Narrow Results

The automatic seizure detection system can effectively help doctors to monitor and diagnose epilepsy thus reducing their workload. Many outstanding studies have given good results in the two-class seizure detection problems, but most of them are based on hand-wrought feature extraction. This study proposes an end-to-end automatic seizure detection system based on deep learning, which does not require heavy preprocessing on the EEG data or feature engineering. The fully convolutional network with three convolution blocks is first used to learn the expressive seizure characteristics from EEG data. Then these robust EEG features pertinent to seizures are presented as an input to the Nested Long Short-Term Memory (NLSTM) model to explore the inherent temporal dependencies in EEG signals. Lastly, the high-level features obtained from the NLSTM model are fed into the softmax layer to output predicted labels. The proposed method yields an accuracy range of 98.44–100% in 10 different experiments based on the Bonn University database. A larger EEG database is then used to evaluate the performance of the proposed method in real-life situations. The average sensitivity of 97.47%, specificity of 96.17%, and false detection rate of 0.487 per hour are yielded. For CHB–MIT Scalp EEG database, the proposed model also achieves a segment-level sensitivity of 94.07% with a false detection rate of 0.66 per hour. The excellent results obtained on three different EEG databases demonstrate that the proposed method has good robustness and generalization power under ideal and real-life conditions.

Epilepsy is one of the most common neurological diseases, which can seriously affect the patient’s psychological well-being and quality of life. An accurate and reliable seizure prediction system can generate alarm before epileptic seizures to provide patients and their caregivers with sufficient time to take appropriate action. This study proposes an efficient seizure prediction system based on deep learning in order to anticipate the onset of the seizure as early as possible. Handcrafted features extracted based on the prior knowledge and hidden deep features are complementarily fused through the feature fusion module, and then the hybrid features are fed into the multiplicative long short-term memory (MLSTM) to explore the temporal dependency in EEG signals. A one-dimensional channel attention mechanism is implemented to emphasize the more representative information in the multi-channel output of the MLSTM. Finally, a transfer learning strategy is proposed to transfer the weights of the base model trained on the EEG data of all patients to the target patient model, and the latter is then continuously trained using the EEG data of the target patient. The proposed method achieves an average sensitivity of 95.56% and a false positive rate (FPR) of 0.27/h on the SWEC-ETHZ intracranial EEG data. For the more challenging CHB-MIT scalp EEG database, an average sensitivity of 89.47% and a FPR of 0.34/h are obtained. Experimental results demonstrate that the proposed method has good robustness and generalization ability in both intracranial and scalp EEG signals.

EEG is useful for the analysis of the functional activity of the brain and a detailed assessment of this non-stationary waveform can provide crucial parameters indicative of the mental state of patients. The complex nature of EEG signals calls for automated analysis using various signal processing methods. This paper attempts to classify the EEG signals of normal and depression patients using well-established signal processing techniques involving relative wavelet energy (RWE) and artificial feedForward neural network. High frequency noise present in the recorded signal is removed using total variation filtering (TVF). Classification of the frequency bands of EEG signals into appropriate detail levels and approximation level is carried out using an eight-level multiresolution decomposition method of discrete wavelet transform (DWT). Parseval's theorem is used for calculating the energy at different resolution levels. RWE analysis gives information about the signal energy distribution at different decomposition levels. Both RWE and feedforward Network are used to classify the signals from normal controls and depression patients. The performance of the artificial neural network was evaluated using the classification accuracy and its value of 98.11% indicates a great potential for classifying normal and depression signals.

Currently, Fourier-based, wavelet-based, and Hilbert-based time–frequency techniques have generated considerable interest in classification studies for emotion recognition in human–computer interface investigations. Empirical mode decomposition (EMD), one of the Hilbert-based time–frequency techniques, has been developed as a tool for adaptive signal processing. Additionally, the multi-variate version strongly influences designing the common oscillation structure of a multi-channel signal by utilizing the common instantaneous concepts of frequency and bandwidth. Additionally, electroencephalographic (EEG) signals are strongly preferred for comprehending emotion recognition perspectives in human–machine interactions. This study aims to herald an emotion detection design via EEG signal decomposition using multi-variate empirical mode decomposition (MEMD). For emotion recognition, the SJTU emotion EEG dataset (SEED) is classified using deep learning methods. Convolutional neural networks (AlexNet, DenseNet-201, ResNet-101, and ResNet50) and AutoKeras architectures are selected for image classification. The proposed framework reaches 99% and 100% classification accuracy when transfer learning methods and the AutoKeras method are used, respectively.

Complexity measures have been enormously used in schizophrenia patients to estimate brain dynamics. However, the conflicting results in terms of both increased and reduced complexity values have been reported in these studies depending on the patients’ clinical status or symptom severity or medication and age status. The objective of this study is to investigate the nonlinear brain dynamics of chronic and medicated schizophrenia patients using distinct complexity estimators. EEG data were collected from 22 relaxed eyes-closed patients and age-matched healthy controls. A single-trial EEG series of 2min was partitioned into identical epochs of 20s intervals. The EEG complexity of participants were investigated and compared using approximate entropy (ApEn), Shannon entropy (ShEn), Kolmogorov complexity (KC) and Lempel–Ziv complexity (LZC). Lower complexity values were obtained in schizophrenia patients. The most significant complexity differences between patients and controls were obtained in especially left frontal (F3) and parietal (P3) regions of the brain when all complexity measures were applied individually. Significantly, we found that KC was more sensitive for detecting EEG complexity of patients than other estimators in all investigated brain regions. Moreover, significant inter-hemispheric complexity differences were found in the frontal and parietal areas of schizophrenics’ brain. Our findings demonstrate that the utilizing of sensitive complexity estimators to analyze brain dynamics of patients might be a useful discriminative tool for diagnostic purposes. Therefore, we expect that nonlinear analysis will give us deeper understanding of schizophrenics’ brain.

Alzheimer’s Disease (AD) is a neurodegenerative disorder and the most common type of dementia with a great prevalence in western countries. The diagnosis of AD and its progression is performed through a variety of clinical procedures including neuropsychological and physical examination, Electroencephalographic (EEG) recording, brain imaging and blood analysis. During the last decades, analysis of the electrophysiological dynamics in AD patients has gained great research interest, as an alternative and cost-effective approach. This paper summarizes recent publications focusing on (a) AD detection and (b) the correlation of quantitative EEG features with AD progression, as it is estimated by Mini Mental State Examination (MMSE) score. A total of 49 experimental studies published from 2009 until 2020, which apply machine learning algorithms on resting state EEG recordings from AD patients, are reviewed. Results of each experimental study are presented and compared. The majority of the studies focus on AD detection incorporating Support Vector Machines, while deep learning techniques have not yet been applied on large EEG datasets. Promising conclusions for future studies are presented.

Diagnosis of learning difficulties is a challenging goal. There are huge number of factors involved in the evaluation procedure that present high variance among the population with the same difficulty. Diagnosis is usually performed by scoring subjects according to results obtained in different neuropsychological (performance-based) tests specifically designed to this end. One of the most frequent disorders is developmental dyslexia (DD), a specific difficulty in the acquisition of reading skills not related to mental age or inadequate schooling. Its prevalence is estimated between 5% and 12% of the population. Traditional tests for DD diagnosis aim to measure different behavioral variables involved in the reading process. In this paper, we propose a diagnostic method not based on behavioral variables but on involuntary neurophysiological responses to different auditory stimuli. The experiments performed use electroencephalography (EEG) signals to analyze the temporal behavior and the spectral content of the signal acquired from each electrode to extract relevant (temporal and spectral) features. Moreover, the relationship of the features extracted among electrodes allows to infer a connectivity-like model showing brain areas that process auditory stimuli in a synchronized way. Then an anomaly detection system based on the reconstruction residuals of an autoencoder using these features has been proposed. Hence, classification is performed by the proposed system based on the differences in the resulting connectivity models that have demonstrated to be a useful tool for differential diagnosis of DD as well as a method to step towards gaining a better knowledge of the brain processes involved in DD. The results corroborate that nonspeech stimulus modulated at specific frequencies related to the sampling processes developed in the brain to capture rhymes, syllables and phonemes produces effects in specific frequency bands that differentiate between controls and DD subjects. The proposed method showed relatively high sensitivity above 0.6, and up to 0.9 in some of the experiments.

Epilepsy is a neurological disorder related to frequent seizures. Automatic seizure prediction is crucial for the prevention and treatment of epilepsy. In this paper, we propose a novel model for seizure prediction that incorporates a convolutional neural network (CNN) with multi-head attention mechanism. In this model, the shallow CNN automatically captures the EEG features, and the multi-headed attention focuses on discriminating the effective information among these features for identifying pre-ictal EEG segments. Compared with current CNN models for seizure prediction, the embedded multi-headed attention empowers the shallow CNN to be more flexible, and enables improvement of the training efficiency. Hence, this compact model is more resistant to being trapped in overfitting. The proposed method was evaluated over the scalp EEG data from the two publicly available epileptic EEG databases, and achieved outperforming values of event-level sensitivity, false prediction rate (FPR), and epoch-level F1. Furthermore, our method achieved the stable length of seizure prediction time that was between 14 and 15 min. The experimental comparisons showed that our method outperformed other prediction methods in terms of prediction and generalization performance.

Emotion and affect play crucial roles in human life that can be disrupted by diseases. Functional brain networks need to dynamically reorganize within short time periods in order to efficiently process and respond to affective stimuli. Documenting these large-scale spatiotemporal dynamics on the same timescale they arise, however, presents a large technical challenge. In this study, the dynamic reorganization of the cortical functional brain network during an affective processing and emotion regulation task is documented using an advanced multi-model electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) technique. Sliding time window correlation and $k$-means clustering are employed to explore the functional brain connectivity (FC) dynamics during the unaltered perception of neutral (moderate valence, low arousal) and negative (low valence, high arousal) stimuli and cognitive reappraisal of negative stimuli. Betweenness centralities are computed to identify central hubs within each complex network. Results from 20 healthy subjects indicate that the cortical mechanism for cognitive reappraisal follows a ‘top-down’ pattern that occurs across four brain network states that arise at different time instants (0–170ms, 170–370ms, 380–620ms, and 620–1000ms). Specifically, the dorsolateral prefrontal cortex (DLPFC) is identified as a central hub to promote the connectivity structures of various affective states and consequent regulatory efforts. This finding advances our current understanding of the cortical response networks of reappraisal-based emotion regulation by documenting the recruitment process of four functional brain sub-networks, each seemingly associated with different cognitive processes, and reveals the dynamic reorganization of functional brain networks during emotion regulation.

Epileptic seizure prediction is one of the most used therapeutic adjuvant strategies for drug-resistant epilepsy. Conventional methods usually adopt handcrafted features and manual parameter setting. The over-reliance on the expertise of specialists may lead to weak exploitation of features and low popularization of clinical application. This paper proposes a novel parameterless patient-specific method based on Fourier Neural Network (FNN), where the Fourier transform and backpropagation learning are synthesized to make the predictor more efficient and practical. The employment of FNN is the first attempt in the field of seizure prediction due to its automatic extraction of immanent spectra in epileptic signals. Despite the self-adaptive superiority of FNN, we introduce Convolutional Neural Network (CNN) to further improve its search capability in high-dimensional feature spaces. The study also develops a multi-layer module to estimate spectral power ratios of raw recordings, which optimizes the prediction by enhancing feature diversity. Based on these modules, this paper proposes a two-channel deep neural network: Fourier Ratio Convolutional Neural Network (FRCNN). To demonstrate the reliability of the model, we explain the mathematical meaning of hidden-layer neurons in FRCNN theoretically. This approach is evaluated on both intracranial and scalp EEG datasets. It shows that the predictor achieved a sensitivity of 91.2% and a false prediction rate (FPR) of 0.06h${}^{-1}$ across intracranial subjects and a sensitivity of 85.4% and an FPR of 0.14h${}^{-1}$ over scalp subjects. The results indicate that FRCNN enables the convenience of epilepsy treatments while preserving a high degree of precision. In the end, a detailed comparison with the previous methods demonstrates that FRCNN has achieved higher performance and generalization ability.

Integration of brain–computer interface (BCI) technique and assistive device is one of chief and promising applications of BCI system. With BCI technique, people with disabilities do not have to communicate with external environment through traditional and natural pathways like peripheral nerves and muscles, and could achieve it only by their brain activities. In this paper, we designed an electroencephalogram (EEG)-based wheelchair which can be steered by users' own thoughts without any other involvements. We evaluated the feasibility of BCI-based wheelchair in terms of accuracies and real-world testing. The results demonstrate that our BCI wheelchair is of good performance not only in accuracy, but also in practical running testing in a real environment. This fact implies that people can steer wheelchair only by their thoughts, and may have a potential perspective in daily application for disabled people.

Aim of this study was to explore the EEG functional connectivity in amnesic mild cognitive impairments (MCI) subjects with multidomain impairment in order to characterize the Default Mode Network (DMN) in converted MCI (cMCI), which converted to Alzheimer’s disease (AD), compared to stable MCI (sMCI) subjects. A total of 59 MCI subjects were recruited and divided -after appropriate follow-up- into cMCI or sMCI. They were further divided in MCI with linguistic domain (LD) impairment and in MCI with executive domain (ED) impairment. Small World (SW) index was measured as index of balance between integration and segregation brain processes. SW, computed restricting to nodes of DMN regions for all frequency bands, evaluated how they differ between MCI subgroups assessed through clinical and neuropsychological four-years follow-up. In addition, SW evaluated how this pattern differs between MCI with LD and MCI with ED. Results showed that SW index significantly decreased in gamma band in cMCI compared to sMCI. In cMCI with LD impairment, the SW index significantly decreased in delta band, while in cMCI with ED impairment the SW index decreased in delta and gamma bands and increased in alpha1 band. We propose that the DMN functional alterations in cognitive impairment could reflect an abnormal flow of brain information processing during resting state possibly associated to a status of pre-dementia.

Epilepsy is a neurological disorder caused by brain dysfunction, which could cause uncontrolled behavior, loss of consciousness and other hazards. Electroencephalography (EEG) is an indispensable auxiliary tool for clinical diagnosis. Great progress has been made by current seizure identification methods. However, the performance of the methods on different patients varies a lot. In order to deal with this problem, we propose an automatic seizure identification method based on brain connectivity learning. The connectivity of different brain regions is modeled by a graph. Different from the manually defined graph structure, our method can extract the optimal graph structure and EEG features in an end-to-end manner. Combined with the popular graph attention neural network (GAT), this method achieves high performance and stability on different patients from the CHB-MIT dataset. The average values of accuracy, sensitivity, specificity, F1-score and AUC of the proposed model are 98.90%, 98.33%, 98.48%, 97.72% and 98.54%, respectively. The standard deviations of the above five indicators are 0.0049, 0.0125, 0.0116 and 0.0094, respectively. Compared with the existing seizure identification methods, the stability of the proposed model is improved by 78–95%.

The Temporal Sampling Framework (TSF) theorizes that the characteristic phonological difficulties of dyslexia are caused by an atypical oscillatory sampling at one or more temporal rates. The LEEDUCA study conducted a series of Electroencephalography (EEG) experiments on children listening to amplitude modulated (AM) noise with slow-rythmic prosodic (0.5–1Hz), syllabic (4–8Hz) or the phoneme (12–40Hz) rates, aimed at detecting differences in perception of oscillatory sampling that could be associated with dyslexia. The purpose of this work is to check whether these differences exist and how they are related to children’s performance in different language and cognitive tasks commonly used to detect dyslexia. To this purpose, temporal and spectral inter-channel EEG connectivity was estimated, and a denoising autoencoder (DAE) was trained to learn a low-dimensional representation of the connectivity matrices. This representation was studied via correlation and classification analysis, which revealed ability in detecting dyslexic subjects with an accuracy higher than 0.8, and balanced accuracy around 0.7. Some features of the DAE representation were significantly correlated ($p<0.005$) with children’s performance in language and cognitive tasks of the phonological hypothesis category such as phonological awareness and rapid symbolic naming, as well as reading efficiency and reading comprehension. Finally, a deeper analysis of the adjacency matrix revealed a reduced bilateral connection between electrodes of the temporal lobe (roughly the primary auditory cortex) in DD subjects, as well as an increased connectivity of the F7 electrode, placed roughly on Broca’s area. These results pave the way for a complementary assessment of dyslexia using more objective methodologies such as EEG.

The detection and quantification of seizures can be achieved through the analysis of nonstationary electroencephalogram (EEG) signals. The detection of these intractable seizures involving human beings is a challenging and difficult task. The analysis of EEG through human inspection is prone to errors and may lead to false conclusions. The computer-aided systems have been developed to assist neurophysiologists in the identification of seizure activities accurately. We propose a new machine learning and signal processing-based automated system that can detect epileptic episodes accurately. The proposed algorithm employs a promising time-frequency tool called tunable-Q wavelet transform (TQWT) to decompose EEG signals into various sub-bands (SBs). The fractal dimensions (FDs) of the SBs have been used as the discriminating features. The TQWT has many attractive features, such as tunable oscillatory attribute and time-invariance property, which are favorable for the analysis of nonstationary and transient signals. Fractal dimension is a nonlinear chaotic trait that has been proven to be very useful in the analysis and classifications of nonstationary signals including EEG. First, we decompose EEG signals into the desired SBs. Then, we compute FD for each SB. These FDs of the SBs have been applied to the least-squares support vector machine (LS-SVM) classifier with radial basis function (RBF) kernel. We have used 10-fold cross-validation to ensure reliable performance and avoid the possible over-fitting of the model. In the proposed study, we investigate the following four popular classification tasks (CTs) related to different classes of EEG signals: (i) normal versus seizure (ii) seizure-free versus seizure (iii) nonseizure versus Seizure (iv) normal versus seizure-free. The proposed model surpassed existing models in the area under the receiver operating characteristics (ROC) curve, Matthew’s correlation coefficient (MCC), average classification accuracy (ACA), and average classification sensitivity (ACS). The proposed system attained perfect 100% ACS for all CTs considered in this study. The method achieved the highest classification accuracy as well as the largest area under ROC curve (AUC) for all classes. The salient feature of our proposed model is that, though many models exist in the literature, which gave high ACA, however, their performance has not been evaluated using MCC and AUC along with ACA simultaneously. The evaluation of the performance in terms of only ACA which may be misleading. Hence, the performance of the proposed model has been assessed not only in terms of ACA but also in terms AUC and MCC. Moreover, the performance of the model has been found to be almost equivalent to a perfect model, and the performance of the proposed model surpasses the existing models for the CTs investigated by us. Therefore, the proposed model is expected to assist clinicians in analyzing seizures accurately in less time without any error.

Aims: QEEG and neuropsychological tests were used to investigate the underlying neural processes in dyslexia.

Methods: A group of dyslexic children were compared with a matched control group from the Brain Resource International Database on measures of cognition and brain function (EEG and coherence).

Results: The dyslexic group showed increased slow activity (Delta and Theta) in the frontal and right temporal regions of the brain. Beta-1 was specifically increased at F7. EEG coherence was increased in the frontal, central and temporal regions for all frequency bands. There was a symmetric increase in coherence for the lower frequency bands (Delta and Theta) and a specific right-temporocentral increase in coherence for the higher frequency bands (Alpha and Beta). Significant correlations were observed between subtests such as Rapid Naming Letters, Articulation, Spelling and Phoneme Deletion and EEG coherence profiles.

Discussion: The results support the double-deficit theory of dyslexia and demonstrate that the differences between the dyslexia and control group might reflect compensatory mechanisms.

Integrative Significance: These findings point to a potential compensatory mechanism of brain function in dyslexia and helps to separate real dysfunction in dyslexia from acquired compensatory mechanisms.

Schizotypy is a latent cluster of personality traits that denote a vulnerability for schizophrenia or a type of spectrum disorder. The aim of the study is to investigate parametric effective brain connectivity features for classifying high versus low schizotypy (LS) status. Electroencephalography (EEG) signals are recorded from 13 high schizotypy (HS) and 11 LS participants during an emotional auditory odd-ball task. The brain connectivity signals for machine learning are taken after the settlement of event-related potentials. A multivariate autoregressive (MVAR)-based connectivity measure is estimated from the EEG signals using the directed transfer functions (DTFs) method. The values of DTF power in five standard frequency bands are used as features. The support vector machines (SVMs) revealed significant differences between HS and LS. The accuracy, specificity, and sensitivity of the results using SVM are as high as 89.21%, 90.3%, and 88.2%, respectively. Our results demonstrate that the effective brain connectivity in prefrontal/parietal and prefrontal/frontal brain regions considerably changes according to schizotypal status. These findings prove that the brain connectivity indices offer valuable biomarkers for detecting schizotypal personality. Further monitoring of the changes in DTF following the diagnosis of schizotypy may lead to the early identification of schizophrenia and other spectrum disorders.

We propose a simple and effective method of characterizing complexity of EEG-signals for monitoring the depth of anaesthesia using Higuchi's fractal dimension method. We demonstrate that the proposed method may compete with the widely used BIS monitoring method.

Brain–computer interfaces (BCIs) for communication can be nonintuitive, often requiring the performance of hand motor imagery or some other conversation-irrelevant task. In this paper, electroencephalography (EEG) was used to develop two intuitive online BCIs based solely on covert speech. The goal of the first BCI was to differentiate between 10s of mental repetitions of the word “no” and an equivalent duration of unconstrained rest. The second BCI was designed to discern between 10s each of covert repetition of the words “yes” and “no”. Twelve participants used these two BCIs to answer yes or no questions. Each participant completed four sessions, comprising two offline training sessions and two online sessions, one for testing each of the BCIs. With a support vector machine and a combination of spectral and time-frequency features, an average accuracy of $75.9\%\pm 11.4$ was reached across participants in the online classification of no versus rest, with 10 out of 12 participants surpassing the chance level (60.0% for $p<0.05$). The online classification of yes versus no yielded an average accuracy of $69.3\%\pm 14.1$, with eight participants exceeding the chance level. Task-specific changes in EEG beta and gamma power in language-related brain areas tended to provide discriminatory information. To our knowledge, this is the first report of online EEG classification of covert speech. Our findings support further study of covert speech as a BCI activation task, potentially leading to the development of more intuitive BCIs for communication.

Decoding brain intention from noninvasively measured neural signals has recently been a hot topic in brain-computer interface (BCI). The motor commands about the movements of fine parts can increase the degrees of freedom under control and be applied to external equipment without stimulus. In the decoding process, the classifier is one of the key factors, and the graph information of the EEG was ignored by most researchers. In this paper, a graph convolutional network (GCN) based on functional connectivity was proposed to decode the motor intention of four fine parts movements (shoulder, elbow, wrist, hand). First, event-related desynchronization was analyzed to reveal the differences between the four classes. Second, functional connectivity was constructed by using synchronization likelihood (SL), phase-locking value (PLV), H index (H), mutual information (MI), and weighted phase-lag index (WPLI) to acquire the electrode pairs with a difference. Subsequently, a GCN and convolutional neural networks (CNN) were performed based on functional topological structures and time points, respectively. The results demonstrated that the proposed method achieved a decoding accuracy of up to 92.81% in the four-class task. Besides, the combination of GCN and functional connectivity can promote the development of BCI.