used a computerized evaluation technique for distinguishing between three school age groups of northerners using the frequency structure of EEG coherence with the purpose of elaborating some criteria for diagnosing school students with a developmental delay in the formation of spatial organization for the interaction of local EEG processes. The recorded age-related changes in the EEG-pattern structure reflect the pattern of morphofunctional brain development in children and adolescents exposed to the severe conditions of the North at different stages of their postnatal ontogenesis.

The article presents the results of the analysis of EEG wave structure formation in children and adolescents aged 7-18 years living under severe conditions of the North. The approaches developed in discrete mathematics (the graph theory, the theory of network flows) were used to assess the time-frequency transformations of EEG patterns. We evaluated conditional probabilities of reciprocal transitions between the components of six frequency bands of E EG (delta, theta, alpha-1, alpha-2, beta-1, beta-2). We described age- and sex-related features as well as regional specificities of the EEG wave structure. We defined the age periods of reorganization of diffuse EEG activities into the main EEG rhythms; the role of distinct rhythms in the maintenance of the EEG wave structure and its dynamic rearrangements was also discussed. The age-related changes of the structure of EEG patterns form some general picture of the morphofunctional development of brain in children and adolescents at different stages of postnatal ontogenesis under severe climate and socio-economic conditions of the North.

Objective This study aimed to find effective approaches to electroencephalographic (EEG) signal analysis and resolve problems of real and imaginary finger movement pattern recognition and categorization for one hand. Methods and materials Eight right-handed subjects (mean age 32.8 [SD = 3.3] years) participated in the study, and activity from sensorimotor zones (central and contralateral to the movements/imagery) was recorded for EEG data analysis. In our study, we explored the decoding accuracy of EEG signals using real and imagined finger (thumb/index of one hand) movements using artificial neural network (ANN) and support vector machine (SVM) algorithms for future brain–computer interface (BCI) applications. Results The decoding accuracy of the SVM based on a Gaussian radial basis function linearly increased with each trial accumulation (mean: 45%, max: 62% with 20 trial summarizations), and the decoding accuracy of the ANN was higher when single-trial discrimination was applied (mean: 38%, max: 42%). The chosen approaches of EEG signal discrimination demonstrated differential sensitivity to data accumulation. Additionally, the time responses varied across subjects and inside sessions but did not influence the discrimination accuracy of the algorithms. Conclusion This work supports the feasibility of the approach, which is presumed suitable for one-hand finger movement (real and imaginary) decoding. These results could be applied in the elaboration of multiclass BCI systems.

The article presents the results of classification of EEG patterns registered during imagined rhytmic movements of the fingers of the right hand (little, thumb, index, middle fingers) in 8 healthy subjects. The subjects imagined finger movements in a given rhythm; no external stimuli were used. A two-level committee of classifiers was developed for decoding: the first level included support vector machines and artificial neural networks; the second level included artificial neural network used for generalizing. As the key parameters for classification, we used the area of zone under the envelope of EEG signal and the length of the envelope calculated in sliding time window for leads F3, C3 and Cz in system 10-20 were chosen as the key features for signals of sensorimotor and adjoining frontal area cortex contralateral to movements. The accuracy of classification of single trials for 4 movements averaged for all subjects for the pair of leads F3-C3 was 50 ± 7% [SD] (maximal - 58%); for the pair of leads C3-Cz, 46 ± 11% [SD] (maximal - 62%); theoretical guessing level is 25%.

This study aimed to finding effective approaches for electroencephalographic (EEG) multiclass classification of imaginary movements. The combined classifier of EEG signals based on artificial neural network (ANN) and support vector machine (SVM) algorithms was applied. Effectiveness of the classifier was shown in 4-class imaginary finger movement classification. Nine right-handed subjects participated in the study. The mean decoding accuracy using combined heterogeneous classifier committee was −60 ± 10 %, max: 77 ± 5 %, while application of homogeneous classifier based on committee of ANNs −52 ± 9 % and 65 ± 5 % correspondingly. This work supports the feasibility of the approach, which is presumed suitable for imaginary movements decoding of four fingers of one hand. These results could be used for development of effective non-invasive BCI with enlarged amount of degrees of freedom.

The paper presents data on the formation of spatial synchronization of brain potentials in 91 children aged 7-18 years living in European North of Russia. We estimated coherence values for 19 derivations (pair 171) in five EEG frequency bands (delta, theta, alpha-1, alpha-2 and beta). We described age-related changes, gender differences and topical specific features of the formation of coherence in the left and right hemispheres, and in inter- and intrahemispheric synchronization. We carried out computer assessment of the differences in EEG coherence between three age groups of children in order to determine criteria for identification of children with retarded formation of spatial organization of local EEG processes. Age-related changes in the structure of EEG patterns observed in the study reflect the processes of morphofunctional brain development in children and adolescents at different stages of postnatal ontogenesis under severe conditions of northern climate.

The article presents the results of the analysis of EEG wave structure formation in children and adolescents aged 7-18 years living under severe conditions of the North. The approaches developed in discrete mathematics (the graph theory, the theory of network flows) were used to assess the time-frequency transformations of EEG patterns. We evaluated conditional probabilities of reciprocal transitions between the components of six frequency bands of E EG (delta, theta, alpha-1, alpha-2, beta-1, beta-2). We described age- and sex-related features as well as regional specificities of the EEG wave structure. We defined the age periods of reorganization of diffuse EEG activities into the main EEG rhythms; the role of distinct rhythms in the maintenance of the EEG wave structure and its dynamic rearrangements was also discussed. The age-related changes of the structure of EEG patterns form some general picture of the morphofunctional development of brain in children and adolescents at different stages of postnatal ontogenesis under severe climate and socio-economic conditions of the North.

Historical background of error detection (ED) studies is restored here from the first suggestion of such a mechanism published (Rabbit, 1966) and the first related anatomo-physiological correlates observed [Bechtereva, N.P., Gretchin, V.B., 1968. Physiological foundations of mental activity. Int. Rev. Neurobiol., vol. 11. Academic Press, N.Y., pp. 239–246; Bechtereva, N.P., 1971. Neurophysiological Aspects of Human Mental Activity. Meditzina, Moskow. 120 pp., (in Russian); Bechtereva, N.P., 1974. Neurophysiological Aspects of Human Mental Activity, second edition, revied and complete Meditzina, Moskow. 151 pp., (in Russian)]. Data from evoked potentials together with new opportunities offered by the technological revolution of the 1980s–1990s provided a large body of knowledge on the ED. The overwhelming majority of the papers stress the spatial relation of ED to Anterior Cingulate Cortex. ED was revealed in a number of other zones to whose role should be specially discussed. The other point of interest is the late appearance of ED after the brain signs of correction which seems particularly important considering the supposed functional role of ED. Data of direct observations of ECoG dynamics in left and right human ACC on correct and erroneous test performance are presented. Research on ED resulted in the development of new ways in treatment of the obsessive–compulsive syndrome. Further psychophysiological research into the ED phenomena is considered as one of the priorities of fundamental and applied investigations for the elucidation of human brain functions. Opinion that ED plays an extremely important role in mechanisms of cognition and creativity is further argumented. Investigations in the field can contribute a lot to clinical neurophysiology as well.

The EEG spatial synchronization in the frequency bands Δ, θ, α1, α2, β1, and β2 was studied on the basis of estimations of corresponding mean values of the coherence function in two groups of subjects (students actors and nonactors) during internal induction of emotional states by means of recall of autobiographic experience and mental performance of given scenic situations as a known character. Emotion-induced changes in the state of the cortex were reflected in statistically significant heterogeneous changes in the EEG spatial synchronization in many cortical areas and different frequency bands. The results obtained are discussed in combination with data presented earlier on changes in EEG local synchronization obtained in the course of the same experiments. The EEG β bands stand out against the background of extensive changes in the local and spatial EEG synchronization: the EEG changes in these bands are most prominent. Also, the changes in the prefrontal and temporal cortical areas, most involved in actualization of emotional states, are noteworthy. Some of the findings can be interpreted as confirming the hypothesis about neurophysiological mechanisms of brain defense from functionally excessive emotions.

An electroencephalogram (EEG) was recorded in 19 standard derivations in two groups of subjects (student actors and nonactors) during emotionally positive and negative mental recalls of personal experiences (test 1) or mental reproduction of scenic situations (test 2), as well as during mental count of time. Statistical comparison of EEG spectral power estimations in the frequency bands Δ, θ, α1, α2, β1, and β2 showed that the induction of emotionally positive and emotionally negative states led to statistically significant changes in the EEG absolute power (local synchronization) simultaneously in many derivations and frequency bands. Analysis of all possible comparisons and changes in frequency bands showed that bilateral prefrontal and temporal cortical areas are most active during the internal induction of emotional states. Emotionally positive states were characterized, predominantly, by higher local EEG power than emotionally negative states. The EEG power changes accompanying internal induction of emotionally negative and positive states were of the same order in the case of reproduction of a presented situation (scenic material), whereas, in the case of recall of personal experience, emotionally negative states were accompanied by substantially weaker EEG spectral changes than emotionally positive states. The internal induction of emotional states on the basis of scenic material was accompanied by a pronounced increase in the power of the EEG Δ range over the whole cortical surface. EEG reactions to induction of emotional states were generally stronger in actors than in nonactors. In case of emotional induction on the basis of scenic material, qualitative differences in the EEG reactions were also revealed between the groups. The findings are discussed in comparison with known data of investigations into regional cerebral blood flow during internal emotional induction and EEG studies of internal and external emotional induction.

The EEG recording was made when the subjects performed tasks that involved overcoming the stereotype (creative) and retrieving information from memory (noncreative) with the usual and complicated presentation of the initial material (incomplete proverbs and sayings without concluding words). The subjective complexity of the task performance under different conditions was assessed. The EEG power from 19 EEG derivations was compared in the β2 and γ frequency bands. The creative task performance was associated with a marked increase in the EEG power; significantly more complicated noncreative tasks were not accompanied by marked changes in the EEG power in these bands.