Motor Imagery Brain-computer Interface Research Articles

Optimizing motor imagery BCI models with hard trials removal and model refinement.

Deep learning models have demonstrated remarkable performance in the classification of motor imagery BCI systems. However, these models exhibit sensitivity to challenging trials, often called hard trials, leading to performance degradation. In this paper, we address this issue by proposing two novel methods for identifying and mitigating the impact of hard trials on model performance. The first method leverages model prediction scores to discern hard trials. The second approach employs a quantitative explainable artificial intelligence (XAI) approach, enabling a more transparent and interpretable means of hard trials identification.
The identified hard trials are removed from the entire motor imagery training and validation dataset, and the deep learning model is further re-trained using the dataset without hard trials.
To evaluate the efficacy of these proposed methods, experiments were conducted on the Open
BMI dataset. The results for hold-out analysis show that, the proposed quantitative XAI based hard trial removal method has statistically improved the average classification accuracy of the baseline deep CNN model from 63.77 % to 68.70 %, with p-value = 7.66 -11 for the subject specific MI classification. Additionally, analyzing the scalp map depicting the average relevance scores of correctly classified trials compared to a misclassified trial provides a deeper insight into identifying hard trials. The results indicates that the proposed quantitative based XAI approach outperforms the prediction-score based approach in hard trial identification.&#xD.

PRISM: Deep metric learning based personal grouping method to reduce intersubject variability for motor imagery brain–computer interface

Motor imagery brain–computer interfaces (MI–BCIs) that use electroencephalogram (EEG) signals are essential for motor rehabilitation. One of the most important problems in MI–BCIs is intersubject variability, as brain signals can vary significantly among individuals due to differences in cognitive processing and mental states. To address this issue, some studies have proposed intrasubject learning models that train and evaluate within specific individuals, while others have suggested intersubject learning models that aim to generalize across subjects. However, insufficient data are available for intrasubject learning, so they cannot use a deep learning model, and still, the problem of intersubject variability is unsolved in intersubject learning. To tackle these challenges, we developed a new framework based on deep metric learning called the Personal grouping method for Reducing InterSubject variability using deep Metric learning (PRISM). PRISM aims to reduce intersubject variability by calculating the similarity between subjects based on their EEG signals and generating clusters. Based on these clusters, the model can train by reducing intersubject variability while maintaining the advantage of intersubject learning. Our thorough evaluation indicates that PRISM is poised to redefine the standards in MI–BCIs, significantly alleviating the challenges of intersubject variability. We compared the performance of PRISM and the existing learning paradigm, and the performance of PRISM was similar to that of the intrasubject learning model (p > 0.4), which is the best existing model, and it performed better than the intersubject learning model (**p < 0.001). Our results indicate that the PRISM framework outperforms intersubject learning and closely matches intrasubject learning models. This suggests that PRISM combines the specificity of intrasubject learning with intersubject model generalizability, thereby enhancing performance.

Large scale investigation of the effect of gender on mu rhythm suppression in motor imagery brain-computer interfaces

ABSTRACT The utmost issue in Motor Imagery Brain-Computer Interfaces (MI-BCI) is the BCI poor performance known as ‘BCI inefficiency’. Although past research has attempted to find a solution by investigating factors influencing users’ MI-BCI performance, the issue persists. One of the factors that has been studied in relation to MI-BCI performance is gender. Research regarding the influence of gender on a user’s ability to control MI-BCIs remains inconclusive, mainly due to the small sample size and unbalanced gender distribution in past studies. To address these issues and obtain reliable results, this study combined four MI-BCI datasets into one large dataset with 248 subjects and equal gender distribution. The datasets included EEG signals from healthy subjects from both gender groups who had executed a right- vs. left-hand motor imagery task following the Graz protocol. The analysis consisted of extracting the Mu Suppression Index from C3 and C4 electrodes and comparing the values between female and male participants. Unlike some of the previous findings which reported an advantage for female BCI users in modulating mu rhythm activity, our results did not show any significant difference between the Mu Suppression Index of both groups, indicating that gender may not be a predictive factor for BCI performance.

Frequency Domain Channel-Wise Attack to CNN Classifiers in Motor Imagery Brain-Computer Interfaces.

Convolutional neural network (CNN), a classical structure in deep learning, has been commonly deployed in the motor imagery brain-computer interface (MIBCI). Many methods have been proposed to evaluate the vulnerability of such CNN models, primarily by attacking them using direct temporal perturbations. In this work, we propose a novel attacking approach based on perturbations in the frequency domain instead. For a given natural MI trial in the frequency domain, the proposed approach, called frequency domain channel-wise attack (FDCA), generates perturbations at each channel one after another to fool the CNN classifiers. The advances of this strategy are two-fold. First, instead of focusing on the temporal domain, perturbations are generated in the frequency domain where discriminative patterns can be extracted for motor imagery (MI) classification tasks. Second, the perturbing optimization is performed based on differential evolution algorithm in a black-box scenario where detailed model knowledge is not required. Experimental results demonstrate the effectiveness of the proposed FDCA which achieves a significantly higher success rate than the baselines and existing methods in attacking three major CNN classifiers on four public MI benchmarks. Perturbations generated in the frequency domain yield highly competitive results in attacking MIBCI deployed by CNN models even in a black-box setting, where the model information is well-protected. To our best knowledge, existing MIBCI attack approaches are all gradient-based methods and require details about the victim model, e.g., the parameters and objective function. We provide a more flexible strategy that does not require model details but still produces an effective attack outcome.

Multi-Source Transfer Learning via Optimal Transport Feature Ranking for EEG Classification

Motor imagery (MI) brain-computer interface (BCI) paradigms have been extensively used in neurological rehabilitation. However, due to the required long calibration time and non-stationary nature of electroencephalogram (EEG) signals, it is challenging to obtain a substantial sample size of EEG data. Transfer learning (TL), which leverages knowledge from existing subjects to enhance the learning performance in new subjects, has been employed to solve this problem. Previous studies often extract sample features directly from the two domains for transfer learning, leading to poor features and negative transfer. To overcome this limitation, an optimal transport feature selection method was developed in this study to select the most suitable features for migration to enhance the generalization capability of the TL. This was achieved by analyzing the diagonal value similarity of the optimal transport coupling matrix. First, the Riemannian tangent space mapping method was used to map the sample covariance matrix of EEG trials from the Riemannian manifold to tangent space. Secondly, the optimal subset of EEG features was selected via the optimal transport feature selection method. Subsequently, the separate distribution alignment method was employed to minimize dissimilarities between the source and target domains. Finally, the weighted voting mechanisms were integrated for decision fusion. The proposed multi-source transfer feature learning (MSTFL) method was validated based on two public BCI Competition IV datasets. Our results demonstrated superior classification accuracy of 85.93% and 79.48%, respectively, on the two public datasets, compared to state-of-the-art models. These findings indicate the effectiveness of our proposed MSTFL model for both feature extraction and classification of MI signals.

Fatigue in children using motor imagery and P300 brain-computer interfaces

BackgroundBrain-computer interface (BCI) technology offers children with quadriplegic cerebral palsy unique opportunities for communication, environmental exploration, learning, and game play. Research in adults demonstrates a negative impact of fatigue on BCI enjoyment, while effects on BCI performance are variable. To date, there have been no pediatric studies of BCI fatigue. The purpose of this study was to assess the effects of two different BCI paradigms, motor imagery and visual P300, on the development of self-reported fatigue and an electroencephalography (EEG) biomarker of fatigue in typically developing children.MethodsThirty-seven typically-developing school-aged children were recruited to a prospective, crossover study. Participants attended three sessions: (A) motor imagery-BCI, (B) visual P300-BCI, and (C) video viewing (control). The motor imagery task involved an imagined left- or right-hand squeeze. The P300 task involved attending to one square on a 3 × 3 grid during a random single flash sequence. Each paradigm had respective calibration periods and a similar visual counting game. Primary outcomes were self-reported fatigue and the power of the EEG alpha band both collected during resting-state periods pre- and post-task. Self-reported fatigue was measured using a 10-point visual analog scale. EEG alpha band power was calculated as the integrated power spectral density from 8 to 12 Hz of the EEG spectrum.ResultsThirty-two children completed the protocol (age range 7–16, 63% female). Self-reported fatigue and EEG alpha band power increased across all sessions (F(1,155) = 33.9, p < 0.001; F = 5.0(1,149), p = 0.027 respectively). No differences in fatigue development were observed between session types. There was no correlation between self-reported fatigue and EEG alpha band power change. BCI performance varied between participants and paradigms as expected but was not associated with self-reported fatigue or EEG alpha band power.ConclusionShort periods (30-mintues) of BCI use can increase self-reported fatigue and EEG alpha band power to a similar degree in children performing motor imagery and P300 BCI paradigms. Performance was not associated with our measures of fatigue; the impact of fatigue on useability and enjoyment is unclear. Our results reflect the variability of fatigue and the BCI experience more broadly in children and warrant further investigation.

EEG rhythm separation and time-frequency analysis of fast multivariate empirical mode decomposition for motor imagery BCI.

Motor imagery electroencephalogram (EEG) is widely employed in brain-computer interface (BCI) systems. As a time-frequency analysis method for nonlinear and non-stationary signals, multivariate empirical mode decomposition (MEMD) and its noise-assisted version (NA-MEMD) has been widely used in the preprocessing step of BCI systems for separating EEG rhythms corresponding to specific brain activities. However, when applied to multichannel EEG signals, MEMD or NA-MEMD often demonstrate low robustness to noise and high computational complexity. To address these issues, we have explored the advantages of our recently proposed fast multivariate empirical mode decomposition (FMEMD) and its noise-assisted version (NA-FMEMD) for analyzing motor imagery data. We emphasize that FMEMD enables a more accurate estimation of EEG frequency information and exhibits a more noise-robust decomposition performance with improved computational efficiency. Comparative analysis with MEMD on simulation data and real-world EEG validates the above assertions. The joint average frequency measure is employed to automatically select intrinsic mode functions that correspond to specific frequency bands. Thus, FMEMD-based classification architecture is proposed. Using FMEMD as a preprocessing algorithm instead of MEMD can improve the classification accuracy by 2.3% on the BCI Competition IV dataset. On the Physiobank Motor/Mental Imagery dataset and BCI Competition IV Dataset 2a, FMEMD-based architecture also attained a comparable performance to complex algorithms. The results indicate that FMEMD proficiently extracts feature information from small benchmark datasets while mitigating dimensionality constraints resulting from computational complexity. Hence, FMEMD or NA-FMEMD can be a powerful time-frequency preprocessing method for BCI.

A robust multi-branch multi-attention-mechanism EEGNet for motor imagery BCI decoding

BackgroundMotor-Imagery-based Brain-Computer Interface (MI-BCI) is a promising technology to assist communication, movement, and neurological rehabilitation for motor-impaired individuals. Electroencephalography (EEG) decoding techniques using deep learning (DL) possess noteworthy advantages due to automatic feature extraction and end-to-end learning. However, the DL-based EEG decoding models tend to show large variations due to intersubject variability of EEG, which results from inconsistencies of different subjects’ optimal hyperparameters. New methodsThis study proposes a multi-branch multi-attention mechanism EEGNet model (MBMANet) for robust decoding. It applies the multi-branch EEGNet structure to achieve various feature extractions. Further, the different attention mechanisms introduced in each branch attain diverse adaptive weight adjustments. This combination of multi-branch and multi-attention mechanisms allows for multi-level feature fusion to provide robust decoding for different subjects. ResultsThe MBMANet model has a four-classification accuracy of 83.18% and kappa of 0.776 on the BCI Competition IV-2a dataset, which outperforms other eight CNN-based decoding models. This consistently satisfactory performance across all nine subjects indicates that the proposed model is robust. ConclusionsThe combine of multi-branch and multi-attention mechanisms empowers the DL-based models to adaptively learn different EEG features, which provides a feasible solution for dealing with data variability. It also gives the MBMANet model more accurate decoding of motion intentions and lower training costs, thus improving the MI-BCI’s utility and robustness.

Applications of Brain Computer Interface for Motor Imagery Using Deep Learning: Review on Recent Trends

Motor Imagery-Brain Computer Interface (MI-BCI) is a very important technology gaining momentum throughout the last decade. This technology enables the linkage of brain activities to computer applications and can give disabled patients who suffer from motor disabilities (e.g., partial paralysis, muscle atrophy, etc.) the ability to interact normally with technologies around them. Currently, the technology is mostly limited to applications within dedicated laboratories and is hardly used in practical settings or in real-life applications. The purpose of this study is to review the latest trends and technologies in the field of MI-BCI, including the major challenges and the state-of-the-art classification techniques. The scope of this review article covers the feature selection algorithms that can help identify the most informative and discriminative features from the recorded brain signals, and the classification techniques that can identify the different types of motor movements.

Multilayer network-based channel selection for motor imagery brain–computer interface

Objective. The number of electrode channels in a motor imagery-based brain–computer interface (MI-BCI) system influences not only its decoding performance, but also its convenience for use in applications. Although many channel selection methods have been proposed in the literature, they are usually based on the univariate features of a single channel. This leads to a loss of the interaction between channels and the exchange of information between networks operating at different frequency bands. Approach. We integrate brain networks containing four frequency bands into a multilayer network framework and propose a multilayer network-based channel selection (MNCS) method for MI-BCI systems. A graph learning-based method is used to estimate the multilayer network from electroencephalogram (EEG) data that are filtered by multiple frequency bands. The multilayer participation coefficient of the multilayer network is then computed to select EEG channels that do not contain redundant information. Furthermore, the common spatial pattern (CSP) method is used to extract effective features. Finally, a support vector machine classifier with a linear kernel is trained to accurately identify MI tasks. Main results. We used three publicly available datasets from the BCI Competition containing data on 12 healthy subjects and one dataset containing data on 15 stroke patients to validate the effectiveness of our proposed method. The results showed that the proposed MNCS method outperforms all channels (85.8% vs. 93.1%, 84.4% vs. 89.0%, 71.7% vs. 79.4%, and 72.7% vs. 84.0%). Moreover, it achieved significantly higher decoding accuracies on MI-BCI systems than state-of-the-art methods (paired t-tests, p < 0.05). Significance. The experimental results showed that the proposed MNCS method can select appropriate channels to improve the decoding performance as well as the convenience of the application of MI-BCI systems.

Transferring a deep learning model from healthy subjects to stroke patients in a motor imagery brain–computer interface

Objective. Motor imagery (MI) brain–computer interfaces (BCIs) based on electroencephalogram (EEG) have been developed primarily for stroke rehabilitation, however, due to limited stroke data, current deep learning methods for cross-subject classification rely on healthy data. This study aims to assess the feasibility of applying MI-BCI models pre-trained using data from healthy individuals to detect MI in stroke patients. Approach. We introduce a new transfer learning approach where features from two-class MI data of healthy individuals are used to detect MI in stroke patients. We compare the results of the proposed method with those obtained from analyses within stroke data. Experiments were conducted using Deep ConvNet and state-of-the-art subject-specific machine learning MI classifiers, evaluated on OpenBMI two-class MI-EEG data from healthy subjects and two-class MI versus rest data from stroke patients. Main results. Results of our study indicate that through domain adaptation of a model pre-trained using healthy subjects’ data, an average MI detection accuracy of 71.15% ( ±12.46% ) can be achieved across 71 stroke patients. We demonstrate that the accuracy of the pre-trained model increased by 18.15% after transfer learning (p < 0.001). Additionally, the proposed transfer learning method outperforms the subject-specific results achieved by Deep ConvNet and FBCSP, with significant enhancements of 7.64% (p < 0.001) and 5.55% (p < 0.001) in performance, respectively. Notably, the healthy-to-stroke transfer learning approach achieved similar performance to stroke-to-stroke transfer learning, with no significant difference (p > 0.05). Explainable AI analyses using transfer models determined channel relevance patterns that indicate contributions from the bilateral motor, frontal, and parietal regions of the cortex towards MI detection in stroke patients. Significance. Transfer learning from healthy to stroke can enhance the clinical use of BCI algorithms by overcoming the challenge of insufficient clinical data for optimal training.

Avatar embodiment prior to motor imagery training in VR does not affect the induced event-related desynchronization: a pilot study

Motor-imagery brain-computer interfaces (MI-BCIs) have the potential to improve motor function in individuals with neurological disorders. Their effectiveness relies on patients’ ability to generate reliable MI-related electroencephalography (EEG) patterns, which can be influenced by the quality of neurofeedback. Virtual Reality (VR) has emerged as a promising tool for enhancing proprioceptive feedback due to its ability to induce a sense of embodiment (SoE), where individuals perceive a virtual body as their own. Although prior research has highlighted the importance of SoE in enhancing MI skills and BCI performance, to date, no study has successfully isolated nor manipulated the SoE in VR before MI training, creating a gap in our understanding of the precise role of the priming effect of embodiment in MI-BCIs. In this study, we aimed to examine whether the virtual SoE when induced, as priming of avatar embodiment, and assessed before MI training, could enhance MI-induced EEG patterns. To achieve this, we divided 26 healthy participants into two groups: the embodied group, which experienced SoE with an avatar before undergoing VR-based MI training, and the non-embodied group, which underwent the same MI training without a prior embodiment phase, serving as a control. We analyzed subjective measures of embodiment, the event-related desynchronization (ERD) power of the sensorimotor rhythms, lateralization of ERD, and offline classification BCI accuracy. Although the embodiment phase effectively induced SoE in the embodied group, both groups exhibited similar MI-induced ERD patterns and BCI classification accuracy. This suggests that the induction of SoE prior to MI training may not significantly influence the training outcomes. Instead, it appears that the integration of embodied VR feedback during MI training itself is sufficient to induce appropriate ERD, as evidenced by previous research.

A time segment adaptive optimization method based on separability criterion and correlation analysis for motor imagery BCIs

Motor imagery (MI) plays a crucial role in brain-computer interface (BCI), and the classification of MI tasks using electroencephalogram (EEG) is currently under extensive investigation. During MI classification, individual differences among subjects in terms of response and time latency need to be considered. Optimizing the time segment for different subjects can enhance subsequent classification performance. In view of the individual differences of subjects in motor imagery tasks, this article proposes a Time Segment Adaptive Optimization method based on Separability criterion and Correlation analysis (TSAOSC). The fundamental principle of this method involves applying the separability criterion to various sizes of time windows within the training data, identifying the optimal raw reference signal, and adaptively adjusting the time segment position for each trial’s data by analyzing its relationship with the optimal reference signal. We evaluated our method on three BCI competition datasets, respectively. The utilization of the TSAOSC method in the experiments resulted in an enhancement of 4.90% in average classification accuracy compared to its absence. Additionally, building upon the TSAOSC approach, this study proposes a Nonlinear-TSAOSC method (N-TSAOSC) for analyzing EEG signals with nonlinearity, which shows improvements in the classification accuracy of certain subjects. The results of the experiments demonstrate that the proposed method is an effective time segment optimization method, and it can be integrated into other algorithms to further improve their accuracy.

Model and Data Dual-Driven Double-Point Observation Network for Ultra-Short MI EEG Classification.

Although deep networks have succeeded in various signal classification tasks, the time sequence samples used to train the deep models are usually required to reach a certain length. Especially, in brain computer interface (BCI) research, around 3.5s-long motor imagery (MI) Electroencephalography (EEG) samples are needed to obtain satisfactory classification performance. This time-span requirement of the training samples makes real-time MI BCI systems impossible to implement based on deep networks, which restricts the related researches within laboratory and makes practical application hard to accomplish. To address this issue, a double-point observation deep network (DoNet) is developed to classify ultra-short samples buried in noise. First, an analytical solution is developed theoretically to perform ultra-short signal classification based on double-point couples. Then, a signal-noise model is constructed to study the interference of noise on classification based on double-point couples. Based on which, an independent identical distribution condition is utilized to improve the classification accuracy in a data-driven manner. Combining the theoretical model and data-driven mechanism, DoNet can construct a steady data-distribution for the double-point couples of the samples with the same label. Therefore, the conditional probability of each double-point couple of a test sample can be obtained. With a voting strategy, the samples can be accurately classified by fusing these conditional probabilities. Meanwhile, the noise interference can be suppressed. DoNet has been evaluated on two public EEG datasets. Compared to most state-of-the-art methods, the 1s-long EEG signal classification accuracy has been improved by more than 3%.

EISATC-Fusion: Inception Self-Attention Temporal Convolutional Network Fusion for Motor Imagery EEG Decoding.

The motor imagery brain-computer interface (MI-BCI) based on electroencephalography (EEG) is a widely used human-machine interface paradigm. However, due to the non-stationarity and individual differences among subjects in EEG signals, the decoding accuracy is limited, affecting the application of the MI-BCI. In this paper, we propose the EISATC-Fusion model for MI EEG decoding, consisting of inception block, multi-head self-attention (MSA), temporal convolutional network (TCN), and layer fusion. Specifically, we design a DS Inception block to extract multi-scale frequency band information. And design a new cnnCosMSA module based on CNN and cos attention to solve the attention collapse and improve the interpretability of the model. The TCN module is improved by the depthwise separable convolution to reduces the parameters of the model. The layer fusion consists of feature fusion and decision fusion, fully utilizing the features output by the model and enhances the robustness of the model. We improve the two-stage training strategy for model training. Early stopping is used to prevent model overfitting, and the accuracy and loss of the validation set are used as indicators for early stopping. The proposed model achieves within-subject classification accuracies of 84.57% and 87.58% on BCI Competition IV Datasets 2a and 2b, respectively. And the model achieves cross-subject classification accuracies of 67.42% and 71.23% (by transfer learning) when training the model with two sessions and one session of Dataset 2a, respectively. The interpretability of the model is demonstrated through weight visualization method.

Recruiting neural field theory for data augmentation in a motor imagery brain-computer interface.

We introduce a novel approach to training data augmentation in brain-computer interfaces (BCIs) using neural field theory (NFT) applied to EEG data from motor imagery tasks. BCIs often suffer from limited accuracy due to a limited amount of training data. To address this, we leveraged a corticothalamic NFT model to generate artificial EEG time series as supplemental training data. We employed the BCI competition IV '2a' dataset to evaluate this augmentation technique. For each individual, we fitted the model to common spatial patterns of each motor imagery class, jittered the fitted parameters, and generated time series for data augmentation. Our method led to significant accuracy improvements of over 2% in classifying the "total power" feature, but not in the case of the "Higuchi fractal dimension" feature. This suggests that the fit NFT model may more favorably represent one feature than the other. These findings pave the way for further exploration of NFT-based data augmentation, highlighting the benefits of biophysically accurate artificial data.

Development of an Adaptive Artifact Subspace Reconstruction Based on Hebbian/Anti-Hebbian Learning Networks for Enhancing BCI Performance.

Brain-computer interface (BCI) actively translates the brain signals into executable actions by establishing direct communication between the human brain and external devices. Recording brain activity through electroencephalography (EEG) is generally contaminated with both physiological and nonphysiological artifacts, which significantly hinders the BCI performance. Artifact subspace reconstruction (ASR) is a well-known statistical technique that automatically removes artifact components by determining the rejection threshold based on the initial reference EEG segment in multichannel EEG recordings. In real-world applications, the fixed threshold may limit the efficacy of the artifact correction, especially when the quality of the reference data is poor. This study proposes an adaptive online ASR technique by integrating the Hebbian/anti-Hebbian neural networks into the ASR algorithm, namely, principle subspace projection ASR (PSP-ASR) and principal subspace whitening ASR (PSW-ASR) that segmentwise self-organize the artifact subspace by updating the synaptic weights according to the Hebbian and anti-Hebbian learning rules. The effectiveness of the proposed algorithm is compared to the conventional ASR approaches on benchmark EEG dataset and three BCI frameworks, including steady-state visual evoked potential (SSVEP), rapid serial visual presentation (RSVP), and motor imagery (MI) by evaluating the root-mean-square error (RMSE), the signal-to-noise ratio (SNR), the Pearson correlation, and classification accuracy. The results demonstrated that the PSW-ASR algorithm effectively removed the EEG artifacts and retained the activity-specific brain signals compared to the PSP-ASR, standard ASR (Init-ASR), and moving-window ASR (MW-ASR) methods, thereby enhancing the SSVEP, RSVP, and MI BCI performances. Finally, our empirical results from the PSW-ASR algorithm suggested the choice of an aggressive cutoff range of c = 1-10 for activity-specific BCI applications and a moderate range of for the benchmark dataset and general BCI applications.

EEG Decoding Based on Normalized Mutual Information for Motor Imagery Brain–Computer Interfaces

EEG Decoding Based on Normalized Mutual Information for Motor Imagery Brain–Computer Interfaces

Effects of transcranial magnetic stimulation on cortical structures during motor imagination performance in the brain-computer interface

The aftermath of a stroke can frequently result in impaired motor functions causing problems performing habitual limb movements. These movement disorders stem from damage to the cerebral cortex and disruptions to neuronal connections in the central pyramidal pathways [1]. Restoring motor skills following a stroke is a time-consuming and challenging process, requiring resources from both external medical sources and the patient. Despite these challenges, it is still feasible for patients to regain control over their limb movements. The common approach for stroke neurorehabilitation consists of therapeutic physical activities and kinesiotherapy. These methods rely on afferent information during motor tasks to repair connections between intact brain areas [2]. Through exercising the impacted limbs, synaptic rearrangement happens in the cortex, awakening dormant neurons and increasing cortical areas adjacent to inactive regions. Although these techniques can partially restore movement, a significant number of stroke patients still face impairments. It is crucial to acknowledge the limitations of the current interventions and explore new approaches for better outcomes. Traditional rehabilitation methods often do not fully restore movement control, prompting researchers to explore alternative approaches. Motor-imagery-based brain-computer interfaces (BCIs) have gained attention recently [3, 4]. They allow for integrating various feedback mechanisms, which can be combined with upper and lower limb exoskeletons. The level of control a subject has over a BCI system directly affects the recovery process [5]. Introducing transcranial magnetic stimulation (TMS) into motor-imagery BCIs shows promise for developing a unified and highly efficient approach to post-stroke rehabilitation. Twenty-nine healthy adult participants (21 females) with a mean age of 20.93±2.14 years were recruited for this experiment. All participants were right-handed and had no prior experience with brain-computer interfaces. Ethical approval was obtained from the local ethics committee of Lobachevsky State University of Nizhny Novgorod (ethical approval No. 2, dated 03/19/2021), and written informed consent was obtained from all participants. Subjects were randomly assigned to receive either a sham intervention (n=15) or active TMS (n=14). The experiment displayed tasks on a 24-inch LCD screen placed 2 meters away. Participants sat in a reclining chair with hands on adjustable armrests for EEG signal recording. Motor-imagery-based BCI training took place over two days with four daily tasks: motor performance, quasi-motor, and two motor imaginations of the dominant hand. EEG activity was recorded prior to and after completing the tasks. Each task comprised of 20 trials, each lasting 10 seconds. TMS was administered in between two motor imagination tasks, and a rest period of 2 minutes followed the stimulation. The researcher objectively monitored electromyography (EMG) in real-time during motor tasks. Certified NVX 52 amplifiers with 32 Cl/Ag electrodes positioned according to the international 10-10 system recorded electroencephalography (EEG) signals. The EEG was digitized at a sampling rate of 1000 Hz and filtered with a 50 Hz Notch filter. Disposable electrodes were used to record EMG data from musculus flexor digitorum superficialis on the right hand. EMG signals were digitized at 1000 Hz and filtered using a 50 Hz Notch filter. TMS was applied to the dorsolateral prefrontal cortex (dlPFC) using a figure-8 coil connected to a Neuro MS/D magnetic stimulator. Sham stimulation was performed with the same parameters, except the coil was tilted 90 degrees to simulate the sound of real stimulation. A nonparametric permutation test was conducted to assess the statistically significant dissimilarities between rest periods and periods subsequent to conducting motor imagery after TMS. Significant clusters of the highest event-related desynchronization (ERD) were observed. The analysis demonstrated a significant negative cluster in the theta rhythm (0–6 Hz) that was distinct from the magnetic stimulation site. Similar significant ERD clusters were identified during the first and second series of imaginary movements. The study examined alterations in motor cortex function following targeted rTMS. The results demonstrated that TMS using particular parameters resulted in the pre-activation of brain regions comparable to those stimulated during motor imagery. The implementation of this stimulation protocol increased activity in cortical regions related to motor imagery. The outcomes propose the probable effectiveness of TMS in elevating motor cortex activation during the rehabilitation process.

TSPNet: a time-spatial parallel network for classification of EEG-based multiclass upper limb motor imagery BCI.

The classification of electroencephalogram (EEG) motor imagery signals has emerged as a prominent research focus within the realm of brain-computer interfaces. Nevertheless, the conventional, limited categories (typically just two or four) offered by brain-computer interfaces fail to provide an extensive array of control modes. To address this challenge, we propose the Time-Spatial Parallel Network (TSPNet) for recognizing six distinct categories of upper limb motor imagery. Within TSPNet, temporal and spatial features are extracted separately, with the time dimension feature extractor and spatial dimension feature extractor performing their respective functions. Following this, the Time-Spatial Parallel Feature Extractor is employed to decouple the connection between temporal and spatial features, thus diminishing feature redundancy. The Time-Spatial Parallel Feature Extractor deploys a gating mechanism to optimize weight distribution and parallelize time-spatial features. Additionally, we introduce a feature visualization algorithm based on signal occlusion frequency to facilitate a qualitative analysis of TSPNet. In a six-category scenario, TSPNet achieved an accuracy of 49.1% ± 0.043 on our dataset and 49.7% ± 0.029 on a public dataset. Experimental results conclusively establish that TSPNet outperforms other deep learning methods in classifying data from these two datasets. Moreover, visualization results vividly illustrate that our proposed framework can generate distinctive classifier patterns for multiple categories of upper limb motor imagery, discerned through signals of varying frequencies. These findings underscore that, in comparison to other deep learning methods, TSPNet excels in intention recognition, which bears immense significance for non-invasive brain-computer interfaces.