The brain-computer interface (BCI) based on motor imagery electroencephalography (MI-EEG) enables direct information interaction between the human brain and external devices. In this paper, a multi-scale EEG feature extraction convolutional neural network model based on time series data enhancement is proposed for decoding MI-EEG signals. First, an EEG signals augmentation method was proposed that could increase the information content of training samples without changing the length of the time series, while retaining its original features completely. Then, multiple holistic and detailed features of the EEG data were adaptively extracted by multi-scale convolution module, and the features were fused and filtered by parallel residual module and channel attention. Finally, classification results were output by a fully connected network. The application experimental results on the BCI Competition IV 2a and 2b datasets showed that the proposed model achieved an average classification accuracy of 91.87% and 87.85% for the motor imagery task, respectively, which had high accuracy and strong robustness compared with existing baseline models. The proposed model does not require complex signals pre-processing operations and has the advantage of multi-scale feature extraction, which has high practical application value.
Aiming at feature selection problem of motor imagery task in brain computer interface (BCI), an algorithm based on mutual information and principal component analysis (PCA) for electroencephalogram (EEG) feature selection is presented. This algorithm introduces the category information, and uses the sum of mutual information matrices between features under different motor imagery category to replace the covariance matrix. The eigenvectors of the sum matrix represent the direction of the principal components and the eigenvalues of the sum matrix are used to determine the dimensionality of principal components. 2005 International BCI competition data set was used in our experiments, and four feature extraction methods were adopted, i. e. power spectrum estimation, continuous wavelet transform, wavelet packet decomposition and Hjorth parameters. The proposed feature selection algorithm was adopted to select and combine the most useful features for classification. The results showed that relative to the PCA algorithm, our algorithm had better performance in dimensionality reduction and in classification accuracy with the assistance of support vector machine classifier under the same dimensionality of principal components.
Motor imagery is often used in the fields of sports training and neurorehabilitation for its advantages of being highly targeted, easy to learn, and requiring no special equipment, and has become a major research paradigm in cognitive neuroscience. Transcranial direct current stimulation (tDCS), an emerging neuromodulation technique, modulates cortical excitability, which in turn affects functions such as locomotion. However, it is unclear whether tDCS has a positive effect on motor imagery task states. In this paper, 16 young healthy subjects were included, and the electroencephalogram (EEG) signals and near-infrared spectrum (NIRS) signals of the subjects were collected when they were performing motor imagery tasks before and after receiving tDCS, and the changes in multiscale sample entropy (MSE) and haemoglobin concentration were calculated and analyzed during the different tasks. The results found that MSE of task-related brain regions increased, oxygenated haemoglobin concentration increased, and total haemoglobin concentration rose after tDCS stimulation, indicating that tDCS increased the activation of task-related brain regions and had a positive effect on motor imagery. This study may provide some reference value for the clinical study of tDCS combined with motor imagery.
Regarding to the channel selection problem during the classification of electroencephalogram (EEG) signals, we proposed a novel method, Relief-SBS, in this paper. Firstly, the proposed method performed EEG channel selection by combining the principles of Relief and sequential backward selection (SBS) algorithms. And then correlation coefficient was used for classification of EEG signals. The selected channels that achieved optimal classification accuracy were considered as optimal channels. The data recorded from motor imagery task experiments were analyzed, and the results showed that the channels selected with our proposed method achieved excellent classification accuracy, and also outperformed other feature selection methods. In addition, the distribution of the optimal channels was proved to be consistent with the neurophysiological knowledge. This demonstrates the effectiveness of our method. It can be well concluded that our proposed method, Relief-SBS, provides a new way for channel selection.
In the research of non-invasive brain-computer interface (BCI), independent component analysis (ICA) has been considered as a promising method of electroencephalogram (EEG) preprocessing and feature enhancement. However, there have been few investigations and implements about online ICA-BCI system up till now. This paper reports the investigation of the ICA-based motor imagery BCI (MIBCI) system, combining the characteristics of unsupervised learning of ICA and event-related desynchronization (ERD) related to motor imagery. We constructed a simple and practical method of ICA spatial filter calculation and discriminate criterion of three-type motor imageries in the study. To validate the online performance of proposed algorithms, an ICA-MIBCI experimental system was fully established based on NeuroScan EEG amplifier and VC++ platform. Four subjects participated in the experiment of MIBCI testing and two of them took part in the online experiment. The average classification accuracies of the three-type motor imageries reached 89.78% and 89.89% in the offline and online testing, respectively. The experimental results showed that the proposed algorithm produced high classification accuracy and required less time consumption, which would have a prospect of cross platform application.
One of the key problems of brain-computer interfaces (BCI) is low signal-to-noise ratio (SNR) of electroencephalogram (EEG) signals. It affects recognition performance. To remove the artifact and noise, block under-determined blind source separation method based on the small number of channels is proposed in this paper. The non-stationary EEG signals are turned into block stationary signals by piecewise. The mixing matrix is estimated by the second-order under-determined blind mixing matrix identification. Then, the beamformer based on minimum mean square error separates the original sources of signals. Eventually, the reconstructed EEG for mixed signals removes the unwanted components of source signals to achieve suppressing artifact. The experiment results on the real motor imagery BCI indicated that the block under-determined blind source separation method could reconstruct signals and remove artifact effectively. The accuracy of motor imagery task of BCI has been greatly improved.
Motor imagery electroencephalogram (EEG) signals are non-stationary time series with a low signal-to-noise ratio. Therefore, the single-channel EEG analysis method is difficult to effectively describe the interaction characteristics between multi-channel signals. This paper proposed a deep learning network model based on the multi-channel attention mechanism. First, we performed time-frequency sparse decomposition on the pre-processed data, which enhanced the difference of time-frequency characteristics of EEG signals. Then we used the attention module to map the data in time and space so that the model could make full use of the data characteristics of different channels of EEG signals. Finally, the improved time-convolution network (TCN) was used for feature fusion and classification. The BCI competition IV-2a data set was used to verify the proposed algorithm. The experimental results showed that the proposed algorithm could effectively improve the classification accuracy of motor imagination EEG signals, which achieved an average accuracy of 83.03% for 9 subjects. Compared with the existing methods, the classification accuracy of EEG signals was improved. With the enhanced difference features between different motor imagery EEG data, the proposed method is important for the study of improving classifier performance.
The brain-computer interface (BCI) based on motor imagery electroencephalography (EEG) shows great potential in neurorehabilitation due to its non-invasive nature and ease of use. However, motor imagery EEG signals have low signal-to-noise ratios and spatiotemporal resolutions, leading to low decoding recognition rates with traditional neural networks. To address this, this paper proposed a three-dimensional (3D) convolutional neural network (CNN) method that learns spatial-frequency feature maps, using Welch method to calculate the power spectrum of EEG frequency bands, converted time-series EEG into a brain topographical map with spatial-frequency information. A 3D network with one-dimensional and two-dimensional convolutional layers was designed to effectively learn these features. Comparative experiments demonstrated that the average decoding recognition rate reached 86.89%, outperforming traditional methods and validating the effectiveness of this approach in motor imagery EEG decoding.
Mental rotation cognitive tasks based on motor imagery (MI) have excellent predictability for individual’s motor imagery ability. In order to explore the relationship between motor imagery and behavioral data, in this study, we asked 10 right-handed male subjects to participate in the experiments of mental rotation tasks based on corresponding body parts pictures, and we therefore obtained the behavioral effects according to their reaction time (RT) and accuracy (ACC). Later on, we performed Pearson correlation analysis between the behavioral data and the scores of the Movement Imagery Questionnaire-Revised(MIQ-R). For each subject, the results showed significant angular and body location effect in the process of mental rotation. For all subjects, the results showed that there were correlations between the behavioral data and the scores of MIQ-R. Subjects who needed the longer reaction time represented lower motor imagery abilities in the same test, and vice versa. This research laid the foundation for the further study on brain electrophysiology in the process of mental rotation based on MI.
The research on brain functional mechanism and cognitive status based on brain network has the vital significance. According to a time–frequency method, partial directed coherence (PDC), for measuring directional interactions over time and frequency from scalp-recorded electroencephalogram (EEG) signals, this paper proposed dynamic PDC (dPDC) method to model the brain network for motor imagery. The parameters attributes (out-degree, in-degree, clustering coefficient and eccentricity) of effective network for 9 subjects were calculated based on dataset from BCI competitions IV in 2008, and then the interaction between different locations for the network character and significance of motor imagery was analyzed. The clustering coefficients for both groups were higher than those of the random network and the path length was close to that of random network. These experimental results show that the effective network has a small world property. The analysis of the network parameter attributes for the left and right hands verified that there was a significant difference on ROI2 (P = 0.007) and ROI3 (P = 0.002) regions for out-degree. The information flows of effective network based dPDC algorithm among different brain regions illustrated the active regions for motor imagery mainly located in fronto-central regions (ROI2 and ROI3) and parieto-occipital regions (ROI5 and ROI6). Therefore, the effective network based dPDC algorithm can be effective to reflect the change of imagery motor, and can be used as a practical index to research neural mechanisms.