Search

Brain–computer interfaces (BCIs) enable direct communication between the brain and external devices using electroencephalography (EEG) signals. BCIs based on code-modulated visual evoked potentials (cVEPs) are based on visual stimuli, thus appropriate visual feedback on the interface is crucial for an effective BCI system. Many previous studies have demonstrated that implementing visual feedback can improve information transfer rate (ITR) and reduce fatigue. This research compares a dynamic interface, where target boxes change their sizes based on detection certainty, with a threshold bar interface in a three-step cVEP speller. In this study, we found that both interfaces perform well, with slight variations in accuracy, ITR, and output characters per minute (OCM). Notably, some participants showed significant performance improvements with the dynamic interface and found it less distracting compared to the threshold bars. These results suggest that while average performance metrics are similar, the dynamic interface can provide significant benefits for certain users. This study underscores the potential for personalized interface choices to enhance BCI user experience and performance. By improving user friendliness, performance, and reducing distraction, dynamic visual feedback could optimize BCI technology for a broader range of users.

Brain-computer interfaces (BCIs) enable direct communication between the brain and external devices using electroencephalography (EEG) signals. BCIs based on code-modulated visual evoked potentials (cVEPs) are based on visual stimuli, thus appropriate visual feedback on the interface is crucial for an effective BCI system. Many previous studies have demonstrated that implementing visual feedback can improve information transfer rate (ITR) and reduce fatigue. This research compares a dynamic interface, where target boxes change their sizes based on detection certainty, with a threshold bar interface in a three-step cVEP speller. In this study, we found that both interfaces perform well, with slight variations in accuracy, ITR, and output characters per minute (OCM). Notably, some participants showed significant performance improvements with the dynamic interface and found it less distracting compared to the threshold bars. These results suggest that while average performance metrics are similar, the dynamic interface can provide significant benefits for certain users. This study underscores the potential for personalized interface choices to enhance BCI user experience and performance. By improving user friendliness, performance, and reducing distraction, dynamic visual feedback could optimize BCI technology for a broader range of users.

An event-related potential (ERP)-based brain–computer interface (BCI) can be used to monitor a user’s cognitive state during a surveillance task in a situational awareness context. The present study explores the use of an ERP-BCI for detecting new planes in an air traffic controller (ATC). Two experiments were conducted to evaluate the impact of different visual factors on target detection. Experiment 1 validated the type of stimulus used and the effect of not knowing its appearance location in an ERP-BCI scenario. Experiment 2 evaluated the effect of the size of the target stimulus appearance area and the stimulus salience in an ATC scenario. The main results demonstrate that the size of the plane appearance area had a negative impact on the detection performance and on the amplitude of the P300 component. Future studies should address this issue to improve the performance of an ATC in stimulus detection using an ERP-BCI.

Convolutional neural networks (CNNs) have shown great potential in the field of brain–computer interfaces (BCIs) due to their ability to directly process raw electroencephalogram (EEG) signals without artificial feature extraction. Some CNNs have achieved better classification accuracy than that of traditional methods. Raw EEG signals are usually represented as a two-dimensional (2-D) matrix composed of channels and time points, ignoring the spatial topological information of electrodes. Our goal is to make a CNN that takes raw EEG signals as inputs have the ability to learn spatial topological features and improve its classification performance while basically maintaining its original structure. We propose an EEG topographic representation module (TRM). This module consists of (1) a mapping block from raw EEG signals to a 3-D topographic map and (2) a convolution block from the topographic map to an output with the same size as the input. According to the size of the convolutional kernel used in the convolution block, we design two types of TRMs, namely TRM-(5,5) and TRM-(3,3). We embed the two TRM types into three widely used CNNs (ShallowConvNet, DeepConvNet and EEGNet) and test them on two publicly available datasets (the Emergency Braking During Simulated Driving Dataset (EBDSDD) and the High Gamma Dataset (HGD)). Results show that the classification accuracies of all three CNNs are improved on both datasets after using the TRMs. With TRM-(5,5), the average classification accuracies of DeepConvNet, EEGNet and ShallowConvNet are improved by 6.54%, 1.72% and 2.07% on the EBDSDD and by 6.05%, 3.02% and 5.14% on the HGD, respectively; with TRM-(3,3), they are improved by 7.76%, 1.71% and 2.17% on the EBDSDD and by 7.61%, 5.06% and 6.28% on the HGD, respectively. We improve the classification performance of three CNNs on both datasets through the use of TRMs, indicating that they have the capability to mine spatial topological EEG information. More importantly, since the output of a TRM has the same size as the input, CNNs with raw EEG signals as inputs can use this module without changing their original structures.

Low transfer rates are a major bottleneck for brain–computer interfaces based on electroencephalography (EEG). This problem has led to the development of more robust and accurate classifiers. In this study, we investigated the performance of variational quantum, quantum-enhanced support vector, and hypergraph case-based reasoning classifiers in the binary classification of EEG data from a P300 experiment. On the one hand, quantum classification is a promising technology to reduce computational time and improve learning outcomes. On the other hand, case-based reasoning has an excellent potential to simplify the preprocessing steps of EEG analysis. We found that the balanced training (prediction) accuracy of each of these three classifiers was 56.95 (51.83), 83.17 (50.25), and 71.10% (52.04%), respectively. In addition, case-based reasoning performed significantly lower with a simplified (49.78%) preprocessing pipeline. These results demonstrated that all classifiers were able to learn from the data and that quantum classification of EEG data was implementable; however, more research is required to enable a greater prediction accuracy because none of the classifiers were able to generalize from the data. This could be achieved by improving the configuration of the quantum classifiers (e.g., increasing the number of shots) and increasing the number of trials for hypergraph case-based reasoning classifiers through transfer learning.

Joint attention skills deficiency in Autism spectrum disorder (ASD) hinders individuals from communicating effectively. The P300 Electroencephalogram (EEG) signal-based brain–computer interface (BCI) helps these individuals in neurorehabilitation training to overcome this deficiency. The detection of the P300 signal is more challenging in ASD as it is noisy, has less amplitude, and has a higher latency than in other individuals. This paper presents a novel application of the variational mode decomposition (VMD) technique in a BCI system involving ASD subjects for P300 signal identification. The EEG signal is decomposed into five modes using VMD. Thirty linear and non-linear time and frequency domain features are extracted for each mode. Synthetic minority oversampling technique data augmentation is performed to overcome the class imbalance problem in the chosen dataset. Then, a comparative analysis of three popular machine learning classifiers is performed for this application. VMD’s fifth mode with a support vector machine (fine Gaussian kernel) classifier gave the best performance parameters, namely accuracy, F1-score, and the area under the curve, as 91.12%, 91.18%, and 96.6%, respectively. These results are better when compared to other state-of-the-art methods.

EEG signals are interpreted, analyzed and classified by many researchers for use in brain–computer interfaces. Although there are many different EEG signal acquisition methods, one of the most interesting is motor imagery signals. Many different signal processing methods, machine learning and deep learning models have been developed for the classification of motor imagery signals. Among these, Convolutional Neural Network models generally achieve better results than other models. Because the size and shape of the data is important for training Convolutional Neural Network models and discovering the right relationships, researchers have designed and experimented with many different input shape structures. However, no study has been found in the literature evaluating the effect of different input shapes on model performance and accuracy. In this study, the effects of different input shapes on model performance and accuracy in the classification of EEG motor imagery signals were investigated, which had not been specifically studied before. In addition, signal preprocessing methods, which take a long time before classification, were not used; rather, two CNN models were developed for training and classification using raw data. Two different datasets, BCI Competition IV 2A and 2B, were used in classification processes. For different input shapes, 53.03–89.29% classification accuracy and 2–23 s epoch time were obtained for 2A dataset, 64.84–84.94% classification accuracy and 4–10 s epoch time were obtained for 2B dataset. This study showed that the input shape has a significant effect on the classification performance, and when the correct input shape is selected and the correct CNN architecture is developed, feature extraction and classification can be done well by the CNN architecture without any signal preprocessing.

In recent years, deep-learning-based motor imagery (MI) electroencephalography (EEG) decoding methods have shown great potential in the field of the brain–computer interface (BCI). The existing literature is relatively mature in decoding methods for two classes of MI tasks. However, with the increase in MI task classes, decoding studies for four classes of MI tasks need to be further explored. In addition, it is difficult to obtain large-scale EEG datasets. When the training data are limited, deep-learning-based decoding models are prone to problems such as overfitting and poor robustness. In this study, we design a data augmentation method for MI-EEG. The original EEG is slid along the time axis and reconstructed to expand the size of the dataset. Second, we combine the gated recurrent unit (GRU) and convolutional neural network (CNN) to construct a parallel-structured feature fusion network to decode four classes of MI tasks. The parallel structure can avoid temporal, frequency and spatial features interfering with each other. Experimenting on the well-known four-class MI dataset BCI Competition IV 2a shows a global average classification accuracy of 80.7% and a kappa value of 0.74. The proposed method improves the robustness of deep learning to decode small-scale EEG datasets and alleviates the overfitting phenomenon caused by insufficient data. The method can be applied to BCI systems with a small amount of daily recorded data.

Brain neural activity decoding is an important branch of neuroscience research and a key technology for the brain–computer interface (BCI). Researchers initially developed simple linear models and machine learning algorithms to classify and recognize brain activities. With the great success of deep learning on image recognition and generation, deep neural networks (DNN) have been engaged in reconstructing visual stimuli from human brain activity via functional magnetic resonance imaging (fMRI). In this paper, we reviewed the brain activity decoding models based on machine learning and deep learning algorithms. Specifically, we focused on current brain activity decoding models with high attention: variational auto-encoder (VAE), generative confrontation network (GAN), and the graph convolutional network (GCN). Furthermore, brain neural-activity-decoding-enabled fMRI-based BCI applications in mental and psychological disease treatment are presented to illustrate the positive correlation between brain decoding and BCI. Finally, existing challenges and future research directions are addressed.

This paper investigates the effects of the repetitive block-wise training process on the classification accuracy for a code-modulated visual evoked potentials (cVEP)-based brain–computer interface (BCI). The cVEP-based BCIs are popular thanks to their autocorrelation feature. The cVEP-based stimuli are generated by a specific code pattern, usually the m-sequence, which is phase-shifted between the individual targets. Typically, the cVEP classification requires a subject-specific template (individually created from the user’s own pre-recorded EEG responses to the same stimulus target), which is compared to the incoming electroencephalography (EEG) data, using the correlation algorithms. The amount of the collected user training data determines the accuracy of the system. In this offline study, previously recorded EEG data collected during an online experiment with 10 participants from multiple sessions were used. A template matching target identification, with similar models as the task-related component analysis (TRCA), was used for target classification. The spatial filter was generated by the canonical correlation analysis (CCA). When comparing the training models from one session with the same session’s data (intra-session) and the model from one session with the data from the other session (inter-session), the accuracies were (94.84%, 94.53%) and (76.67%, 77.34%) for intra-sessions and inter-sessions, respectively. In order to investigate the most reliable configuration for accurate classification, the training data blocks from different sessions (days) were compared interchangeably. In the best training set composition, the participants achieved an average accuracy of 82.66% for models based only on two training blocks from two different sessions. Similarly, at least five blocks were necessary for the average accuracy to exceed 90%. The presented method can further improve cVEP-based BCI performance by reusing previously recorded training data.

This is the first pilot study with children that has assessed the effects of a brain–computer interface-assisted mindfulness program on neural mechanisms and associated cognitive performance. The participants were 31 children aged 9–10 years who were randomly assigned to either an eight-session mindfulness training with EEG-feedback or a passive control group. Mindfulness-related brain activity was measured during the training, while cognitive tests and resting-state brain activity were measured pre- and post-test. The within-group measurement of calm/focused brain states and mind-wandering revealed a significant linear change. Significant positive changes were detected in children’s inhibition, information processing, and resting-state brain activity (alpha, theta) compared to the control group. Elevated baseline alpha activity was associated with less reactivity in reaction time on a cognitive test. Our exploratory findings show some preliminary support for a potential executive function-enhancing effect of mindfulness supplemented with EEG-feedback, which may have some important implications for children’s self-regulated learning and academic achievement.

The ability to control external devices through thought is increasingly becoming a reality. Human beings can use the electrical signals of their brain to interact or change the surrounding environment and more. The development of this technology called brain-computer interface (BCI) will increasingly allow people with motor disabilities to communicate or use assistive devices to walk, manipulate objects and communicate. Using data from the PhysioNet database, this study implemented a pattern classification system for use in a BCI on 109 healthy volunteers during real movement activities and motor imagery recorded by 64-channels electroencephalography (EEG) system. Different classifiers such as Support Vector Machine (SVM), K-Nearest Neighbors (KNN), and Decision Trees (TREE) were applied on different combinations of EEG channels. Starting from two channels (C3, C4 and CP3 and CP4) positioned on the contralateral and ipsilateral sensorimotor cortex, the Region of Interest (RoI) centred on C3/Cp3 and C4/Cp4 and, finally, a data-driven automatic channels selection was tested to explore the best channel combination able to increase the classification accuracy. The results showed that the proposed automatic channels selection was able to significantly improve the performance of each classifier achieving 98% of accuracy for classification of real and imagined hand movement (sensitivity = 97%, specificity = 99%, AUC = 0.99) by SVM. While the accuracy of the classification between the imagery of hand and foot movements was 91% (sensitivity = 87%, specificity = 86%, AUC = 0.93) also with SVM. In the proposed approach, the data-driven automatic channels selection outperforms classical a priori channel selection models such as C3/C4, Cp3/Cp4, or RoIs around those channels with the utmost accuracy to help remove the boundaries of human communication and improve the quality of life of people with disabilities.

The steady-state visual evoked potential (SSVEP), measured by the electroencephalograph (EEG), has high rates of information transfer and signal-to-noise ratio, and has been used to construct brain–computer interface (BCI) spellers. In BCI spellers, the targets of alphanumeric characters are assigned different visual stimuli and the fixation of each target generates a unique SSVEP. Matching the SSVEP to the stimulus allows users to select target letters and numbers. Many BCI spellers that harness the SSVEP have been proposed over the past two decades. Various paradigms of visual stimuli, including the procedure of target selection, layout of targets, stimulus encoding, and the combination with other triggering methods are used and considered to influence on the BCI speller performance significantly. This paper reviews these stimulus paradigms and analyzes factors influencing their performance. The fundamentals of BCI spellers are first briefly described. SSVEP-based BCI spellers, where only the SSVEP is used, are classified by stimulus paradigms and described in chronological order. Furthermore, hybrid spellers that involve the use of the SSVEP are presented in parallel. Factors influencing the performance and visual fatigue of BCI spellers are provided. Finally, prevailing challenges and prospective research directions are discussed to promote the development of BCI spellers.

Studies so far have analyzed the effect of distractor stimuli in different types of brain–computer interface (BCI). However, the effect of a background speech has not been studied using an auditory event-related potential (ERP-BCI), a convenient option when the visual path cannot be adopted by users. Thus, the aim of the present work is to examine the impact of a background speech on selection performance and user workload in auditory BCI systems. Eleven participants tested three conditions: (i) auditory BCI control condition, (ii) auditory BCI with a background speech to ignore (non-attentional condition), and (iii) auditory BCI while the user has to pay attention to the background speech (attentional condition). The results demonstrated that, despite no significant differences in performance, shared attention to auditory BCI and background speech required a higher cognitive workload. In addition, the P300 target stimuli in the non-attentional condition were significantly higher than those in the attentional condition for several channels. The non-attentional condition was the only condition that showed significant differences in the amplitude of the P300 between target and non-target stimuli. The present study indicates that background speech, especially when it is attended to, is an important interference that should be avoided while using an auditory BCI.

Background: Brain-Computer Interface (BCI) is becoming more reliable, thanks to the advantages of Artificial Intelligence (AI). Recently, hybrid Deep Learning (hDL), which combines different DL algorithms, has gained momentum over the past five years. In this work, we proposed a review on hDL-based BCI starting from the seminal studies in 2015. Objectives: We have reviewed 47 papers that apply hDL to the BCI system published between 2015 and 2020 extracting trends and highlighting relevant aspects to the topic. Methods: We have queried four scientific search engines (Google Scholar, PubMed, IEEE Xplore and Elsevier Science Direct) and different data items were extracted from each paper such as the database used, kind of application, online/offline training, tasks used for the BCI, pre-processing methodology adopted, type of normalization used, which kind of features were extracted, type of DL architecture used, number of layers implemented and which optimization approach were used as well. All these items were then investigated one by one to uncover trends. Results: Our investigation reveals that Electroencephalography (EEG) has been the most used technique. Interestingly, despite the lower Signal-to-Noise Ratio (SNR) of the EEG data that makes pre-processing of that data mandatory, we have found that the pre-processing has only been used in 21.28% of the cases by showing that hDL seems to be able to overcome this intrinsic drawback of the EEG data. Temporal-features seem to be the most effective with 93.94% accuracy, while spatial-temporal features are the most used with 33.33% of the cases investigated. The most used architecture has been Convolutional Neural Network-Recurrent Neural Network CNN-RNN with 47% of the cases. Moreover, half of the studies have used a low number of layers to achieve a good compromise between the complexity of the network and computational efficiency. Significance: To give useful information to the scientific community, we make our summary table of hDL-based BCI papers available and invite the community to published work to contribute to it directly. We have indicated a list of open challenges, emphasizing the need to use neuroimaging techniques other than EEG, such as functional Near-Infrared Spectroscopy (fNIRS), deeper investigate the advantages and disadvantages of using pre-processing and the relationship with the accuracy obtained. To implement new combinations of architectures, such as RNN-based and Deep Belief Network DBN-based, it is necessary to better explore the frequency and temporal-frequency features of the data at hand.

Motor deficiencies constitute a significant problem affecting millions of people worldwide. Such people suffer from a debility in daily functioning, which may lead to decreased and incoherence in daily routines and deteriorate their quality of life (QoL). Thus, there is an essential need for assistive systems to help those people achieve their daily actions and enhance their overall QoL. This study proposes a novel brain–computer interface (BCI) system for assisting people with limb motor disabilities in performing their daily life activities by using their brain signals to control assistive devices. The extraction of useful features is vital for an efficient BCI system. Therefore, the proposed system consists of a hybrid feature set that feeds into three machine-learning (ML) classifiers to classify motor Imagery (MI) tasks. This hybrid feature selection (FS) system is practical, real-time, and an efficient BCI with low computation cost. We investigate different combinations of channels to select the combination that has the highest impact on performance. The results indicate that the highest achieved accuracies using a support vector machine (SVM) classifier are 93.46% and 86.0% for the BCI competition III–IVa dataset and the autocalibration and recurrent adaptation dataset, respectively. These datasets are used to test the performance of the proposed BCI. Also, we verify the effectiveness of the proposed BCI by comparing its performance with recent studies. We show that the proposed system is accurate and efficient. Future work can apply the proposed system to individuals with limb motor disabilities to assist them and test their capability to improve their QoL. Moreover, the forthcoming work can examine the system’s performance in controlling assistive devices such as wheelchairs or artificial limbs.

Brain–computer interface (BCI) technology provides a direct interface between the brain and an external device. BCIs have facilitated the monitoring of conscious brain electrical activity via electroencephalogram (EEG) signals and the detection of human emotion. Recently, great progress has been made in the development of novel paradigms for EEG-based emotion detection. These studies have also attempted to apply BCI research findings in varied contexts. Interestingly, advances in BCI technologies have increased the interest of scientists because such technologies’ practical applications in human–machine relationships seem promising. This emphasizes the need for a building process for an EEG-based emotion detection system that is lightweight, in terms of a smaller EEG dataset size and no involvement of feature extraction methods. In this study, we investigated the feasibility of using a spiking neural network to build an emotion detection system from a smaller version of the DEAP dataset with no involvement of feature extraction methods while maintaining decent accuracy. The results showed that by using a NeuCube-based spiking neural network, we could detect the valence emotion level using only 60 EEG samples with 84.62% accuracy, which is a comparable accuracy to that of previous studies.

With the continuous development of portable noninvasive human sensor technologies such as brain–computer interfaces (BCI), multimodal emotion recognition has attracted increasing attention in the area of affective computing. This paper primarily discusses the progress of research into multimodal emotion recognition based on BCI and reviews three types of multimodal affective BCI (aBCI): aBCI based on a combination of behavior and brain signals, aBCI based on various hybrid neurophysiology modalities and aBCI based on heterogeneous sensory stimuli. For each type of aBCI, we further review several representative multimodal aBCI systems, including their design principles, paradigms, algorithms, experimental results and corresponding advantages. Finally, we identify several important issues and research directions for multimodal emotion recognition based on BCI.

Motion-based visual evoked potentials (mVEP) is a new emerging trend in the field of steady-state visual evoked potentials (SSVEP)-based brain–computer interfaces (BCI). In this paper, we introduce different movement-based stimulus patterns (steady-state motion visual evoked potentials—SSMVEP), without employing the typical flickering. The tested movement patterns for the visual stimuli included a pendulum-like movement, a flipping illusion, a checkerboard pulsation, checkerboard inverse arc pulsations, and reverse arc rotations, all with a spelling task consisting of 18 trials. In an online experiment with nine participants, the movement-based BCI systems were evaluated with an online four-target BCI-speller, in which each letter may be selected in three steps (three trials). For classification, the minimum energy combination and a filter bank approach were used. The following frequencies were utilized: 7.06 Hz, 7.50 Hz, 8.00 Hz, and 8.57 Hz, reaching an average accuracy between 97.22% and 100% and an average information transfer rate (ITR) between 15.42 bits/min and 33.92 bits/min. All participants successfully used the SSMVEP-based speller with all types of stimulation pattern. The most successful SSMVEP stimulus was the SSMVEP1 (pendulum-like movement), with the average results reaching 100% accuracy and 33.92 bits/min for the ITR.

Among brain-computer interface studies, electroencephalography (EEG)-based emotion recognition is receiving attention and some studies have performed regression analyses to recognize small-scale emotional changes; however, effective brain regions in emotion regression analyses have not been identified yet. Accordingly, this study sought to identify neural activities correlating with emotional states in the source space. We employed independent component analysis, followed by a source localization method, to obtain distinct neural activities from EEG signals. After the identification of seven independent component (IC) clusters in a k-means clustering analysis, group-level regression analyses using frequency band power of the ICs were performed based on Russell’s valence–arousal model. As a result, in the regression of the valence level, an IC cluster located in the cuneus predicted both high- and low-valence states and two other IC clusters located in the left precentral gyrus and the precuneus predicted the low-valence state. In the regression of the arousal level, the IC cluster located in the cuneus predicted both high- and low-arousal states and two posterior IC clusters located in the cingulate gyrus and the precuneus predicted the high-arousal state. In this proof-of-concept study, we revealed neural activities correlating with specific emotional states across participants, despite individual differences in emotional processing.

Sensorimotor rhythm (SMR)-based brain–computer interface (BCI) controlled Functional Electrical Stimulation (FES) has gained importance in recent years for the rehabilitation of motor deficits. However, there still remain many research questions to be addressed, such as unstructured Motor Imagery (MI) training procedures; a lack of methods to classify different MI tasks in a single hand, such as grasping and opening; and difficulty in decoding voluntary MI-evoked SMRs compared to FES-driven passive-movement-evoked SMRs. To address these issues, a study that is composed of two phases was conducted to develop and validate an SMR-based BCI-FES system with 2-class MI tasks in a single hand (Phase 1), and investigate the feasibility of the system with stroke and traumatic brain injury (TBI) patients (Phase 2). The results of Phase 1 showed that the accuracy of classifying 2-class MIs (approximately 71.25%) was significantly higher than the true chance level, while that of distinguishing voluntary and passive SMRs was not. In Phase 2, where the patients performed goal-oriented tasks in a semi-asynchronous mode, the effects of the FES existence type and adaptive learning on task performance were evaluated. The results showed that adaptive learning significantly increased the accuracy, and the accuracy after applying adaptive learning under the No-FES condition (61.9%) was significantly higher than the true chance level. The outcomes of the present research would provide insight into SMR-based BCI-controlled FES systems that can connect those with motor disabilities (e.g., stroke and TBI patients) to other people by greatly improving their quality of life. Recommendations for future work with a larger sample size and kinesthetic MI were also presented.

In this paper we compared different visual feedback methods, informing users about classification progress in a steady-state visual evoked potential (SSVEP)-based brain–computer interface (BCI) speller application. According to results from our previous studies, changes in stimulus size and contrast as online feedback of classification progress have great impact on BCI performance in SSVEP-based spellers. In this experiment we further investigated these effects, and tested a 4-target SSVEP speller interface with a much higher number of subjects. Five different scenarios were used with variations in stimulus size and contrast, “no feedback”, “size increasing”, “size decreasing”, “contrast increasing”, and “contrast decreasing”. With each of the five scenarios, 24 participants had to spell six letter words (at least 18 selections with this three-steps speller). The fastest feedback modalities were different for the users, there was no visual feedback which was generally better than the others. With the used interface, six users achieved significantly better Information Transfer Rates (ITRs) compared to the “no feedback” condition. Their average improvement by using the individually fastest feedback method was 46.52%. This finding is very important for BCI experiments, as by determining the optimal feedback for the user, the speed of the BCI can be improved without impairing the accuracy.

Human facial expressions are regarded as a vital indicator of one’s emotion and intention, and even reveal the state of health and wellbeing. Emotional states have been associated with information processing within and between subcortical and cortical areas of the brain, including the amygdala and prefrontal cortex. In this study, we evaluated the relationship between spontaneous human facial affective expressions and multi-modal brain activity measured via non-invasive and wearable sensors: functional near-infrared spectroscopy (fNIRS) and electroencephalography (EEG) signals. The affective states of twelve male participants detected via fNIRS, EEG, and spontaneous facial expressions were investigated in response to both image-content stimuli and video-content stimuli. We propose a method to jointly evaluate fNIRS and EEG signals for affective state detection (emotional valence as positive or negative). Experimental results reveal a strong correlation between spontaneous facial affective expressions and the perceived emotional valence. Moreover, the affective states were estimated by the fNIRS, EEG, and fNIRS + EEG brain activity measurements. We show that the proposed EEG + fNIRS hybrid method outperforms fNIRS-only and EEG-only approaches. Our findings indicate that the dynamic (video-content based) stimuli triggers a larger affective response than the static (image-content based) stimuli. These findings also suggest joint utilization of facial expression and wearable neuroimaging, fNIRS, and EEG, for improved emotional analysis and affective brain−computer interface applications.

An event-related potential (ERP)-based brain-computer interface (BCI) can be used to monitor a user's cognitive state during a surveillance task in a situational awareness context. The present study explores the use of an ERP-BCI for detecting new planes in an air traffic controller (ATC). Two experiments were conducted to evaluate the impact of different visual factors on target detection. Experiment 1 validated the type of stimulus used and the effect of not knowing its appearance location in an ERP-BCI scenario. Experiment 2 evaluated the effect of the size of the target stimulus appearance area and the stimulus salience in an ATC scenario. The main results demonstrate that the size of the plane appearance area had a negative impact on the detection performance and on the amplitude of the P300 component. Future studies should address this issue to improve the performance of an ATC in stimulus detection using an ERP-BCI.

Tactile attention tasks are used in the diagnosis and treatment of neurological and sensory processing disorders, while somatosensory event-related potentials (ERP) measured by electroencephalography (EEG) are used as neural correlates of attention processes. Brain-computer interface (BCI) technology provides an opportunity for the training of mental task execution via providing online feedback based on ERP measures. Our recent work introduced a novel electrotactile BCI for sensory training, based on somatosensory ERP; however, no previous studies have addressed specific somatosensory ERP morphological features as measures of sustained endogenous spatial tactile attention in the context of BCI control. Here we show the morphology of somatosensory ERP responses induced by a novel task introduced within our electrotactile BCI platform i.e., the sustained endogenous spatial electrotactile attention task. By applying pulsed electrical stimuli to the two proximal stimulation hotspots at the user's forearm, stimulating sequentially the mixed branches of radial and median nerves with equal probability of stimuli occurrence, we successfully recorded somatosensory ERPs for both stimulation locations, in the attended and unattended conditions. Waveforms of somatosensory ERP responses for both mixed nerve branches showed similar morphology in line with previous reports on somatosensory ERP components obtained by stimulation of exclusively sensory nerves. Moreover, we found statistically significant increases in ERP amplitude on several components, at both stimulation hotspots, while sustained endogenous spatial electrotactile attention task is performed. Our results revealed the existence of general ERP windows of interest and signal features that can be used to detect sustained endogenous tactile attention and classify between spatial attention locations in 11 healthy subjects. The current results show that features of N140, P3a and P3b somatosensory ERP components are the most prominent global markers of sustained spatial electrotactile attention, over all subjects, within our novel electrotactile BCI task/paradigm, and this work proposes the features of those components as markers of sustained endogenous spatial tactile attention in online BCI control. Immediate implications of this work are the possible improvement of online BCI control within our novel electrotactile BCI system, while these finding can be used for other tactile BCI applications in the diagnosis and treatment of neurological disorders by employing mixed nerve somatosensory ERPs and sustained endogenous electrotactile attention task as control paradigms.

A Brain–Computer Interface (BCI) provides a novel non-muscular communication method via brain signals. A BCI-speller can be considered as one of the first published BCI applications and has opened the gate for many advances in the field. Although many BCI-spellers have been developed during the last few decades, to our knowledge, no reviews have described the different spellers proposed and studied in this vital field. The presented speller systems are categorized according to major BCI paradigms: P300, steady-state visual evoked potential (SSVEP), and motor imagery (MI). Different BCI paradigms require specific electroencephalogram (EEG) signal features and lead to the development of appropriate Graphical User Interfaces (GUIs). The purpose of this review is to consolidate the most successful BCI-spellers published since 2010, while mentioning some other older systems which were built explicitly for spelling purposes. We aim to assist researchers and concerned individuals in the field by illustrating the highlights of different spellers and presenting them in one review. It is almost impossible to carry out an objective comparison between different spellers, as each has its variables, parameters, and conditions. However, the gathered information and the provided taxonomy about different BCI-spellers can be helpful, as it could identify suitable systems for first-hand users, as well as opportunities of development and learning from previous studies for BCI researchers.

Joint attention skills deficiency in Autism spectrum disorder (ASD) hinders individuals from communicating effectively. The P300 Electroencephalogram (EEG) signal-based brain-computer interface (BCI) helps these individuals in neurorehabilitation training to overcome this deficiency. The detection of the P300 signal is more challenging in ASD as it is noisy, has less amplitude, and has a higher latency than in other individuals. This paper presents a novel application of the variational mode decomposition (VMD) technique in a BCI system involving ASD subjects for P300 signal identification. The EEG signal is decomposed into five modes using VMD. Thirty linear and non-linear time and frequency domain features are extracted for each mode. Synthetic minority oversampling technique data augmentation is performed to overcome the class imbalance problem in the chosen dataset. Then, a comparative analysis of three popular machine learning classifiers is performed for this application. VMD's fifth mode with a support vector machine (fine Gaussian kernel) classifier gave the best performance parameters, namely accuracy, F1-score, and the area under the curve, as 91.12%, 91.18%, and 96.6%, respectively. These results are better when compared to other state-of-the-art methods.

Low transfer rates are a major bottleneck for brain-computer interfaces based on electroencephalography (EEG). This problem has led to the development of more robust and accurate classifiers. In this study, we investigated the performance of variational quantum, quantum-enhanced support vector, and hypergraph case-based reasoning classifiers in the binary classification of EEG data from a P300 experiment. On the one hand, quantum classification is a promising technology to reduce computational time and improve learning outcomes. On the other hand, case-based reasoning has an excellent potential to simplify the preprocessing steps of EEG analysis. We found that the balanced training (prediction) accuracy of each of these three classifiers was 56.95 (51.83), 83.17 (50.25), and 71.10% (52.04%), respectively. In addition, case-based reasoning performed significantly lower with a simplified (49.78%) preprocessing pipeline. These results demonstrated that all classifiers were able to learn from the data and that quantum classification of EEG data was implementable; however, more research is required to enable a greater prediction accuracy because none of the classifiers were able to generalize from the data. This could be achieved by improving the configuration of the quantum classifiers (e.g., increasing the number of shots) and increasing the number of trials for hypergraph case-based reasoning classifiers through transfer learning., Competing Interests: The authors declare no conflict of interest.

Convolutional neural networks (CNNs) have shown great potential in the field of brain-computer interfaces (BCIs) due to their ability to directly process raw electroencephalogram (EEG) signals without artificial feature extraction. Some CNNs have achieved better classification accuracy than that of traditional methods. Raw EEG signals are usually represented as a two-dimensional (2-D) matrix composed of channels and time points, ignoring the spatial topological information of electrodes. Our goal is to make a CNN that takes raw EEG signals as inputs have the ability to learn spatial topological features and improve its classification performance while basically maintaining its original structure. We propose an EEG topographic representation module (TRM). This module consists of (1) a mapping block from raw EEG signals to a 3-D topographic map and (2) a convolution block from the topographic map to an output with the same size as the input. According to the size of the convolutional kernel used in the convolution block, we design two types of TRMs, namely TRM-(5,5) and TRM-(3,3). We embed the two TRM types into three widely used CNNs (ShallowConvNet, DeepConvNet and EEGNet) and test them on two publicly available datasets (the Emergency Braking During Simulated Driving Dataset (EBDSDD) and the High Gamma Dataset (HGD)). Results show that the classification accuracies of all three CNNs are improved on both datasets after using the TRMs. With TRM-(5,5), the average classification accuracies of DeepConvNet, EEGNet and ShallowConvNet are improved by 6.54%, 1.72% and 2.07% on the EBDSDD and by 6.05%, 3.02% and 5.14% on the HGD, respectively; with TRM-(3,3), they are improved by 7.76%, 1.71% and 2.17% on the EBDSDD and by 7.61%, 5.06% and 6.28% on the HGD, respectively. We improve the classification performance of three CNNs on both datasets through the use of TRMs, indicating that they have the capability to mine spatial topological EEG information. More importantly, since the output of a TRM has the same size as the input, CNNs with raw EEG signals as inputs can use this module without changing their original structures.

Steady state visual evoked potentials (SSVEPs)-based Brain-Computer interfaces (BCIs), as well as eyetracking devices, provide a pathway for re-establishing communication for people with severe disabilities. We fused these control techniques into a novel eyetracking/SSVEP hybrid system, which utilizes eye tracking for initial rough selection and the SSVEP technology for fine target activation. Based on our previous studies, only four stimuli were used for the SSVEP aspect, granting sufficient control for most BCI users. As Eye tracking data is not used for activation of letters, false positives due to inappropriate dwell times are avoided. This novel approach combines the high speed of eye tracking systems and the high classification accuracies of low target SSVEP-based BCIs, leading to an optimal combination of both methods. We evaluated accuracy and speed of the proposed hybrid system with a 30-target spelling application implementing all three control approaches (pure eye tracking, SSVEP and the hybrid system) with 32 participants. Although the highest information transfer rates (ITRs) were achieved with pure eye tracking, a considerable amount of subjects was not able to gain sufficient control over the stand-alone eye-tracking device or the pure SSVEP system (78.13% and 75% of the participants reached reliable control, respectively). In this respect, the proposed hybrid was most universal (over 90% of users achieved reliable control), and outperformed the pure SSVEP system in terms of speed and user friendliness. The presented hybrid system might offer communication to a wider range of users in comparison to the standard techniques.

EEG signals are interpreted, analyzed and classified by many researchers for use in brain-computer interfaces. Although there are many different EEG signal acquisition methods, one of the most interesting is motor imagery signals. Many different signal processing methods, machine learning and deep learning models have been developed for the classification of motor imagery signals. Among these, Convolutional Neural Network models generally achieve better results than other models. Because the size and shape of the data is important for training Convolutional Neural Network models and discovering the right relationships, researchers have designed and experimented with many different input shape structures. However, no study has been found in the literature evaluating the effect of different input shapes on model performance and accuracy. In this study, the effects of different input shapes on model performance and accuracy in the classification of EEG motor imagery signals were investigated, which had not been specifically studied before. In addition, signal preprocessing methods, which take a long time before classification, were not used; rather, two CNN models were developed for training and classification using raw data. Two different datasets, BCI Competition IV 2A and 2B, were used in classification processes. For different input shapes, 53.03-89.29% classification accuracy and 2-23 s epoch time were obtained for 2A dataset, 64.84-84.94% classification accuracy and 4-10 s epoch time were obtained for 2B dataset. This study showed that the input shape has a significant effect on the classification performance, and when the correct input shape is selected and the correct CNN architecture is developed, feature extraction and classification can be done well by the CNN architecture without any signal preprocessing.

Over the last few decades, brain signals have been significantly exploited for brain-computer interface (BCI) applications. In this paper, we study the extraction of features using event-related desynchronization/synchronization techniques to improve the classification accuracy for three-class motor imagery (MI) BCI. The classification approach is based on combining the features of the phase and amplitude of the brain signals using fast Fourier transform (FFT) and autoregressive (AR) modeling of the reconstructed phase space as well as the modification of the BCI parameters (trial length, trial frequency band, classification method). We report interesting results compared with those present in the literature by utilizing sequential forward floating selection (SFFS) and a multi-class linear discriminant analysis (LDA), our findings showed superior classification results, a classification accuracy of 86.06% and 93% for two BCI competition datasets, with respect to results from previous studies.

In recent years, deep-learning-based motor imagery (MI) electroencephalography (EEG) decoding methods have shown great potential in the field of the brain-computer interface (BCI). The existing literature is relatively mature in decoding methods for two classes of MI tasks. However, with the increase in MI task classes, decoding studies for four classes of MI tasks need to be further explored. In addition, it is difficult to obtain large-scale EEG datasets. When the training data are limited, deep-learning-based decoding models are prone to problems such as overfitting and poor robustness. In this study, we design a data augmentation method for MI-EEG. The original EEG is slid along the time axis and reconstructed to expand the size of the dataset. Second, we combine the gated recurrent unit (GRU) and convolutional neural network (CNN) to construct a parallel-structured feature fusion network to decode four classes of MI tasks. The parallel structure can avoid temporal, frequency and spatial features interfering with each other. Experimenting on the well-known four-class MI dataset BCI Competition IV 2a shows a global average classification accuracy of 80.7% and a kappa value of 0.74. The proposed method improves the robustness of deep learning to decode small-scale EEG datasets and alleviates the overfitting phenomenon caused by insufficient data. The method can be applied to BCI systems with a small amount of daily recorded data.

This paper investigates the effects of the repetitive block-wise training process on the classification accuracy for a code-modulated visual evoked potentials (cVEP)-based brain-computer interface (BCI). The cVEP-based BCIs are popular thanks to their autocorrelation feature. The cVEP-based stimuli are generated by a specific code pattern, usually the m-sequence, which is phase-shifted between the individual targets. Typically, the cVEP classification requires a subject-specific template (individually created from the user's own pre-recorded EEG responses to the same stimulus target), which is compared to the incoming electroencephalography (EEG) data, using the correlation algorithms. The amount of the collected user training data determines the accuracy of the system. In this offline study, previously recorded EEG data collected during an online experiment with 10 participants from multiple sessions were used. A template matching target identification, with similar models as the task-related component analysis (TRCA), was used for target classification. The spatial filter was generated by the canonical correlation analysis (CCA). When comparing the training models from one session with the same session's data (intra-session) and the model from one session with the data from the other session (inter-session), the accuracies were (94.84%, 94.53%) and (76.67%, 77.34%) for intra-sessions and inter-sessions, respectively. In order to investigate the most reliable configuration for accurate classification, the training data blocks from different sessions (days) were compared interchangeably. In the best training set composition, the participants achieved an average accuracy of 82.66% for models based only on two training blocks from two different sessions. Similarly, at least five blocks were necessary for the average accuracy to exceed 90%. The presented method can further improve cVEP-based BCI performance by reusing previously recorded training data.

Brain neural activity decoding is an important branch of neuroscience research and a key technology for the brain-computer interface (BCI). Researchers initially developed simple linear models and machine learning algorithms to classify and recognize brain activities. With the great success of deep learning on image recognition and generation, deep neural networks (DNN) have been engaged in reconstructing visual stimuli from human brain activity via functional magnetic resonance imaging (fMRI). In this paper, we reviewed the brain activity decoding models based on machine learning and deep learning algorithms. Specifically, we focused on current brain activity decoding models with high attention: variational auto-encoder (VAE), generative confrontation network (GAN), and the graph convolutional network (GCN). Furthermore, brain neural-activity-decoding-enabled fMRI-based BCI applications in mental and psychological disease treatment are presented to illustrate the positive correlation between brain decoding and BCI. Finally, existing challenges and future research directions are addressed.

This is the first pilot study with children that has assessed the effects of a brain-computer interface-assisted mindfulness program on neural mechanisms and associated cognitive performance. The participants were 31 children aged 9-10 years who were randomly assigned to either an eight-session mindfulness training with EEG-feedback or a passive control group. Mindfulness-related brain activity was measured during the training, while cognitive tests and resting-state brain activity were measured pre- and post-test. The within-group measurement of calm/focused brain states and mind-wandering revealed a significant linear change. Significant positive changes were detected in children's inhibition, information processing, and resting-state brain activity (alpha, theta) compared to the control group. Elevated baseline alpha activity was associated with less reactivity in reaction time on a cognitive test. Our exploratory findings show some preliminary support for a potential executive function-enhancing effect of mindfulness supplemented with EEG-feedback, which may have some important implications for children's self-regulated learning and academic achievement.

The ability to control external devices through thought is increasingly becoming a reality. Human beings can use the electrical signals of their brain to interact or change the surrounding environment and more. The development of this technology called brain-computer interface (BCI) will increasingly allow people with motor disabilities to communicate or use assistive devices to walk, manipulate objects and communicate. Using data from the PhysioNet database, this study implemented a pattern classification system for use in a BCI on 109 healthy volunteers during real movement activities and motor imagery recorded by 64-channels electroencephalography (EEG) system. Different classifiers such as Support Vector Machine (SVM), K-Nearest Neighbors (KNN), and Decision Trees (TREE) were applied on different combinations of EEG channels. Starting from two channels (C3, C4 and CP3 and CP4) positioned on the contralateral and ipsilateral sensorimotor cortex, the Region of Interest (RoI) centred on C3/Cp3 and C4/Cp4 and, finally, a data-driven automatic channels selection was tested to explore the best channel combination able to increase the classification accuracy. The results showed that the proposed automatic channels selection was able to significantly improve the performance of each classifier achieving 98% of accuracy for classification of real and imagined hand movement (sensitivity = 97%, specificity = 99%, AUC = 0.99) by SVM. While the accuracy of the classification between the imagery of hand and foot movements was 91% (sensitivity = 87%, specificity = 86%, AUC = 0.93) also with SVM. In the proposed approach, the data-driven automatic channels selection outperforms classical a priori channel selection models such as C3/C4, Cp3/Cp4, or RoIs around those channels with the utmost accuracy to help remove the boundaries of human communication and improve the quality of life of people with disabilities.

The steady-state visual evoked potential (SSVEP), measured by the electroencephalograph (EEG), has high rates of information transfer and signal-to-noise ratio, and has been used to construct brain-computer interface (BCI) spellers. In BCI spellers, the targets of alphanumeric characters are assigned different visual stimuli and the fixation of each target generates a unique SSVEP. Matching the SSVEP to the stimulus allows users to select target letters and numbers. Many BCI spellers that harness the SSVEP have been proposed over the past two decades. Various paradigms of visual stimuli, including the procedure of target selection, layout of targets, stimulus encoding, and the combination with other triggering methods are used and considered to influence on the BCI speller performance significantly. This paper reviews these stimulus paradigms and analyzes factors influencing their performance. The fundamentals of BCI spellers are first briefly described. SSVEP-based BCI spellers, where only the SSVEP is used, are classified by stimulus paradigms and described in chronological order. Furthermore, hybrid spellers that involve the use of the SSVEP are presented in parallel. Factors influencing the performance and visual fatigue of BCI spellers are provided. Finally, prevailing challenges and prospective research directions are discussed to promote the development of BCI spellers.

Background: Brain-Computer Interface (BCI) is becoming more reliable, thanks to the advantages of Artificial Intelligence (AI). Recently, hybrid Deep Learning (hDL), which combines different DL algorithms, has gained momentum over the past five years. In this work, we proposed a review on hDL-based BCI starting from the seminal studies in 2015., Objectives: We have reviewed 47 papers that apply hDL to the BCI system published between 2015 and 2020 extracting trends and highlighting relevant aspects to the topic., Methods: We have queried four scientific search engines (Google Scholar, PubMed, IEEE Xplore and Elsevier Science Direct) and different data items were extracted from each paper such as the database used, kind of application, online/offline training, tasks used for the BCI, pre-processing methodology adopted, type of normalization used, which kind of features were extracted, type of DL architecture used, number of layers implemented and which optimization approach were used as well. All these items were then investigated one by one to uncover trends., Results: Our investigation reveals that Electroencephalography (EEG) has been the most used technique. Interestingly, despite the lower Signal-to-Noise Ratio (SNR) of the EEG data that makes pre-processing of that data mandatory, we have found that the pre-processing has only been used in 21.28% of the cases by showing that hDL seems to be able to overcome this intrinsic drawback of the EEG data. Temporal-features seem to be the most effective with 93.94% accuracy, while spatial-temporal features are the most used with 33.33% of the cases investigated. The most used architecture has been Convolutional Neural Network-Recurrent Neural Network CNN-RNN with 47% of the cases. Moreover, half of the studies have used a low number of layers to achieve a good compromise between the complexity of the network and computational efficiency., Significance: To give useful information to the scientific community, we make our summary table of hDL-based BCI papers available and invite the community to published work to contribute to it directly. We have indicated a list of open challenges, emphasizing the need to use neuroimaging techniques other than EEG, such as functional Near-Infrared Spectroscopy (fNIRS), deeper investigate the advantages and disadvantages of using pre-processing and the relationship with the accuracy obtained. To implement new combinations of architectures, such as RNN-based and Deep Belief Network DBN-based, it is necessary to better explore the frequency and temporal-frequency features of the data at hand.