Several inaccurate or erroneous conceptions and misleading propaganda about brain-computer interfaces.

Brain-computer interface (BCI) is a revolutionizing human-computer interaction, which has potential applications for specific individuals or groups in specific scenarios. Extensive research has been conducted on the principles and implementation methods of BCI, and efforts are currently being made to bridge the gap from research to real-world applications. However, there are inaccurate or erroneous conceptions about BCI among some members of the public, and certain media outlets, as well as some BCI researchers, developers, manufacturers, and regulators, propagate misleading or overhyped claims about BCI technology. Therefore, this article summarizes the several misconceptions and misleading propaganda about BCI, including BCI being capable of "mind-controlled," "controlling brain," "mind reading," and the ability to "download" or "upload" information from or to the brain using BCI, among others. Finally, the limitations (shortcomings) and limits (boundaries) of BCI, as well as the necessity of conducting research aimed at countering BCI systems are discussed, and several suggestions are offered to reduce misconceptions and misleading claims about BCI.

Differential Brain Activation for Four Emotions in VR-2D and VR-3D Modes.

Similar to traditional imaging, virtual reality (VR) imagery encompasses nonstereoscopic (VR-2D) and stereoscopic (VR-3D) modes. Currently, Russell's emotional model has been extensively studied in traditional 2D and VR-3D modes, but there is limited comparative research between VR-2D and VR-3D modes. In this study, we investigate whether Russell's emotional model exhibits stronger brain activation states in VR-3D mode compared to VR-2D mode. By designing an experiment covering four emotional categories (high arousal-high pleasure (HAHV), high arousal-low pleasure (HALV), low arousal-low pleasure (LALV), and low arousal-high pleasure (LAHV)), EEG signals were collected from 30 healthy undergraduate and graduate students while watching videos in both VR modes. Initially, power spectral density (PSD) computations revealed distinct brain activation patterns in different emotional states across the two modes, with VR-3D videos inducing significantly higher brainwave energy, primarily in the frontal, temporal, and occipital regions. Subsequently, Differential entropy (DE) feature sets, selected via a dual ten-fold cross-validation Support Vector Machine (SVM) classifier, demonstrate satisfactory classification accuracy, particularly superior in the VR-3D mode. The paper subsequently presents a deep learning-based EEG emotion recognition framework, adeptly utilizing the frequency, spatial, and temporal information of EEG data to improve recognition accuracy. The contribution of each individual feature to the prediction probabilities is discussed through machine-learning interpretability based on Shapley values. The study reveals notable differences in brain activation states for identical emotions between the two modes, with VR-3D mode showing more pronounced activation.

Emotion Classification Based on Transformer and CNN for EEG Spatial-Temporal Feature Learning.

OBJECTIVES: The temporal and spatial information of electroencephalogram (EEG) signals is crucial for recognizing features in emotion classification models, but it excessively relies on manual feature extraction. The transformer model has the capability of performing automatic feature extraction; however, its potential has not been fully explored in the classification of emotion-related EEG signals. To address these challenges, the present study proposes a novel model based on transformer and convolutional neural networks (TCNN) for EEG spatial-temporal (EEG ST) feature learning to automatic emotion classification. METHODS: The proposed EEG ST-TCNN model utilizes position encoding (PE) and multi-head attention to perceive channel positions and timing information in EEG signals. Two parallel transformer encoders in the model are used to extract spatial and temporal features from emotion-related EEG signals, and a CNN is used to aggregate the EEG's spatial and temporal features, which are subsequently classified using Softmax. RESULTS: The proposed EEG ST-TCNN model achieved an accuracy of 96.67% on the SEED dataset and accuracies of 95.73%, 96.95%, and 96.34% for the arousal-valence, arousal, and valence dimensions, respectively, for the DEAP dataset. CONCLUSIONS: The results demonstrate the effectiveness of the proposed ST-TCNN model, with superior performance in emotion classification compared to recent relevant studies. SIGNIFICANCE: The proposed EEG ST-TCNN model has the potential to be used for EEG-based automatic emotion recognition.

[Ethical considerations for medical applications of implantable brain-computer interfaces].

Implantable brain-computer interfaces (BCIs) have potentially important clinical applications due to the high spatial resolution and signal-to-noise ratio of electrodes that are closer to or implanted in the cerebral cortex. However, the surgery and electrodes of implantable BCIs carry safety risks of brain tissue damage, and their medical applications face ethical challenges, with little literature to date systematically considering ethical norms for the medical applications of implantable BCIs. In order to promote the clinical translation of this type of BCI, we considered the ethics of practice for the medical application of implantable BCIs, including: reducing the risk of brain tissue damage from implantable BCI surgery and electrodes, providing patients with customized and personalized implantable BCI treatments, ensuring multidisciplinary collaboration in the clinical application of implantable BCIs, and the responsible use of implantable BCIs, among others. It is expected that this article will provide thoughts and references for the research and development of ethics of the medical application of implantable BCI.

Successful learning of alpha up-regulation through neurofeedback training modulates sustained attention.

As a fundamental attention function, sustained attention plays a critical role in general cognitive abilities and is closely linked to EEG alpha oscillations. Neurofeedback training (NFT) of alpha activity on different aspects of attention has been studied previously. However, it remains unclear how NFT with up- or down-regulation directions modulates sustained attention. Here we employed a counterbalanced single-blind sham-controlled crossover design, in which healthy young adults underwent one NFT session of alpha up-regulation, one NFT session of alpha down-regulation, and one sham-control NFT session over the posterior area. The session order was counterbalanced with a 7-day interval between each session. After each NFT session, the participants completed a visual continuous temporal expectancy task (vCTET) to assess their sustained attention performance. The results showed that compared to sham-control NFT, successful learning of alpha up-regulation resulted in increased reaction time at the beginning of the attention task but a slower increase over vCTET blocks. On the other hand, successful learning of alpha down-regulation had no impact on attention performance compared to sham-control NFT. These findings suggest that successful learning of alpha up-regulation through NFT could impair initial attention performance but slow down visual attention deterioration over time, i.e., alpha enhancement by NFT stabilizing visual attention.

Cross-Domain Identification of Multisite Major Depressive Disorder Using End-to-End Brain Dynamic Attention Network.

Establishing objective and quantitative imaging markers at individual level can assist in accurate diagnosis of Major Depressive Disorder (MDD). However, the clinical heterogeneity of MDD and the shift to multisite data decreased identification accuracy. To address these issues, the Brain Dynamic Attention Network (BDANet) is innovatively proposed, and analyzed bimodal scans from 2055 participants of the Rest-meta-MDD consortium. The end-to-end BDANet contains two crucial components. The Dynamic BrainGraph Generator dynamically focuses and represents topological relationships between Regions of Interest, overcoming limitations of static methods. The Ensemble Classifier is constructed to obfuscate domain sources to achieve inter-domain alignment. Finally, BDANet dynamically generates sample-specific brain graphs by downstream recognition tasks. The proposed BDANet achieved an accuracy of 81.6%. The regions with high attribution for classification were mainly located in the insula, cingulate cortex and auditory cortex. The level of brain connectivity in p24 region was negatively correlated ( [Formula: see text]) with the severity of MDD. Additionally, sex differences in connectivity strength were observed in specific brain regions and functional subnetworks ( [Formula: see text] or [Formula: see text]). These findings based on a large multisite dataset support the conclusion that BDANet can better solve the problem of the clinical heterogeneity of MDD and the shift of multisite data. It also illustrates the potential utility of BDANet for personalized accurate identification, treatment and intervention of MDD.

Analysis of Artifactual Components Rejection Threshold towards Enhanced Characterization of Neural Activity in Post-Stroke Survivor.

Research advancement has spurred the usage of electroencephalography (EEG)-based neural oscillatory rhythms as a biomarker to complement clinical rehabilitation strategies for the recovery of motor functions in stroke survivors. However, the inevitable contamination of EEG signals with artifacts from various sources limits its utilization and effectiveness. Thus, the integration of Independent Component Analysis (ICA) and Independent Component Label (ICLabel) has been widely employed to separate neural activity from artifacts. A crucial step in the ICLabel preprocessing pipeline is the artifactual ICs rejection threshold (TH) parameter, which determines the overall signal's quality. For instance, selecting a high TH will cause many ICs to be rejected, thereby leading to signal over-cleaning, and choosing a low TH may result in under-cleaning of the signal. Toward determining the optimal TH parameter, this study investigates the effect of six different TH groups (NO-TH and TH1-TH6) on EEG signals recorded from post-stroke patients who performed four distinct motor imagery (MI) tasks including wrist and grasping movements. Utilizing the EEG-beta band signal at the brain's sensorimotor cortex, the performance of the TH groups was evaluated using three notable EEG quantifiers. Overall, the obtained result shows that the considered THs will significantly alter neural oscillatory patterns. Comparing the performance of the TH-groups, TH-3 with a confidence level of 60% showed consistently stronger signal desynchronization and lateralization. The correlation result shows that most of the electrode pairs with high correlation values are replicable across all the MI tasks. It also revealed that brain activity correlates linearly with distance, and a strong correlation between electrode pairs is independent of the different brain cortices. The study outcome may facilitate adequate therapeutic intervention for stroke rehab.Clinical Relevance: This study indicated that optimal selection of the ICLabel artifactual rejection threshold is essential for EEG enhancement for adequate signal characterization. Thus, a TH-values with a confidence level between 50% - 70% would be suggested for artifactual ICs rejection in MI-EEG.

Neural Mechanisms of Visual-Spatial Judgment Behavior under Visual and Auditory Constraints: Evidence from an Electroencephalograph during Handgun Shooting.

Light and noise are important factors affecting shooting performance, and shooters can exhibit physiological processes that differ from normal shooting when they are subjected to disturbed visual and auditory conditions. The purpose of this study was to explore the neural mechanism of shooting preparation in skilled shooters with visual and auditory limitations. We designed an experiment and recorded the electroencephalograph (EEG) and shooting performance indexes of 40 individuals skilled in marksmanship during the shooting preparation stage under three conditions: low light, noise interference, and a normal environment. EEG relative band power features and event-related desynchronization/synchronization (ERD/ERS) features were extracted and analyzed. The results showed that (1) the average score of the shooters was 8.55 under normal conditions, 7.71 under visually restricted conditions, and 8.50 under auditorily restricted conditions; (2) the relative EEG band power in the frontal lobe (Fp1, Fp2), frontal lobe (F4, F8), left temporal region (T7), central lobe (CP2), and parietal lobe (P3, PO3) in the theta band was significantly lower than in the other two environments (p < 0.05), and there was no significant difference between the power intensity of the shooter in the noisy environment and that in the normal environment; and (3) in the low-light environment, a significant negative correlation was found between the central region, the left and right temporal regions, and the parietal lobe (p < 0.05). These findings provide a basis for further understanding neural mechanisms in the brain during the shooting preparation phase under visually and auditorily restricted conditions.

Optimization of SSVEP-BCI Virtual Reality Stereo Stimulation Parameters Based on Knowledge Graph.

The steady-state visually evoked potential (SSVEP) is an important type of BCI that has various potential applications, including in virtual environments using virtual reality (VR). However, compared to VR research, the majority of visual stimuli used in the SSVEP-BCI are plane stimulation targets (PSTs), with only a few studies using stereo stimulation targets (SSTs). To explore the parameter optimization of the SSVEP-BCI virtual SSTs, this paper presents a parameter knowledge graph. First, an online VR stereoscopic stimulation SSVEP-BCI system is built, and a parameter dictionary for VR stereoscopic stimulation parameters (shape, color, and frequency) is established. The online experimental results of 10 subjects under different parameter combinations were collected, and a knowledge graph was constructed to optimize the SST parameters. The best classification performances of the shape, color, and frequency parameters were sphere (91.85%), blue (94.26%), and 13Hz (95.93%). With various combinations of virtual reality stereo stimulation parameters, the performance of the SSVEP-BCI varies. Using the knowledge graph of the stimulus parameters can help intuitively and effectively select appropriate SST parameters. The knowledge graph of the stereo target stimulation parameters presented in this work is expected to offer a way to convert the application of the SSVEP-BCI and VR.

[Ethics considerations on brain-computer interface technology].

The development and potential application of brain-computer interface (BCI) technology is closely related to the human brain, so that the ethical regulation of BCI has become an important issue attracting the consideration of society. Existing literatures have discussed the ethical norms of BCI technology from the perspectives of non-BCI developers and scientific ethics, while few discussions have been launched from the perspective of BCI developers. Therefore, there is a great need to study and discuss the ethical norms of BCI technology from the perspective of BCI developers. In this paper, we present the user-centered and non-harmful BCI technology ethics, and then discuss and look forward on them. This paper argues that human beings can cope with the ethical issues arising from BCI technology, and as BCI technology develops, its ethical norms will be improved continuously. It is expected that this paper can provide thoughts and references for the formulation of ethical norms related to BCI technology.

Computer-aided diagnosis of schizophrenia based on node2vec and Transformer.

OBJECTIVE: Compared with the healthy control (HC) group, the brain structure and function of schizophrenia (SZ) patients are significantly abnormal, so brain imaging methods can be used to achieve the aided diagnosis of SZ. However, a brain network based on brain imaging data is non-Euclidean, and its intrinsic features cannot be learned effectively by general deep learning models. Furthermore, in the majority of existing studies, brain network features were manually specified as the input of machine learning models. METHODS: In this study, brain functional network constructed from the subject's fMRI data is analyzed, and its small-world value is calculated and t-tested; the node2vec algorithm in graph embedding is introduced to transform the constructed brain network into low-dimensional dense vectors, and the brain network's non-Euclidean spatial structure characteristics are retained to the greatest extent, so that its intrinsic features can be extracted by deep learning models; GridMask is used to randomly mask part of the information in the vectors to enhance the data; and then features can be extracted using the Transformer model to identify SZ. RESULTS: It is again shown that the small-world value of the brain network in SZ is significantly lower than that in HC by t-test (p=0.014¡0.05). 97.78% classification accuracy is achieved by the proposed methods (node2vec + GridMask + Transformer) in 30 SZ patients and 30 healthy people. CONCLUSION: The experiment shows that the node2vec used in this paper can effectively solve the problem of brain network features being difficult to learn by general deep learning models. The high-precision computer-aided diagnosis of SZ can be obtained by combining node2vec with Transformer and GridMask. SIGNIFICANCE: The proposed methods in the paper are expected to be used for aided diagnosis of SZ.

Human factors engineering of BCI: an evaluation for satisfaction of BCI based on motor imagery.

Existing brain-computer interface (BCI) research has made great progress in improving the accuracy and information transfer rate (ITR) of BCI systems. However, the practicability of BCI is still difficult to achieve. One of the important reasons for this difficulty is that human factors are not fully considered in the research and development of BCI. As a result, BCI systems have not yet reached users' expectations. In this study, we investigate a BCI system of motor imagery for lower limb synchronous rehabilitation as an example. From the perspective of human factors engineering of BCI, a comprehensive evaluation method of BCI system development is proposed based on the concept of human-centered design and evaluation. Subjects' satisfaction ratings for BCI sensors, visual analog scale (VAS), subjects' satisfaction rating of the BCI system, and the mental workload rating for subjects manipulating the BCI system, as well as interview/follow-up comprehensive evaluation of motor imagery of BCI (MI-BCI) system satisfaction were used. The methods and concepts proposed in this study provide useful insights for the design of personalized MI-BCI. We expect that the human factors engineering of BCI could be applied to the design and satisfaction evaluation of MI-BCI, so as to promote the practical application of this kind of BCI.

Brain-computer interface paradigms and neural coding.

Brain signal patterns generated in the central nervous system of brain-computer interface (BCI) users are closely related to BCI paradigms and neural coding. In BCI systems, BCI paradigms and neural coding are critical elements for BCI research. However, so far there have been few references that clearly and systematically elaborated on the definition and design principles of the BCI paradigm as well as the definition and modeling principles of BCI neural coding. Therefore, these contents are expounded and the existing main BCI paradigms and neural coding are introduced in the review. Finally, the challenges and future research directions of BCI paradigm and neural coding were discussed, including user-centered design and evaluation for BCI paradigms and neural coding, revolutionizing the traditional BCI paradigms, breaking through the existing techniques for collecting brain signals and combining BCI technology with advanced AI technology to improve brain signal decoding performance. It is expected that the review will inspire innovative research and development of the BCI paradigm and neural coding.

[Neurofeedback technology based on functional near infrared spectroscopy imaging and its applications].

Neurofeedback (NF) technology based on electroencephalogram (EEG) data or functional magnetic resonance imaging (fMRI) has been widely studied and applied. In contrast, functional near infrared spectroscopy (fNIRS) has become a new technique in NF research in recent years. fNIRS is a neuroimaging technology based on hemodynamics, which has the advantages of low cost, good portability and high spatial resolution, and is more suitable for use in natural environments. At present, there is a lack of comprehensive review on fNIRS-NF technology (fNIRS-NF) in China. In order to provide a reference for the research of fNIRS-NF technology, this paper first describes the principle, key technologies and applications of fNIRS-NF, and focuses on the application of fNIRS-NF. Finally, the future development trend of fNIRS-NF is prospected and summarized. In conclusion, this paper summarizes fNIRS-NF technology and its application, and concludes that fNIRS-NF technology has potential practicability in neurological diseases and related fields. fNIRS can be used as a good method for NF training. This paper is expected to provide reference information for the development of fNIRS-NF technology.

Multi-source joint domain adaptation for cross-subject and cross-session emotion recognition from electroencephalography.

As an important component to promote the development of affective brain-computer interfaces, the study of emotion recognition based on electroencephalography (EEG) has encountered a difficult challenge; the distribution of EEG data changes among different subjects and at different time periods. Domain adaptation methods can effectively alleviate the generalization problem of EEG emotion recognition models. However, most of them treat multiple source domains, with significantly different distributions, as one single source domain, and only adapt the cross-domain marginal distribution while ignoring the joint distribution difference between the domains. To gain the advantages of multiple source distributions, and better match the distributions of the source and target domains, this paper proposes a novel multi-source joint domain adaptation (MSJDA) network. We first map all domains to a shared feature space and then align the joint distributions of the further extracted private representations and the corresponding classification predictions for each pair of source and target domains. Extensive cross-subject and cross-session experiments on the benchmark dataset, SEED, demonstrate the effectiveness of the proposed model, where more significant classification results are obtained on the more difficult cross-subject emotion recognition task.

Brain Network Research of Skilled Shooters in the Shooting Preparation Stage under the Condition of Limited Sensory Function.

Shooting is a sport dominated by psychological factors. Hence, disturbing the shooter's sensory function during aiming will seriously affect his psychological state and shooting performance. Electroencephalograph (EEG) measurements of 30 skilled marksmen in the shooting preparation stage under noisy disturbance, weak light, and normal conditions were recorded. Therefore, the differences in neural mechanisms in the shooter's brain during shooting aiming in different disturbance conditions were explored using an analytical approach that employs functional connectivity and brain network analysis based on graph theory. The relationship between these brain network characteristics and shooting performance was also compared. The results showed that (1) the average connection strength in the beta frequency band and connection intensity in the left and right temporal lobes of the shooters under noise disturbance were significantly higher than those under the other two conditions, and their brain networks also showed a higher global and local efficiency. In addition, (2) the functional connection intensity in the occipital region of the beta band was higher than that in the normal condition in the weak-light condition. The information interaction in the left parietal region also increased continually during the shooting process. (3) Furthermore, the shooters' eigenvector centrality in the temporal and occipital regions with limited sensory function in the two conditions was lower than those in the normal condition. These findings suggest that noise disturbance activates the arousal level of the shooter's brain and enhances the information processing efficiency of the brain network; however, it increases the mental workload. In weak-light conditions, shooters focus more on visual information processing during aiming and strengthen the inhibition of functions in the brain regions unrelated to shooting behavior. Audiovisual disturbance renders the cortical regions equivalent to the audiovisual perception function in the shooter's brain less important in the entire brain network than in the normal condition. Therefore, these findings reveal the effect of audiovisual disturbance on the functional network of the cortex in the shooting preparation stage and provide a theoretical basis for further understanding the neural mechanism of the shooting process under sensory disturbances.

Research on Top Archer's EEG Microstates and Source Analysis in Different States.

The electroencephalograph (EEG) microstate is a method used to describe the characteristics of the EEG signal through the brain scalp electrode potential's spatial distribution; as such, it reflects the changes in the brain's functional state. The EEGs of 13 elite archers from China's national archery team and 13 expert archers from China's provincial archery team were recorded under the alpha rhythm during the resting state (with closed eyes) and during archery aiming. By analyzing the differences between the EEG microstate parameters and the correlation between these parameters with archery performance, as well as by combining our findings through standardized low-resolution brain electromagnetic tomography source analysis (sLORETA), we explored the changes in the neural activity of professional archers of different levels, under different states. The results of the resting state study demonstrated that the duration, occurrence, and coverage in microstate D of elite archers were significantly higher than those of expert archers and that their other microstates had the greatest probability of transferring to microstate D. During the archery aiming state, the average transition probability of the other microstates transferring to microstate in the left temporal region was the highest observed in the two groups of archers. Moreover, there was a significant negative correlation between the duration and coverage of microstates in the frontal region of elite archers and their archery performance. Our findings indicate that elite archers are more active in the dorsal attention system and demonstrate a higher neural efficiency during the resting state. When aiming, professional archers experience an activation of brain regions associated with archery by suppressing brain regions unrelated to archery tasks. These findings provide a novel theoretical basis for the study of EEG microstate dynamics in archery and related cognitive motor tasks, particularly from the perspective of the subject's mental state.

Enhanced beetle antennae search algorithm for complex and unbiased optimization.

Beetle Antennae Search algorithm is a kind of intelligent optimization algorithms, which has the advantages of few parameters and simplicity. However, due to its inherent limitations, BAS has poor performance in complex optimization problems. The existing improvements of BAS are mainly based on the utilization of multiple beetles or combining BAS with other algorithms. The present study improves BAS from its origin and keeps the simplicity of the algorithm. First, an adaptive step size reduction method is used to increase the usability of the algorithm, which is based on an accurate factor and curvilinearly reduces the step size; second, the calculated information of fitness functions during each iteration are fully utilized with a contemporary optimal update strategy to promote the optimization processes; third, the theoretical analysis of the multi-directional sensing method is conducted and utilized to further improve the efficiency of the algorithm. Finally, the proposed Enhanced Beetle Antennae Search algorithm is compared with many other algorithms based on unbiased test functions. The test functions are unbiased when their solution space does not contain simple patterns, which may be used to facilitate the searching processes. As a result, EBAS outperformed BAS with at least 1 orders of magnitude difference. The performance of EBAS was even better than several state-of-the-art swarm-based algorithms, such as Slime Mold Algorithm and Grey Wolf Optimization, with similar running times. In addition, a WSN coverage optimization problem is tested to demonstrate the applicability of EBAS on real-world optimizations.

[Key technology of brain-computer interaction based on speech imagery].

Speech expression is an important high-level cognitive behavior of human beings. The realization of this behavior is closely related to human brain activity. Both true speech expression and speech imagination can activate part of the same brain area. Therefore, speech imagery becomes a new paradigm of brain-computer interaction. Brain-computer interface (BCI) based on speech imagery has the advantages of spontaneous generation, no training, and friendliness to subjects, so it has attracted the attention of many scholars. However, this interactive technology is not mature in the design of experimental paradigms and the choice of imagination materials, and there are many issues that need to be discussed urgently. Therefore, in response to these problems, this article first expounds the neural mechanism of speech imagery. Then, by reviewing the previous BCI research of speech imagery, the mainstream methods and core technologies of experimental paradigm, imagination materials, data processing and so on are systematically analyzed. Finally, the key problems and main challenges that restrict the development of this type of BCI are discussed. And the future development and application perspective of the speech imaginary BCI system are prospected.

Improved Brain-Computer Interface Signal Recognition Algorithm Based on Few-Channel Motor Imagery.

Common spatial pattern (CSP) is an effective algorithm for extracting electroencephalogram (EEG) features of motor imagery (MI); however, CSP mainly aims at multichannel EEG signals, and its effect in extracting EEG features with fewer channels is poor-even worse than before using CSP. To solve the above problem, a new combined feature extraction method has been proposed in this study. For EEG signals from fewer channels (three channels), wavelet packet transform, fast ensemble empirical mode decomposition, and local mean decomposition were used to decompose the band-pass filtered EEG into multiple time-frequency components, and the corresponding components were selected according to the frequency characteristics of MI or the correlation coefficient between its time-frequency components and the original EEG signal. Furthermore, phase space reconstruction (PSR) was performed on the selected components after the three time-frequency decompositions, the maximum Lyapunov index was calculated, and the features were reconstructed; then, CSP projection mapping was used for the reconstructed features. The support vector machine probability output model was trained by the obtained three mappings. Probability outputs by three different support vector machines were then obtained. Finally, the classification of test samples was determined by the fusion of the Dempster-Shafer evidence theory at the decision level. The results showed that the accuracy of the proposed method was 95.71% on data set III of BCI competition II (left- and right-hand MI), which was 2.88% higher than the existing methods. On data set IIb of BCI competition IV, the average accuracy was 86.60%, which was 2.3% higher than the existing methods. This study verified the effectiveness of the proposed method and provided an approach for the research and development of the MI-BCI system based on fewer channels.