BCI Competition 2003—Data Set IIb: Support Vector Machines for the P300 Speller Paradigm

Operation of a P300-based brain-computer interface by individuals with cervical spinal cord injury

OBJECTIVE

This study evaluates the efficacy of a P300-based brain-computer interface (BCI) with green/blue flicker matrices for individuals with cervical spinal cord injury (SCI).

METHODS

Ten individuals with cervical SCI (age 26-53, all male) and 10 age- and sex-matched able-bodied controls (age 27-52, all male) with no prior BCI experience were asked to input hiragana (Japanese alphabet) characters using the P300 BCI with two distinct types of visual stimuli, white/gray and green/blue, in an 8×10 flicker matrix. Both online and offline performance were evaluated.

RESULTS

The mean online accuracy of the SCI subjects was 88.0% for the white/gray and 90.7% for the green/blue flicker matrices. The accuracy of the control subjects was 77.3% and 86.0% for the white/gray and green/blue, respectively. There was a significant difference in online accuracy between the two types of flicker matrix. SCI subjects performed with greater accuracy than controls, but the main effect was not significant.

CONCLUSIONS

Individuals with cervical SCI successfully controlled the P300 BCI, and the green/blue flicker matrices were associated with significantly higher accuracy than the white/gray matrices.

SIGNIFICANCE

The P300 BCI with the green/blue flicker matrices is effective for use not only in able-bodied subjects, but also in individuals with cervical SCI.

Documenting, modelling and exploiting P300 amplitude changes due to variable target delays in Donchin's speller

The P300 is an endogenous event-related potential (ERP) that is naturally elicited by rare and significant external stimuli. P300s are used increasingly frequently in brain-computer interfaces (BCIs) because the users of ERP-based BCIs need no special training. However, P300 waves are hard to detect and, therefore, multiple target stimulus presentations are needed before an interface can make a reliable decision. While significant improvements have been made in the detection of P300s, no particular attention has been paid to the variability in shape and timing of P300 waves in BCIs. In this paper we start filling this gap by documenting, modelling and exploiting a modulation in the amplitude of P300s related to the number of non-targets preceding a target in a Donchin speller. The basic idea in our approach is to use an appropriately weighted average of the responses produced by a classifier during multiple stimulus presentations, instead of the traditional plain average. This makes it possible to weigh more heavily events that are likely to be more informative, thereby increasing the accuracy of classification. The optimal weights are determined through a mathematical model that precisely estimates the accuracy of our speller as well as the expected performance improvement w.r.t. the traditional approach. Tests with two independent datasets show that our approach provides a marked statistically significant improvement in accuracy over the top-performing algorithm presented in the literature to date. The method and the theoretical models we propose are general and can easily be used in other P300-based BCIs with minimal changes.

An EEG-based BCI system for 2-D cursor control by combining Mu/Beta rhythm and P300 potential

Two-dimensional cursor control is an important and challenging issue in EEG-based brain-computer interfaces (BCIs). To address this issue, here we propose a new approach by combining two brain signals including Mu/Beta rhythm during motor imagery and P300 potential. In particular, a motor imagery detection mechanism and a P300 potential detection mechanism are devised and integrated such that the user is able to use the two signals to control, respectively, simultaneously, and independently, the horizontal and the vertical movements of the cursor in a specially designed graphic user interface. A real-time BCI system based on this approach is implemented and evaluated through an online experiment involving six subjects performing 2-D control tasks. The results attest to the efficacy of obtaining two independent control signals by the proposed approach. Furthermore, the results show that the system has merit compared with prior systems: it allows cursor movement between arbitrary positions.

A semi-supervised support vector machine approach for parameter setting in motor imagery-based brain computer interfaces

Parameter setting plays an important role for improving the performance of a brain computer interface (BCI). Currently, parameters (e.g. channels and frequency band) are often manually selected. It is time-consuming and not easy to obtain an optimal combination of parameters for a BCI. In this paper, motor imagery-based BCIs are considered, in which channels and frequency band are key parameters. First, a semi-supervised support vector machine algorithm is proposed for automatically selecting a set of channels with given frequency band. Next, this algorithm is extended for joint channel-frequency selection. In this approach, both training data with labels and test data without labels are used for training a classifier. Hence it can be used in small training data case. Finally, our algorithms are applied to a BCI competition data set. Our data analysis results show that these algorithms are effective for selection of frequency band and channels when the training data set is small.

(C)overt attention and visual speller design in an ERP-based brain-computer interface

BACKGROUND

In a visual oddball paradigm, attention to an event usually modulates the event-related potential (ERP). An ERP-based brain-computer interface (BCI) exploits this neural mechanism for communication. Hitherto, it was unclear to what extent the accuracy of such a BCI requires eye movements (overt attention) or whether it is also feasible for targets in the visual periphery (covert attention). Also unclear was how the visual design of the BCI can be improved to meet peculiarities of peripheral vision such as low spatial acuity and crowding.

METHOD

Healthy participants (N = 13) performed a copy-spelling task wherein they had to count target intensifications. EEG and eye movements were recorded concurrently. First, (c)overt attention was investigated by way of a target fixation condition and a central fixation condition. In the latter, participants had to fixate a dot in the center of the screen and allocate their attention to a target in the visual periphery. Second, the effect of visual speller layout was investigated by comparing the symbol Matrix to an ERP-based Hex-o-Spell, a two-levels speller consisting of six discs arranged on an invisible hexagon.

RESULTS

We assessed counting errors, ERP amplitudes, and offline classification performance. There is an advantage (i.e., less errors, larger ERP amplitude modulation, better classification) of overt attention over covert attention, and there is also an advantage of the Hex-o-Spell over the Matrix. Using overt attention, P1, N1, P2, N2, and P3 components are enhanced by attention. Using covert attention, only N2 and P3 are enhanced for both spellers, and N1 and P2 are modulated when using the Hex-o-Spell but not when using the Matrix. Consequently, classifiers rely mainly on early evoked potentials in overt attention and on later cognitive components in covert attention.

CONCLUSIONS

Both overt and covert attention can be used to drive an ERP-based BCI, but performance is markedly lower for covert attention. The Hex-o-Spell outperforms the Matrix, especially when eye movements are not permitted, illustrating that performance can be increased if one accounts for peculiarities of peripheral vision.

A brain controlled wheelchair to navigate in familiar environments

While brain-computer interfaces (BCIs) can provide communication to people who are locked-in, they suffer from a very low information transfer rate. Further, using a BCI requires a concentration effort and using it continuously can be tiring. The brain controlled wheelchair (BCW) described in this paper aims at providing mobility to BCI users despite these limitations, in a safe and efficient way. Using a slow but reliable P300 based BCI, the user selects a destination amongst a list of predefined locations. While the wheelchair moves on virtual guiding paths ensuring smooth, safe, and predictable trajectories, the user can stop the wheelchair by using a faster BCI. Experiments with nondisabled subjects demonstrated the efficiency of this strategy. Brain control was not affected when the wheelchair was in motion, and the BCW enabled the users to move to various locations in less time and with significantly less control effort than other control strategies proposed in the literature.

A dictionary-driven P300 speller with a modified interface

P300 spellers are mainly composed of an interface, by which alphanumerical characters are presented to users, and a classification system, which identifies the target character by using acquired EEG data. In this study, we proposed modifications both to the interface and to the classification system, in order to reduce the number of required stimulus repetitions and consequently boost the information transfer rate. We initially incorporated a custom-built dictionary into the classification system, and conducted a study on 14 healthy subjects who copy-spelled 15 four letter words. Incorporating the dictionary, the mean accuracy at five trials increased from 72.86% to 95.71%. To further increase the system performance, we first validated the hypothesis that for a conventional P300 system, most target-error pairs lie on the same row or column. Then based on the validated hypothesis, we adjusted letter positions on the well-known from A to Z interface. The same subjects spelled the same 15 words using the modified interface as well, and the mean information transfer rate at two trials reached 55.32 bits/min.

Single trial independent component analysis for P300 BCI system

A Brain Computer Interface (BCI) is a device that allows the user to communicate with the world without utilizing voluntary muscle activity (i.e., using only the electrical activity of the brain). It makes use of the well-studied observation that the brain reacts differently to different stimuli, as a function of the level of attention allotted to the stimulus stream and the specific processing triggered by the stimulus. In this article we present a single trial independent component analysis (ICA) method that is working with a BCI system proposed by Farwell and Donchin. It can dramatically reduce the signal processing time and improve the data communicating rate. This ICA method achieved 76.67% accuracy on single trial P300 response identification.

A novel P300-based brain-computer interface stimulus presentation paradigm: moving beyond rows and columns

OBJECTIVE

An electroencephalographic brain-computer interface (BCI) can provide a non-muscular means of communication for people with amyotrophic lateral sclerosis (ALS) or other neuromuscular disorders. We present a novel P300-based BCI stimulus presentation - the checkerboard paradigm (CBP). CBP performance is compared to that of the standard row/column paradigm (RCP) introduced by Farwell and Donchin (1988).

METHODS

Using an 8x9 matrix of alphanumeric characters and keyboard commands, 18 participants used the CBP and RCP in counter-balanced fashion. With approximately 9-12 min of calibration data, we used a stepwise linear discriminant analysis for online classification of subsequent data.

RESULTS

Mean online accuracy was significantly higher for the CBP, 92%, than for the RCP, 77%. Correcting for extra selections due to errors, mean bit rate was also significantly higher for the CBP, 23 bits/min, than for the RCP, 17 bits/min. Moreover, the two paradigms produced significantly different waveforms. Initial tests with three advanced ALS participants produced similar results. Furthermore, these individuals preferred the CBP to the RCP.

CONCLUSIONS

These results suggest that the CBP is markedly superior to the RCP in performance and user acceptability.

SIGNIFICANCE

The CBP has the potential to provide a substantially more effective BCI than the RCP. This is especially important for people with severe neuromuscular disabilities.

Adaptive active auditory brain computer interface

An active paradigm was employed to produce reliable and prominent target response in an auditory brain computer interface (BCI), in which subject's voluntary recognition of the property of a target human voice enhances the discriminability between target and non-target EEG response. Furthermore, to adaptively decide the optimal number of trials being averaged for SVM classification, a statistical approach was proposed to convert each sample's margin in support vector space into probabilities of each voice choice being the target. In a testing of 8 subjects' EEG data from the active auditory BCI experiment, the proposed adaptive approach needs only about 4-6 trials to reach the equivalent accuracy of 15-trial averaging. The improved information transfer rate suggests the advantage of adaptive strategy in an active auditory BCI.

Visual stimuli for the P300 brain-computer interface: a comparison of white/gray and green/blue flicker matrices

OBJECTIVE

The white/gray flicker matrix has been used as a visual stimulus for the so-called P300 brain-computer interface (BCI), but the white/gray flash stimuli might induce discomfort. In this study, we investigated the effectiveness of green/blue flicker matrices as visual stimuli.

METHODS

Ten able-bodied, non-trained subjects performed Alphabet Spelling (Japanese Alphabet: Hiragana) using an 8 x 10 matrix with three types of intensification/rest flicker combinations (L, luminance; C, chromatic; LC, luminance and chromatic); both online and offline performances were evaluated.

RESULTS

The accuracy rate under the online LC condition was 80.6%. Offline analysis showed that the LC condition was associated with significantly higher accuracy than was the L or C condition (Tukey-Kramer, p < 0.05). No significant difference was observed between L and C conditions.

CONCLUSIONS

The LC condition, which used the green/blue flicker matrix was associated with better performances in the P300 BCI.

SIGNIFICANCE

The green/blue chromatic flicker matrix can be an efficient tool for practical BCI application.

A simple classification tool for single-trial analysis of ERP components

Event-related potentials (ERPs) were recorded by measuring a dense sensor EEG from eight healthy volunteers in a visual oddball experiment. Single trials were analyzed with an extremely simple high-dimensional version of discriminant analysis. The question was how many of the target trials contribute to the average P3, and to test whether other components in the ERP are sensitive to discriminate between target and non-target trials. One common classification rule for all participants expressing the P3 component correctly classified 88% of the ERPs of all subjects in response to a target or non-target trial. For four of the eight participants, there were strong differences in an early ERP component over the occipital recording sites. Their individual classification rules, obtained from the training data in the time interval up to 200 ms, correctly classified 85% of the trials of the test data.

Unsupervised brain computer interface based on intersubject information and online adaptation

Conventional brain computer interfaces rely on a guided calibration procedure to address the problem of considerable variations in electroencephalography (EEG) across human subjects. This calibration, however, implies inconvenience to the end users. In this paper, we propose an online-adaptive-learning method to address this problem for P300-based brain computer interfaces. By automatically capturing subject-specific EEG characteristics during online operation, this method allows a new user to start operating a P300-based brain-computer interface without guided (supervised) calibration. The basic principle is to first learn a generic model termed subject-independent model offline from EEG of a pool of subjects to capture common P300 characteristics. For a new user, a new model termed subject-specific model is then adapted online based on EEG recorded from the new subject and the corresponding labels predicted by either the subject-independent model or the adapted subject-specific model, depending on a confidence score. To verify the proposed method, a study involving 10 healthy subjects is carried out and positive results are obtained. For instance, after 2-4 min online adaptation (spelling of 10-20 characters), the accuracy of the adapted model converges to that of a fully trained supervised subject-specific model.

Supervised adaptive downsampling for P300-based brain computer interface

To realize Brain Computer Interface, a recording electroencephalogram (EEG) and determining whether or not P300 is evoked by the presented stimulus have become increasingly important. Using the machine learning method for this classification is effective, but constructing feature vectors with all data points might result in very high-dimensional data. Because such redundant features are undesirable from the viewpoint of computation and classification performance, EEG has been downsampled in several studies. In the present study, we propose a new downsampling method aiming at the improvement of P300 classification accuracy. In particular, each single trial EEG is segmented at non-uniform intervals and then averaged in each segment. The segmentation is decided in such a way that the degree of separating two classes from training data is increased by applying a time series segmentation algorithm. Our experiment using the BCI Competition III P300 Speller paradigm data set demonstrated that our method resulted in higher accuracy than traditional downsampling methods.

Automated seizure onset detection for accurate onset time determination in intracranial EEG

OBJECTIVE

A novel algorithm for automated seizure onset detection is presented. The method allows for precise identification of electrographic seizure onset times within large databases of electrographic data.

METHODS

The patient-specific algorithm extracts salient spectral and temporal features in five frequency bands within a sliding window of an electrographic recording. Feature windows are classified as containing or not containing a seizure onset via support vector machines. A clustering and regression analysis is utilized to accurately localize seizure onsets in time. User-adjustable parameters allow for tuning of detection sensitivity, false positive rate, and latency. The method was tested on intracranial electrographic data recorded from six patients with a total of 1792 recorded seizure onsets from 8246 total electrographic recordings.

RESULTS

Testing of algorithm performance via cross-validation resulted in sensitivities between 80% and 98%, false positive rates from 0.002 to 0.046 per minute (0.12-2.8 per hour), and median detection time within 100ms of the electrographic onset for all patients. In five of the six patients, more than 90% of all detected onsets were less than 3s from the electrographic onset.

CONCLUSIONS

The detection system was able to detect seizure onset times in a temporally unbiased fashion with low latency while maintaining reasonable sensitivities and false positive rates. The regression algorithm for temporal localization of onsets confers a considerable benefit in terms of detection latency.

SIGNIFICANCE

With the use of our algorithm, large databases of electrographic data can be rapidly processed and seizure onset times accurately marked, facilitating research and analyses of peri-onset events that require precise seizure onset alignment.

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).

Environmental Noise and P300-Based Brain-Computer Interface (BCI)

The primary purpose of this study was to systematically evaluate the overall effect of simulated environmental noise on the P300 Speller in order to obtain usability and performance data. P300 Speller is a P300-based brain-computer interface (BCI) that allows people with motor disabilities to type characters just by thinking. Two environmental noise simulations (quiet [20–40 dB], and noisy [70–120 dB]) were examined to simulate the effects of real-world noise. Results of the study indicated that although there were differences in accuracy rate and information transfer rate (ITR) between the noise and quiet conditions, the environmental noise factor was not statistically significant. On the other hand, the P300 amplitude was significantly higher in the noisy condition than in the quiet condition. Unlike the common knowledge that BCI applications are generally preferred to be performed in quiet conditions, higher noise levels seem to increase user concentration. The outcomes of this research should have a broad impact on future user interface design of BCI applications.

Fusion of classic P300 detection methods' inferences in a framework of fuzzy labels

OBJECTIVE

Designing a reliable and accurate brain-computer interface (BCI) is one of the most challenging fields in biomedical signal processing. To achieve this goal, different methods have been adopted in different blocks of a typical BCI system (i.e., in preprocessing, feature extraction, feature classification and feature selection blocks). Since BCI's speed plays a crucial role in its success in real-life applications, using mathematically simple techniques with accurate and reliable performance can improve this aspect of BCI systems' design.

METHODS AND MATERIALS

In this paper, a new method is introduced, which combines information from different classic time series similarity measures, using a simple fuzzy fusion framework. This method is accurate and reliable in P300 (a positive event-related component occurring 300 ms after stimulus onset) detection. This framework is used to combine two computationally simple signal detection methods: "peak picking" and "template matching". Fusion takes place in the last step (decision-making step) by means of a fuzzy rule-base.

RESULTS AND CONCLUSIONS

Compared to similar works on electroencephalogram-based (EEG-based) BCI datasets, in spite of being computationally simple, this new technique's performance is comparable to very complicated methods, like support vector machines. This research indicates that, using both spatial and temporal information content of EEG trials (from all electrodes or a subset of them), even under a non-complicated mathematical framework can yield an accurate and powerful classification.

A P300-based brain-computer interface for people with amyotrophic lateral sclerosis

OBJECTIVE

The current study evaluates the efficacy of a P300-based brain-computer interface (BCI) communication device for individuals with advanced ALS.

METHODS

Participants attended to one cell of a N x N matrix while the N rows and N columns flashed randomly. Each cell of the matrix contained one character. Every flash of an attended character served as a rare event in an oddball sequence and elicited a P300 response. Classification coefficients derived using a stepwise linear discriminant function were applied to the data after each set of flashes. The character receiving the highest discriminant score was presented as feedback.

RESULTS

In Phase I, six participants used a 6 x 6 matrix on 12 separate days with a mean rate of 1.2 selections/min and mean online and offline accuracies of 62% and 82%, respectively. In Phase II, four participants used either a 6 x 6 or a 7 x 7 matrix to produce novel and spontaneous statements with a mean online rate of 2.1 selections/min and online accuracy of 79%. The amplitude and latency of the P300 remained stable over 40 weeks.

CONCLUSIONS

Participants could communicate with the P300-based BCI and performance was stable over many months.

SIGNIFICANCE

BCIs could provide an alternative communication and control technology in the daily lives of people severely disabled by ALS.

Towards an independent brain-computer interface using steady state visual evoked potentials

OBJECTIVE

Brain-computer interface (BCI) systems using steady state visual evoked potentials (SSVEPs) have allowed healthy subjects to communicate. However, these systems may not work in severely disabled users because they may depend on gaze shifting. This study evaluates the hypothesis that overlapping stimuli can evoke changes in SSVEP activity sufficient to control a BCI. This would provide evidence that SSVEP BCIs could be used without shifting gaze.

METHODS

Subjects viewed a display containing two images that each oscillated at a different frequency. Different conditions used overlapping or non-overlapping images to explore dependence on gaze function. Subjects were asked to direct attention to one or the other of these images during each of 12 one-minute runs.

RESULTS

Half of the subjects produced differences in SSVEP activity elicited by overlapping stimuli that could support BCI control. In all remaining users, differences did exist at corresponding frequencies but were not strong enough to allow effective control.

CONCLUSIONS

The data demonstrate that SSVEP differences sufficient for BCI control may be elicited by selective attention to one of two overlapping stimuli. Thus, some SSVEP-based BCI approaches may not depend on gaze control. The nature and extent of any BCI's dependence on muscle activity is a function of many factors, including the display, task, environment, and user.

SIGNIFICANCE

SSVEP BCIs might function in severely disabled users unable to reliably control gaze. Further research with these users is necessary to explore the optimal parameters of such a system and validate online performance in a home environment.

Analysis of p300 classifiers in brain computer interface speller

In this paper, the performance of five classifiers in P300 speller paradigm are compared. Theses classifiers are Linear Support Vector Machine (LSVM), Gaussian Support Vector Machine (GSVM), Neural Network (NN), Fisher Linear Discriminant (FLD), and Kernel Fisher Discriminant (KFD). In classification of P300 waves, there has been a trend to use SVM classifiers. Although they have shown a good performance, in this paper, it is shown that the FLD classifiers outperform the SVM classifiers. FLD classifier uses only ten channels of the recorded electroencephalogram (EEG) signals. This makes them a very good candidate for real-time applications. In addition, FLD approach does not need any optimization similar to other methods. In addition, in this paper, it is shown that the efficiency of using Principal Component Analysis (PCA) for feature reduction results in decreasing the time for the classification and increasing the accuracy.

Brain-computer interface systems: progress and prospects

Brain-computer interface (BCI) systems support communication through direct measures of neural activity without muscle activity. BCIs may provide the best and sometimes the only communication option for users disabled by the most severe neuromuscular disorders and may eventually become useful to less severely disabled and/or healthy individuals across a wide range of applications. This review discusses the structure and functions of BCI systems, clarifies terminology and addresses practical applications. Progress and opportunities in the field are also identified and explicated.

Toward enhanced P300 speller performance

This study examines the effects of expanding the classical P300 feature space on the classification performance of data collected from a P300 speller paradigm [Farwell LA, Donchin E. Talking off the top of your head: toward a mental prosthesis utilizing event-related brain potentials. Electroenceph Clin Neurophysiol 1988;70:510-23]. Using stepwise linear discriminant analysis (SWLDA) to construct a classifier, the effects of spatial channel selection, channel referencing, data decimation, and maximum number of model features are compared with the intent of establishing a baseline not only for the SWLDA classifier, but for related P300 speller classification methods in general. By supplementing the classical P300 recording locations with posterior locations, online classification performance of P300 speller responses can be significantly improved using SWLDA and the favorable parameters derived from the offline comparative analysis.

An efficient P300-based brain–computer interface for disabled subjects

A brain-computer interface (BCI) is a communication system that translates brain-activity into commands for a computer or other devices. In other words, a BCI allows users to act on their environment by using only brain-activity, without using peripheral nerves and muscles. In this paper, we present a BCI that achieves high classification accuracy and high bitrates for both disabled and able-bodied subjects. The system is based on the P300 evoked potential and is tested with five severely disabled and four able-bodied subjects. For four of the disabled subjects classification accuracies of 100% are obtained. The bitrates obtained for the disabled subjects range between 10 and 25bits/min. The effect of different electrode configurations and machine learning algorithms on classification accuracy is tested. Further factors that are possibly important for obtaining good classification accuracy in P300-based BCI systems for disabled subjects are discussed.

Defining brain–machine interface applications by matching interface performance with device requirements

Interaction with machines is mediated by human-machine interfaces (HMIs). Brain-machine interfaces (BMIs) are a particular class of HMIs and have so far been studied as a communication means for people who have little or no voluntary control of muscle activity. In this context, low-performing interfaces can be considered as prosthetic applications. On the other hand, for able-bodied users, a BMI would only be practical if conceived as an augmenting interface. In this paper, a method is introduced for pointing out effective combinations of interfaces and devices for creating real-world applications. First, devices for domotics, rehabilitation and assistive robotics, and their requirements, in terms of throughput and latency, are described. Second, HMIs are classified and their performance described, still in terms of throughput and latency. Then device requirements are matched with performance of available interfaces. Simple rehabilitation and domotics devices can be easily controlled by means of BMI technology. Prosthetic hands and wheelchairs are suitable applications but do not attain optimal interactivity. Regarding humanoid robotics, the head and the trunk can be controlled by means of BMIs, while other parts require too much throughput. Robotic arms, which have been controlled by means of cortical invasive interfaces in animal studies, could be the next frontier for non-invasive BMIs. Combining smart controllers with BMIs could improve interactivity and boost BMI applications.

Preprocessing and meta-classification for brain-computer interfaces

A brain-computer interface (BCI) is a system which allows direct translation of brain states into actions, bypassing the usual muscular pathways. A BCI system works by extracting user brain signals, applying machine learning algorithms to classify the user's brain state, and performing a computer-controlled action. Our goal is to improve brain state classification. Perhaps the most obvious way to improve classification performance is the selection of an advanced learning algorithm. However, it is now well known in the BCI community that careful selection of preprocessing steps is crucial to the success of any classification scheme. Furthermore, recent work indicates that combining the output of multiple classifiers (meta-classification) leads to improved classification rates relative to single classifiers (Dornhege et al., 2004). In this paper, we develop an automated approach which systematically analyzes the relative contributions of different preprocessing and meta-classification approaches. We apply this procedure to three data sets drawn from BCI Competition 2003 (Blankertz et al., 2004) and BCI Competition III (Blankertz et al., 2006), each of which exhibit very different characteristics. Our final classification results compare favorably with those from past BCI competitions. Additionally, we analyze the relative contributions of individual preprocessing and meta-classification choices and discuss which types of BCI data benefit most from specific algorithms.

Robustness of the Farwell & Donchin BCI protocol to visual stimulus parameter changes

In this paper a number of visual modifications were carried out upon the Farwell & Donchin protocol. The effects of these modifications were studied both in the classification accuracy of the classifiers and the electrophysiological morphology of the P3 potential. The classifiers used were a Support Vector Machine with a gaussian kernel and a Fisher Linear Discriminant. The electrophysiological aspects of the P3 potential studied were the amplitude and latency. The results indicate that although small fluctuations in the classifier accuracy were observed between the differing visual protocols, the relative changes were not statistically significant. This means that in this set of experiments the Farwell & Donchin has proved to be robust to visual stimulus parameter changes. The experiments also demonstrate the difficulties of using Brain Computer Interfaces and the inconsistent results they often provide across subjects. Furthermore, the experiments have introduced some interesting changes to the visual layout of the Farwell & Donchin protocol.

Assessment of preprocessing on classifiers used in the p300 speller paradigm

Artifact removal is an essential part in electroencephalogram (EEG) recording and the raw EEG signals require preprocessing before feature extraction. In this work, we implemented three filtering methods and demonstrated their effects on the performance of different classifiers. Bandpass digital filtering, median filtering and facet method are three preprocessing approaches investigated in this paper. We used data set lib from the BCI competition 2003 for training and testing phase. Our accuracy varied between 80% and 96%. In our work, we demonstrated that the problems of choosing the classifier and preprocessing methods are not independent of each other. Two of our approaches could achieve the 96% accuracy i.e. 31 of 32 characters were predicted correctly. These two approaches have different classifier and different preprocessing method. It means that the performance of each classifier can be enhanced with a specific preprocessing method. In our approach, we used only three electrodes of 64 applied electrodes. Therefore it can noticeably reduce the time and cost of EEG measurement.

Application of a hybrid wavelet feature selection method in the design of a self-paced brain interface system

BACKGROUND

Recently, successful applications of the discrete wavelet transform have been reported in brain interface (BI) systems with one or two EEG channels. For a multi-channel BI system, however, the high dimensionality of the generated wavelet features space poses a challenging problem.

METHODS

In this paper, a feature selection method that effectively reduces the dimensionality of the feature space of a multi-channel, self-paced BI system is proposed. The proposed method uses a two-stage feature selection scheme to select the most suitable movement-related potential features from the feature space. The first stage employs mutual information to filter out the least discriminant features, resulting in a reduced feature space. Then a genetic algorithm is applied to the reduced feature space to further reduce its dimensionality and select the best set of features.

RESULTS

An offline analysis of the EEG signals (18 bipolar EEG channels) of four able-bodied subjects showed that the proposed method acquires low false positive rates at a reasonably high true positive rate. The results also show that features selected from different channels varied considerably from one subject to another.

CONCLUSION

The proposed hybrid method effectively reduces the high dimensionality of the feature space. The variability in features among subjects indicates that a user-customized BI system needs to be developed for individual users.

A survey of signal processing algorithms in brain-computer interfaces based on electrical brain signals

Brain-computer interfaces (BCIs) aim at providing a non-muscular channel for sending commands to the external world using the electroencephalographic activity or other electrophysiological measures of the brain function. An essential factor in the successful operation of BCI systems is the methods used to process the brain signals. In the BCI literature, however, there is no comprehensive review of the signal processing techniques used. This work presents the first such comprehensive survey of all BCI designs using electrical signal recordings published prior to January 2006. Detailed results from this survey are presented and discussed. The following key research questions are addressed: (1) what are the key signal processing components of a BCI, (2) what signal processing algorithms have been used in BCIs and (3) which signal processing techniques have received more attention?

EMG and EOG artifacts in brain computer interface systems: A survey

It is widely accepted in the brain computer interface (BCI) research community that neurological phenomena are the only source of control in any BCI system. Artifacts are undesirable signals that can interfere with neurological phenomena. They may change the characteristics of neurological phenomena or even be mistakenly used as the source of control in BCI systems. Electrooculography (EOG) and electromyography (EMG) artifacts are considered among the most important sources of physiological artifacts in BCI systems. Currently, however, there is no comprehensive review of EMG and EOG artifacts in BCI literature. This paper reviews EOG and EMG artifacts associated with BCI systems and the current methods for dealing with them. More than 250 refereed journal and conference papers are reviewed and categorized based on the type of neurological phenomenon used and the methods employed for handling EOG and EMG artifacts. This study reveals weaknesses in BCI studies related to reporting the methods of handling EMG and EOG artifacts. Most BCI papers do not report whether or not they have considered the presence of EMG and EOG artifacts in the brain signals. Only a small percentage of BCI papers report automated methods for rejection or removal of artifacts in their systems. As the lack of dealing with artifacts may result in the deterioration of the performance of a particular BCI system during practical applications, it is necessary to develop automatic methods to handle artifacts or to design BCI systems whose performance is robust to the presence of artifacts.

Single-trial estimation of imitating-natural-reading evoked potentials in single-channel

Using Imitating-Natural-Reading Induced Potentials as communication carriers, we are constructing a Brain-computer interface based mental speller which enable users to interaction with computers. The potentials were induced in this way: In a trial, strings consisted of target and non-target symbols were moving smoothly from right to left through a little visual window at the center of computer screen. Subject was instructed to stare at the visual window to count the target, and thus potentials were evoked. In practical applications, fewer electroencephalograph recording channels are preferred. We explored the single-trial estimating of event-related potentials recorded in single-channel using support vector machines in three subjects. With carefully feature selections, we obtained satisfying results of correct classification rate, which is 92.1%, 94.1% and 91.5%, respectively. The results demonstrated the advantages of the inducing paradigm used in our experiments.

Generalizing to new subjects in brain-computer interfacing

This paper evaluates an algorithm based on support vector machines to analyze EEG data from the P300 speller brain-computer interface paradigm. We evaluated the performance of this technique on own experimental data from 8 subjects and achieved high transfer rates of up to 97.57 bits/min (mean 47.26 bits/min) within subjects. We then investigated how well the classifier generalizes when it is trained on data from a set of several subjects and then applied on data from a new subject to use this BCI in a pretrained fashion. Transfer rates up to 61.04 bits/min were achieved (mean 17.64 bits/min) for this situation indicating an encouraging generalization performance.

A review of classification algorithms for EEG-based brain–computer interfaces

In this paper we review classification algorithms used to design brain-computer interface (BCI) systems based on electroencephalography (EEG). We briefly present the commonly employed algorithms and describe their critical properties. Based on the literature, we compare them in terms of performance and provide guidelines to choose the suitable classification algorithm(s) for a specific BCI.

A Comprehensive Survey of Brain Interface Technology Designs

In this work we present the first comprehensive survey of Brain Interface (BI) technology designs published prior to January 2006. Detailed results from this survey, which was based on the Brain Interface Design Framework proposed by Mason and Birch, are presented and discussed to address the following research questions: (1) which BI technologies are directly comparable, (2) what technology designs exist, (3) which application areas (users, activities and environments) have been targeted in these designs, (4) which design approaches have received little or no research and are possible opportunities for new technology, and (5) how well are designs reported. The results of this work demonstrate that meta-analysis of high-level BI design attributes is possible and informative. The survey also produced a valuable, historical cross-reference where BI technology designers can identify what types of technology have been proposed and by whom.