Facilitating effects of transcranial direct current stimulation on motor imagery brain-computer interface with robotic feedback for stroke rehabilitation.

OBJECTIVE: To investigate the efficacy and effects of transcranial direct current stimulation (tDCS) on motor imagery brain-computer interface (MI-BCI) with robotic feedback for stroke rehabilitation. DESIGN: A sham-controlled, randomized controlled trial. SETTING: Patients recruited through a hospital stroke rehabilitation program. PARTICIPANTS: Subjects (N=19) who incurred a stroke 0.8 to 4.3 years prior, with moderate to severe upper extremity functional impairment, and passed BCI screening. INTERVENTIONS: Ten sessions of 20 minutes of tDCS or sham before 1 hour of MI-BCI with robotic feedback upper limb stroke rehabilitation for 2 weeks. Each rehabilitation session comprised 8 minutes of evaluation and 1 hour of therapy. MAIN OUTCOME MEASURES: Upper extremity Fugl-Meyer Motor Assessment (FMMA) scores measured end-intervention at week 2 and follow-up at week 4, online BCI accuracies from the evaluation part, and laterality coefficients of the electroencephalogram (EEG) from the therapy part of the 10 rehabilitation sessions. RESULTS: FMMA score improved in both groups at week 4, but no intergroup differences were found at any time points. Online accuracies of the evaluation part from the tDCS group were significantly higher than those from the sham group. The EEG laterality coefficients from the therapy part of the tDCS group were significantly higher than those of the sham group. CONCLUSIONS: The results suggest a role for tDCS in facilitating motor imagery in stroke.

A hierarchical dynamic Bayesian learning network for EMG-based early prediction of voluntary movement intention.

Decoding human action intention prior to motion onset with surface electromyograms (sEMG) is an emerging neuroengineering topic with interesting clinical applications such as intelligent control of powered prosthesis/exoskeleton devices. Despite extensive prior works in the related fields, it remains a technical challenge due to considerable variability of complex multi-muscle activation patterns in terms of volatile spatio-temporal characteristics. To address this issue, we first hypothesize that the inherent variability of the idle state immediately preceding the motion initiation needs to be addressed explicitly. We therefore design a hierarchical dynamic Bayesian learning network model that integrates an array of Gaussian mixture model - hidden Markov models (GMM-HMMs), where each GMM-HMM learns the multi-sEMG processes either during the idle state, or during the motion initiation phase of a particular motion task. To test the hypothesis and evaluate the new learning network, we design and build a upper-limb sEMG-joystick motion study system, and collect data from 11 healthy volunteers. The data collection protocol adapted from the psychomotor vigilance task includes repeated and randomized binary hand motion tasks (push or pull) starting from either of two designated idle states: relaxed (with minimal muscle tones), or prepared (with muscle tones). We run a series of cross-validation tests to examine the performance of the method in comparison with the conventional techniques. The results suggest that the idle state recognition favors the dynamic Bayesian model over a static classification model. The results also show a statistically significant improvement in motion prediction accuracy by the proposed method (93.83±6.41%) in comparison with the conventional GMM-HMM method (89.71±8.98%) that does not explicitly account for the idle state. Moreover, we examine the progress of prediction accuracy over the course of motion initiation and identify the important hidden states that warrant future research.

Information sparseness in cortical microelectrode channels while decoding movement direction using an artificial neural network.

Implanted microelectrode arrays can directly pick up electrode signals from the primary motor cortex (M1) during movement, and brain-machine interfaces (BMIs) can decode these signals to predict the directions of contemporaneous movements. However, it is not well known how much each individual input is responsible for the overall performance of a BMI decoder. In this paper, we seek to quantify how much each channel contributes to an artificial neural network (ANN)-based decoder, by measuring how much the removal of each individual channel degrades the accuracy of the output. If information on movement direction was equally distributed among channels, then the removal of one would have a minimal effect on decoder accuracy. On the other hand, if that information was distributed sparsely, then the removal of specific information-rich channels would significantly lower decoder accuracy. We found that for most channels, their removal did not significantly affect decoder performance. However, for a subset of channels (16 out of 61), removing them significantly reduced the decoder accuracy. This suggests that information is not uniformly distributed among the recording channels. We propose examining these channels further to optimize BMIs more effectively, as well as understand how M1 functions at the neuronal level.

Scalp EEG-Based Pain Detection Using Convolutional Neural Network.

Pain is an integrative phenomenon coupled with dynamic interactions between sensory and contextual processes in the brain, often associated with detectable neurophysiological changes. Recent advances in brain activity recording tools and machine learning technologies have intrigued research and development of neurocomputing techniques for objective and neurophysiology-based pain detection. This paper proposes a pain detection framework based on Electroencephalogram (EEG) and deep convolutional neural networks (CNN). The feasibility of CNN is investigated for distinguishing induced pain state from resting state in the recruitment of 10 chronic back pain patients. The experimental study recorded EEG signals in two phases: 1. movement stimulation (MS), where induces back pain by executing predefined movement tasks; 2. video stimulation (VS), where induces back pain perception by watching a set of video clips. A multi-layer CNN classifies the EEG segments during the resting state and the pain state. The novel approach offers high and robust performance and hence is significant in building a powerful pain detection algorithm. The area under the receiver operating characteristic curve (AUC) of our approach is 0.83 ± 0.09 and 0.81 ± 0.15, in MS and VS, respectively, higher than the state-of-the-art approaches. The sub-brain-areas are also analyzed, to examine distinct brain topographies relevant for pain detection. The results indicate that MS-induced pain tends to evoke a generalized brain area, while the evoked area is relatively partial under VS-induced pain. This work may provide a new solution for researchers and clinical practitioners on pain detection.

An Affective Interaction System using Virtual Reality and Brain-Computer Interface.

Affective Computing is a multidisciplinary area of research that allows computers to perform human emotion recognition, with potential applications in areas such as healthcare, gaming and intuitive human computer interface design. Hence, this paper proposes an affective interaction system using dry EEG-based Brain-Computer Interface and Virtual Reality (BCI-VR). The proposed BCI-VR system integrates existing low-cost consumer devices such as an EEG headband with frontal and temporal dry electrodes for brain signal acquisition, and a low-cost VR headset that houses an Android handphone. The handphone executes an in-house developed software that connects wirelessly to the headband, processes the acquired EEG signals, and displays VR content to elicit emotional responses. The proposed BCI-VR system was used to collect EEG data from 13 subjects while they watched VR content that elicits positive or negative emotional responses. EEG bandpower features were extracted to train Linear Discriminant and Support Vector Machine classifiers. The classification performances of these classifiers on this dataset and the results of a public dataset (SEED-IV) are then evaluated. The results in classifying positive vs negative emotions in both datasets (~66% for 2-class) show promise that positive and negative emotions can be detected by the proposed low cost BCIVR system, yielding nearly the same performance on the public dataset that used wet EEG electrodes. Hence the results show promise of the proposed BCI-VR system for real-time affective interaction applications in future.

Wireless Multi-sensor Physio-Motion Measurement and Synchronization System and Method for HRI Research.

There is a strong demand for acquisition, processing and understanding of a variety of physiological and behavioral signals from the measurements in human-robot interface (HRI). However, multiple data streams from these measurements bring considerable challenges for their synchronizations, either for offline analysis or for online HRI applications, especially when the sensors are wirelessly connected, without synchronization mechanisms, such as a network-time-protocol. In this paper, we presented a full wireless multi-modality sensor system comprising biopotential measurements such as EEG, EMG and inertial parameter data of articulated body-limb motions. In the paper, we propose two methods to synchronize and calibrate the transmission latencies from different wireless channels. The first method employs the traditional artificial electrical timing signal. The other one employs the force-acceleration relationship governed by Newton's Second Law to facilitate reconstruction of the sample-to-sample alignment between the two wireless sensors. The measured latencies are investigated and the result show that they could be determined consistently and accurately by the devised techniques.

Infant EEG Band Power Analysis at 6 Months and 18 Months.

Neural development of infants has drawn increasing research interests from the community. In this paper, we investigated the frequency band power of 112 infants who participated in an auditory oddball experiment, and the visual expectation (VE) score of 177 infants who went through a visual expectation paradigm test. Analysis found that the frequency band power decreases in the delta and theta bands, and increases in the alpha and beta bands when the infants grow up from 6 months old to 18 months old. We also proposed a sustainability index to measure the capability of a subject to maintain their band power in the auditory oddball experiment when infants grow up from 6 months old to 18 months old. Analysis shows that the sustainability index increased significantly in the alpha and beta band, decreased in the delta and theta bands. Correlation between the VE score and frequency band power was investigated on 47 infants who participated in both auditory oddball experiment and visual expectation paradigm test. Analysis shows that the reaction speed to stimulus have statistical a significant correlation with the changes of band power and sustainability index in posterior and temporal section, and in the higher frequency bands.

Decoding movement direction from cortical microelectrode recordings using an LSTM-based neural network.

Brain-machine interfaces (BMIs) allow individuals to communicate with computers using neural signals, and Kalman Filter (KF) are prevailingly used to decode movement directions from these neural signals. In this paper, we implemented a multi-layer long short-term memory (LSTM)based artificial neural network (ANN) for decoding BMI neural signals. We collected motor cortical neural signals from a nonhuman primate (NHP), implanted with microelectrode array (MEA) while performing a directional joystick task. Next, we compared the LSTM model in decoding the joystick trajectories from the neural signals against the prevailing KF model. The results showed that the LSTM model yielded significantly improved decoding accuracy measured by mean correlation coefficient (0.84, p < 10-7) than the KF model (0.72). In addition, using a principal component analysis (PCA)-based dimensionality reduction technique yielded slightly deteriorated accuracies for both the LSTM (0.80) and KF (0.70) models, but greatly reduced the computational complexity. The results showed that the LSTM decoding model holds promise to improve decoding in BMIs for paralyzed humans.

Asynchronous P300-based brain-computer interfaces: a computational approach with statistical models.

Asynchronous control is an important issue for brain-computer interfaces (BCIs) working in real-life settings, where the machine should determine from brain signals not only the desired command but also when the user wants to input it. In this paper, we propose a novel computational approach for robust asynchronous control using electroencephalogram (EEG) and a P300-based oddball paradigm. In this approach, we first address the mathematical modeling of target P300, nontarget P300, and noncontrol signals, by using Gaussian distribution models in a support vector margin space. Furthermore, we derive a method to compute the likelihood of control state in a time window of EEG. Finally, we devise a recursive algorithm to detect control states in ongoing EEG for online application. We conducted experiments with four subjects to study both the asynchronous BCI's receiver operating characteristics and its performance in actual online tests. The results show that the BCI is able to achieve an averaged information transfer rate of approximately 20 b/min at a low false positive rate (one event per minute).

EEG-based discrimination of different cognitive workload levels from mental arithmetic.

Cognitive workload, which is the level of mental effort required for a cognitive task, can be assessed by monitoring the changes in neurophysiological measures such as electroencephalogram (EEG). This study investigates the performance of an EEG-based Brain-Computer Interface (BCI) to discriminate different difficulty levels in performing a mental arithmetic task. EEG data from 10 subjects were collected while performing mental addition with 3 difficulty levels (easy, medium and hard). EEG features were then extracted using band power and Common Spatial Pattern features and subsequently features were selected using Fisher Ratio to train a Linear Discriminant Classifier. The results from 10-fold cross-validation yielded averaged accuracy of 90% for 2 classes (easy versus hard tasks) and 66% for 3 classes (easy versus medium versus hard tasks). Hence the results showed the feasibility of using EEG-based BCI to measure cognitive workload in performing mental arithmetic.

Modeling EEG-based Motor Imagery with Session to Session Online Adaptation.

Subject-specific calibration plays an important role in electroencephalography (EEG)-based Brain-Computer Interface (BCI) for Motor Imagery (MI) detection. A calibration session is often introduced to build a subject specific model, which then can be deployed into BCI system for MI detection in the following rehabilitation sessions. The model is termed as a fixed calibration model. Progressive adaptive models can also be built by using data not only from calibration session, but also from available rehabilitation sessions. It was reported that the progressive adaptive model yielded significant improved MI detection compared to the fixed model in a retrospective clinical study. In this work, we deploy the progressive adaptation model in a BCI-based stroke rehabilitation system and bring it online. We dub this system nBETTER (Neurostyle Brain Exercise Therapy Towards Enhanced Recovery). A clinical trial using the nBETTER system was conducted to evaluate the performance of 11 stroke patients who underwent a calibration session followed by 18 rehabilitation sessions over 6 weeks. We conduct retrospective analysis to compare the performance of various modeling strategies: the fixed calibration model, the online progressive adaptation model and a light-weight adaptation model, where the second one is generated online by nBETTER system and the other two models are obtained retrospectively. The mean accuracy of the three models across 11 subjects are 68.17%, 74.04% and 74.53% respectively. Statistical test conducted on the three groups using ANOVA yields a p-value of 9.83-e06. The test result shows that the two adaptation models both have significant different mean from fixed mode. Hence our study confirmed the effectiveness of using the progressive adaptive model for EEGbased BCI to detect MI in an online setting.

Spectrum and Phase Adaptive CCA for SSVEP-based Brain Computer Interface.

Among various brain activity patterns, Steady State Visual Evoked Potential (SSVEP) based Brain Computer Inter-face (BCI) requires the least training time while carrying the fastest information transfer rate, making it highly suitable for deploying efficient self-paced BCI systems. In this study, we propose a Spectrum and Phase Adaptive CCA (SPACCA) for subject-and device-specific SSVEP-based BCI. Cross subject heterogeneity of spectrum distribution is taken into consideration to improve the prediction accuracy. We design a library of phase shifting reference signals to accommodate subjective and device-related response time lag. With the flexible reference signal generating approach, the system can be optimized for any specific flickering source, include LED, computer screen and mobile devices. We evaluated the performance of SPACCA using three sets of data that use LED, computer screen and mobile device (tablet) as stimuli sources respectively. The first two data sets are publicly available whereas the third data set is collected in our BCI lab. Across different data sets, SPACCA consistently performs better than the baseline, i.e. standard CCA approach. Statistical test to compare the overall results across three data sets yield a p-value of 1.66e-6, implying the improvement is significant.

Effectiveness of a Personalized Brain-Computer Interface System for Cognitive Training in Healthy Elderly: A Randomized Controlled Trial.

BACKGROUND: Cognitive training has been demonstrated to improve cognitive performance in older adults. To date, no study has explored personalized training that targets the brain activity of each individual. OBJECTIVE: This is the first large-scale trial that examines the usefulness of personalized neurofeedback cognitive training. METHODS: We conducted a randomized-controlled trial with participants who were 60-80 years old, with Clinical Dementia Rating (CDR) score of 0-0.5, Mini-Mental State Examination (MMSE) score of 24 and above, and with no neuropsychiatric diagnosis. Participants were randomly assigned to the Intervention or Waitlist-Control group. The training system, BRAINMEM, has attention, working memory, and delayed recall game components. The intervention schedule comprised 24 sessions over eight weeks and three monthly booster sessions. The primary outcome was the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS) total score after the 24-session training. RESULTS: There were no significant between-subjects differences in overall cognitive performance post-intervention. However, a sex moderation effect (p = 0.014) was present. Men in the intervention group performed better than those in the waitlist group (mean difference, +4.03 (95% CI 0.1 to 8.0), p = 0.046. Among females, however, both waitlist-control and intervention participants improved from baseline, although the between-group difference in improvement did not reach significance. BRAINMEM also received positive appraisal and intervention adherence from the participants. CONCLUSION: A personalized neurofeedback intervention is potentially feasible for use in cognitive training for older males. The sex moderation effect warrants further investigation and highlights the importance of taking sex into account during cognitive training.

Facilitating motor imagery-based brain-computer interface for stroke patients using passive movement.

Motor imagery-based brain-computer interface (MI-BCI) has been proposed as a rehabilitation tool to facilitate motor recovery in stroke. However, the calibration of a BCI system is a time-consuming and fatiguing process for stroke patients, which leaves reduced time for actual therapeutic interaction. Studies have shown that passive movement (PM) (i.e., the execution of a movement by an external agency without any voluntary motions) and motor imagery (MI) (i.e., the mental rehearsal of a movement without any activation of the muscles) induce similar EEG patterns over the motor cortex. Since performing PM is less fatiguing for the patients, this paper investigates the effectiveness of calibrating MI-BCIs from PM for stroke subjects in terms of classification accuracy. For this purpose, a new adaptive algorithm called filter bank data space adaptation (FB-DSA) is proposed. The FB-DSA algorithm linearly transforms the band-pass-filtered MI data such that the distribution difference between the MI and PM data is minimized. The effectiveness of the proposed algorithm is evaluated by an offline study on data collected from 16 healthy subjects and 6 stroke patients. The results show that the proposed FB-DSA algorithm significantly improved the classification accuracies of the PM and MI calibrated models (p < 0.05). According to the obtained classification accuracies, the PM calibrated models that were adapted using the proposed FB-DSA algorithm outperformed the MI calibrated models by an average of 2.3 and 4.5 % for the healthy and stroke subjects respectively. In addition, our results suggest that the disparity between MI and PM could be stronger in the stroke patients compared to the healthy subjects, and there would be thus an increased need to use the proposed FB-DSA algorithm in BCI-based stroke rehabilitation calibrated from PM.

A Randomized Controlled Trial of EEG-Based Motor Imagery Brain-Computer Interface Robotic Rehabilitation for Stroke.

Electroencephalography (EEG)-based motor imagery (MI) brain-computer interface (BCI) technology has the potential to restore motor function by inducing activity-dependent brain plasticity. The purpose of this study was to investigate the efficacy of an EEG-based MI BCI system coupled with MIT-Manus shoulder-elbow robotic feedback (BCI-Manus) for subjects with chronic stroke with upper-limb hemiparesis. In this single-blind, randomized trial, 26 hemiplegic subjects (Fugl-Meyer Assessment of Motor Recovery After Stroke [FMMA] score, 4-40; 16 men; mean age, 51.4 years; mean stroke duration, 297.4 days), prescreened with the ability to use the MI BCI, were randomly allocated to BCI-Manus or Manus therapy, lasting 18 hours over 4 weeks. Efficacy was measured using upper-extremity FMMA scores at weeks 0, 2, 4 and 12. ElEG data from subjects allocated to BCI-Manus were quantified using the revised brain symmetry index (rBSI) and analyzed for correlation with the improvements in FMMA score. Eleven and 15 subjects underwent BCI-Manus and Manus therapy, respectively. One subject in the Manus group dropped out. Mean total FMMA scores at weeks 0, 2, 4, and 12 weeks improved for both groups: 26.3±10.3, 27.4±12.0, 30.8±13.8, and 31.5±13.5 for BCI-Manus and 26.6±18.9, 29.9±20.6, 32.9±21.4, and 33.9±20.2 for Manus, with no intergroup differences (P=.51). More subjects attained further gains in FMMA scores at week 12 from BCI-Manus (7 of 11 [63.6%]) than Manus (5 of 14 [35.7%]). A negative correlation was found between the rBSI and FMMA score improvement (P=.044). BCI-Manus therapy was well tolerated and not associated with adverse events. In conclusion, BCI-Manus therapy is effective and safe for arm rehabilitation after severe poststroke hemiparesis. Motor gains were comparable to those attained with intensive robotic therapy (1,040 repetitions/session) despite reduced arm exercise repetitions using EEG-based MI-triggered robotic feedback (136 repetitions/session). The correlation of rBSI with motor improvements suggests that the rBSI can be used as a prognostic measure for BCI-based stroke rehabilitation.

Discriminative channel addition and reduction for filter bank common spatial pattern in motor imagery BCI.

An electroencephalography (EEG)-based Motor Imagery Brain-Computer Interface (MI-BCI) requires a long setup time if a large number of channels is used, and EEG from noisy or irrelevant channels may adversely affect the classification performance. To address this issue, this paper proposed 2 approaches to systematically select discriminative channels for EEG-based MI-BCI. The proposed Discriminative Channel Addition (DCA) approach and the Discriminative Channel Reduction (DCR) approach selects subject-specific discriminative channels by iteratively adding or removing channels based on the cross-validation classification accuracies obtained using the Filter Bank Common Spatial Pattern algorithm. The performances of the proposed approaches were evaluated on the BCI Competition IV Dataset 2a. The results on 2-class and 4-class MI data showed that DCA, which iteratively adds channels, selected 13~14 channels that consistently yielded better cross-validation accuracies on the training data and session-to-session transfer accuracies on the evaluation data compared to the use of a full 22-channel setup. Hence, this results in a reduced channel setup that could improve the classification accuracy of the MI-BCI after removing less discriminative channels.

Selection of effective EEG channels in brain computer interfaces based on inconsistencies of classifiers.

This paper proposed a novel method to select the effective Electroencephalography (EEG) channels for the motor imagery tasks based on the inconsistencies from multiple classifiers. The inconsistency criterion for channel selection was designed based on the fluctuation of the classification accuracies among different classifiers when the noisy channels were included. These noisy channels were then identified and removed till a required number of channels was selected or a predefined classification accuracy with reference to baseline was obtained. Experiments conducted on a data set of 13 healthy subjects performing hand grasping and idle revealed that the EEG channels from the motor area were most frequently selected. Furthermore, the mean increases of 4.07%, 3.10% and 1.77% of the averaged accuracies in comparison with the four existing channel selection methods were achieved for the non-feedback, feedback and calibration sessions, respectively, by selecting as low as seven channels. These results further validated the effectiveness of our proposed method.

Brain-computer interface-based robotic end effector system for wrist and hand rehabilitation: results of a three-armed randomized controlled trial for chronic stroke.

The objective of this study was to investigate the efficacy of an Electroencephalography (EEG)-based Motor Imagery (MI) Brain-Computer Interface (BCI) coupled with a Haptic Knob (HK) robot for arm rehabilitation in stroke patients. In this three-arm, single-blind, randomized controlled trial; 21 chronic hemiplegic stroke patients (Fugl-Meyer Motor Assessment (FMMA) score 10-50), recruited after pre-screening for MI BCI ability, were randomly allocated to BCI-HK, HK or Standard Arm Therapy (SAT) groups. All groups received 18 sessions of intervention over 6 weeks, 3 sessions per week, 90 min per session. The BCI-HK group received 1 h of BCI coupled with HK intervention, and the HK group received 1 h of HK intervention per session. Both BCI-HK and HK groups received 120 trials of robot-assisted hand grasping and knob manipulation followed by 30 min of therapist-assisted arm mobilization. The SAT group received 1.5 h of therapist-assisted arm mobilization and forearm pronation-supination movements incorporating wrist control and grasp-release functions. In all, 14 males, 7 females, mean age 54.2 years, mean stroke duration 385.1 days, with baseline FMMA score 27.0 were recruited. The primary outcome measure was upper extremity FMMA scores measured mid-intervention at week 3, end-intervention at week 6, and follow-up at weeks 12 and 24. Seven, 8 and 7 subjects underwent BCI-HK, HK and SAT interventions respectively. FMMA score improved in all groups, but no intergroup differences were found at any time points. Significantly larger motor gains were observed in the BCI-HK group compared to the SAT group at weeks 3, 12, and 24, but motor gains in the HK group did not differ from the SAT group at any time point. In conclusion, BCI-HK is effective, safe, and may have the potential for enhancing motor recovery in chronic stroke when combined with therapist-assisted arm mobilization.

Motor imagery BCI for upper limb stroke rehabilitation: An evaluation of the EEG recordings using coherence analysis.

Brain-computer interface (BCI) technology has the potential as a post-stroke rehabilitation tool, and the efficacy of the technology is most often demonstrated through output peripherals such as robots, orthosis and computers. In this study, the EEG signals recorded during the course of upper limb stroke rehabilitaion using motor imagery BCI were analyzed to better understand the effect of BCI therapy for post-stroke rehabilitation. The stroke patients recruited underwent 10 sessions of 1-hour BCI with robotic feedback for 2 weeks, 5 times a week. The analysis was performed by computing the coherences of the EEG in the lesion and contralesion side of the hemisphere from each session, and the coherence index of the lesion hemisphere (0 ≤ CI ≤ 1) was computed. The coherence index represents the rate of activation of the lesion hemisphere, and the correlation with the Fugl-Meyer assessment (FMA) before and after the BCI therapy was investigated. Significant improvement in the FMA scores was reported for five of the six patients (p = 0.01). The analysis showed that the number of sessions with CI ≥ 0.5 correlated with the change in the FMA scores. This suggests that post-stroke motor recovery best results from the activation in the lesion hemisphere, which is in agreement with previous studies performed using multimodal imaging technologies.

Feature consistency-based model adaptation in session-to-session classification: a study using motor imagery of swallow EEG signals.

The performance degradation for session to session classification in brain computer interface is a critical problem. This paper proposes a novel method for model adaptation based on motor imagery of swallow EEG signal for dysphagia rehabilitation. A small amount of calibration testing data is utilized to select the model catering for test data. The features of the training and calibration testing data are firstly clustered and each cluster is labeled by the dominant label of the training data. The cluster with the minimum impurity is selected and the number of features consistent with the cluster label are calculated for both training and calibration testing data. Finally, the training model with the maximum number of consistent features is selected. Experiments conducted on motor imagery of swallow EEG data achieved an average accuracy of 74.29% and 72.64% with model adaptation for Laplacian derivates of power features and wavelet features, respectively. Further, an average accuracy increase of 2.9% is achieved with model adaptation using wavelet features, in comparison with that achieved without model adaptation, which is significant at 5% significance level as demonstrated in the statistical test.