Minpeng Xu

A hybrid BCI speller paradigm combining P300 potential and the SSVEP blocking feature

OBJECTIVE Hybrid brain-computer interfaces (BCIs) have been proved to be more effective in mental control by combining another channel of physiologic control signals. Among those studies, little attention has been paid to the combined use of a steady-state visual evoked potential (SSVEP) and P300 potential, both providing the fastest and the most reliable EEG based BCIs. In this paper, a novel hybrid BCI speller is developed to elicit P300 potential and SSVEP blocking (SSVEP-B) distinctly and simultaneously with the same target stimulus. APPROACH Twelve subjects were involved in the study and every one performed offline spelling twice in succession with two different speller paradigms for comparison: hybrid speller and control P300-speller. Feature analysis was adopted from the view of time domain, frequency domain and spatial distribution; the performances were evaluated by character accuracy and information transfer rate (ITR). MAIN RESULTS Signal analysis of the hybrid paradigm shows that SSVEPs are an evident EEG component during the nontarget phase but are dismissed and replaced by P300 potentials after target stimuli. The absence of an SSVEP, called SSVEP-B, mostly appearing in channel Oz, presents a sharp distinction between target responses and nontarget responses. The r(2) value of SSVEP-B in channel Oz is comparable to that of P300 in channel Cz. Compared with the control P300-speller, the hybrid speller achieves significantly higher accuracy and ITR with combined features. SIGNIFICANCE The results indicate that the combination of P300 with an SSVEP-B improves target discrimination greatly; the proposed hybrid paradigm is superior to the control paradigm in spelling performance. Thus, our findings provide a new approach to improve BCI performances.

Channel selection based on phase measurement in P300-based brain-computer interface.

Most EEG-based brain-computer interface (BCI) paradigms include specific electrode positions. As the structures and activities of the brain vary with each individual, contributing channels should be chosen based on original records of BCIs. Phase measurement is an important approach in EEG analyses, but seldom used for channel selections. In this paper, the phase locking and concentrating value-based recursive feature elimination approach (PLCV-RFE) is proposed to produce robust-EEG channel selections in a P300 speller. The PLCV-RFE, deriving from the phase resetting mechanism, measures the phase relation between EEGs and ranks channels by the recursive strategy. Data recorded from 32 electrodes on 9 subjects are used to evaluate the proposed method. The results show that the PLCV-RFE substantially reduces channel sets and improves recognition accuracies significantly. Moreover, compared with other state-of-the-art feature selection methods (SSNRSF and SVM-RFE), the PLCV-RFE achieves better performance. Thus the phase measurement is available in the channel selection of BCI and it may be an evidence to indirectly support that phase resetting is at least one reason for ERP generations.

A visual parallel-BCI speller based on the time–frequency coding strategy

OBJECTIVE Spelling is one of the most important issues in brain-computer interface (BCI) research. This paper is to develop a visual parallel-BCI speller system based on the time-frequency coding strategy in which the sub-speller switching among four simultaneously presented sub-spellers and the character selection are identified in a parallel mode. APPROACH The parallel-BCI speller was constituted by four independent P300+SSVEP-B (P300 plus SSVEP blocking) spellers with different flicker frequencies, thereby all characters had a specific time-frequency code. To verify its effectiveness, 11 subjects were involved in the offline and online spellings. A classification strategy was designed to recognize the target character through jointly using the canonical correlation analysis and stepwise linear discriminant analysis. MAIN RESULTS Online spellings showed that the proposed parallel-BCI speller had a high performance, reaching the highest information transfer rate of 67.4 bit min(-1), with an average of 54.0 bit min(-1) and 43.0 bit min(-1) in the three rounds and five rounds, respectively. SIGNIFICANCE The results indicated that the proposed parallel-BCI could be effectively controlled by users with attention shifting fluently among the sub-spellers, and highly improved the BCI spelling performance.

Incorporation of Inter-Subject Information to Improve the Accuracy of Subject-Specific P300 Classifiers

Although the inter-subject information has been demonstrated to be effective for a rapid calibration of the P300-based brain-computer interface (BCI), it has never been comprehensively tested to find if the incorporation of heterogeneous data could enhance the accuracy. This study aims to improve the subject-specific P300 classifier by adding other subjects data. A classifier calibration strategy, weighted ensemble learning generic information (WELGI), was developed, in which elementary classifiers were constructed by using both the intra- and inter-subject information and then integrated into a strong classifier with a weight assessment. 55 subjects were recruited to spell 20 characters offline using the conventional P300-based BCI, i.e. the P300-speller. Four different metrics, the P300 accuracy and precision, the round accuracy, and the character accuracy, were performed for a comprehensive investigation. The results revealed that the classifier constructed on the training dataset in combination with adding other subjects data was significantly superior to that without the inter-subject information. Therefore, the WELGI is an effective classifier calibration strategy which uses the inter-subject information to improve the accuracy of subject-specific P300 classifiers, and could also be applied to other BCI paradigms.

Randomly dividing homologous samples leads to overinflated accuracies for emotion recognition

There are numerous studies measuring the brain emotional status by analyzing EEGs under the emotional stimuli that have occurred. However, they often randomly divide the homologous samples into training and testing groups, known as randomly dividing homologous samples (RDHS), despite considering the impact of the non-emotional information among them, which would inflate the recognition accuracy. This work proposed a modified method, the integrating homologous samples (IHS), where the homologous samples were either used to build a classifier, or to be tested. The results showed that the classification accuracy was much lower for the IHS than for the RDHS. Furthermore, a positive correlation was found between the accuracy and the overlapping rate of the homologous samples. These findings implied that the overinflated accuracy did exist in those previous studies where the RDHS method was employed for emotion recognition. Moreover, this study performed a feature selection for the IHS condition based on the support vector machine-recursive feature elimination, after which the average accuracies were greatly improved to 85.71% and 77.18% in the picture-induced and video-induced tasks, respectively.

Inter-subject information contributes to the ERP classification in the P300 speller

This study aims to investigate whether the inter-subject information is beneficial to the event-related potential (ERP) classification in the P300-speller. To this end, a classification strategy of weighted ensemble learning generic information (WELGI) was developed, in which the base classifiers constructed by combining both intra- and inter-subject information were integrated into a strong classifier with weight assessments. To verify the algorithms validity, 55 subjects were recruited to spell 20 characters offline by using the conventional P300-speller paradigm, and the ERP accuracy and precision were investigated. Compared with the traditional classification strategy only using the intra-subject information, the WELGI could achieve significantly higher ERP accuracy and precision. It was demonstrated that the inter-subject information was beneficial to the ERP classification in the P300-speller.

Feature selection study of P300 speller using support vector machine

P300 speller is a traditional brain computer interface paradigm and focused by lots of current BCI researches. In this paper a support vector machine based recursive feature elimination method was adapted to select the optimal channels for character recognition. The margin distance between target and nontarget stimulus in feature space was evaluated by training SVM classifier and then the features from single channel were eliminated one by one, eventually, channel set provided best recognition performance was left as the optimal set. The results showed that using optimal channel set would achieve a higher recognition correct ratio compared with no channel eliminating. Furthermore the optimal features localized on parietal and occipital areas, on which not only P300 components but VEP components also present a high amplitude waveform. It may suggest that row/column intensification in speller matrix arouses a visual evoked potential and contributes a lot to character identification as well as P300.

Fast detection of covert visuospatial attention using hybrid N2pc and SSVEP features

OBJECTIVE Detecting the shift of covert visuospatial attention (CVSA) is vital for gaze-independent brain-computer interfaces (BCIs), which might be the only communication approach for severely disabled patients who cannot move their eyes. Although previous studies had demonstrated that it is feasible to use CVSA-related electroencephalography (EEG) features to control a BCI system, the communication speed remains very low. This study aims to improve the speed and accuracy of CVSA detection by fusing EEG features of N2pc and steady-state visual evoked potential (SSVEP). APPROACH A new paradigm was designed to code the left and right CVSA with the N2pc and SSVEP features, which were then decoded by a classification strategy based on canonical correlation analysis. Eleven subjects were recruited to perform an offline experiment in this study. Temporal waves, amplitudes, and topographies for brain responses related to N2pc and SSVEP were analyzed. The classification accuracy derived from the hybrid EEG features (SSVEP and N2pc) was compared with those using the single EEG features (SSVEP or N2pc). MAIN RESULTS The N2pc could be significantly enhanced under certain conditions of SSVEP modulations. The hybrid EEG features achieved significantly higher accuracy than the single features. It obtained an average accuracy of 72.9% by using a data length of 400 ms after the attention shift. Moreover, the average accuracy reached ∼80% (peak values above 90%) when using 2 s long data. SIGNIFICANCE The results indicate that the combination of N2pc and SSVEP is effective for fast detection of CVSA. The proposed method could be a promising approach for implementing a gaze-independent BCI.

Enhancing performance of P300-Speller under mental workload by incorporating dual-task data during classifier training

As one of the most important brain-computer interface (BCI) paradigms, P300-Speller was shown to be significantly impaired once applied in practical situations due to effects of mental workload. This study aims to provide a new method of building training models to enhance performance of P300-Speller under mental workload. Three experiment conditions based on row-column P300-Speller paradigm were performed including speller-only, 3-back-speller and mental-arithmetic-speller. Data under dual-task conditions were introduced to speller-only data respectively to build new training models. Then performance of classifiers with different models was compared under the same testing condition. The results showed that when tasks of imported training data and testing data were the same, character recognition accuracies and round accuracies of P300-Speller with mixed-data training models significantly improved (FDR, p < 0.005). When they were different, performance significantly improved when tested on mental-arithmetic-speller (FDR, p < 0.05) while the improvement was modest when tested on n-back-speller (FDR, p < 0.1). The analysis of ERPs revealed that ERP difference between training data and testing data was significantly diminished when the dual-task data was introduced to training data (FDR, p < 0.05). The new method of training classifier on mixed data proved to be effective in enhancing performance of P300-Speller under mental workload, confirmed the feasibility to build a universal training model and overcome the effects of mental workload in its practical applications.

The parallel-BCI speller based on the P300 and SSVEP features

This paper developed a parallel-BCI speller system, consisting of four simultaneously presented sub-spellers, where the sub-speller switching and character selection were simultaneously identified through decoding the dominant frequency of steady-state visual-evoked potential (SSVEP) and the time of P300 occurrence in a parallel mode. Five subjects took part in the offline and online experiments. The canonical correlation analysis (CCA) and stepwise linear discriminant analysis (SWLDA) were jointly used to recognize the target character. Online tests showed that the parallel-BCI speller reached the peak information transfer rate (ITR) of 67.4 bit/min with an average of 48.0 bit/min. The results indicate that the proposed parallel-BCI system can be effectively controlled by users with attention shifting fluently among the sub-spellers, and highly improve the BCI spelling performance.