Comparing Different Classifiers in Sensory Motor Brain Computer Interfaces.

A problem that impedes the progress in Brain-Computer Interface (BCI) research is the difficulty in reproducing the results of different papers. Comparing different algorithms at present is very difficult. Some improvements have been made by the use of standard datasets to evaluate different algorithms. However, the lack of a comparison framework still exists. In this paper, we construct a new general comparison framework to compare different algorithms on several standard datasets. All these datasets correspond to sensory motor BCIs, and are obtained from 21 subjects during their operation of synchronous BCIs and 8 subjects using self-paced BCIs. Other researchers can use our framework to compare their own algorithms on their own datasets. We have compared the performance of different popular classification algorithms over these 29 subjects and performed statistical tests to validate our results. Our findings suggest that, for a given subject, the choice of the classifier for a BCI system depends on the feature extraction method used in that BCI system. This is in contrary to most of publications in the field that have used Linear Discriminant Analysis (LDA) as the classifier of choice for BCI systems.

Automatic artefact removal in a self-paced hybrid brain- computer interface system.

BACKGROUND: A novel artefact removal algorithm is proposed for a self-paced hybrid brain-computer interface (BCI) system. This hybrid system combines a self-paced BCI with an eye-tracker to operate a virtual keyboard. To select a letter, the user must gaze at the target for at least a specific period of time (dwell time) and then activate the BCI by performing a mental task. Unfortunately, electroencephalogram (EEG) signals are often contaminated with artefacts. Artefacts change the quality of EEG signals and subsequently degrade the BCI's performance. METHODS: To remove artefacts in EEG signals, the proposed algorithm uses the stationary wavelet transform combined with a new adaptive thresholding mechanism. To evaluate the performance of the proposed algorithm and other artefact handling/removal methods, semi-simulated EEG signals (i.e., real EEG signals mixed with simulated artefacts) and real EEG signals obtained from seven participants are used. For real EEG signals, the hybrid BCI system's performance is evaluated in an online-like manner, i.e., using the continuous data from the last session as in a real-time environment. RESULTS: With semi-simulated EEG signals, we show that the proposed algorithm achieves lower signal distortion in both time and frequency domains. With real EEG signals, we demonstrate that for dwell time of 0.0s, the number of false-positives/minute is 2 and the true positive rate (TPR) achieved by the proposed algorithm is 44.7%, which is more than 15.0% higher compared to other state-of-the-art artefact handling methods. As dwell time increases to 1.0s, the TPR increases to 73.1%. CONCLUSIONS: The proposed artefact removal algorithm greatly improves the BCI's performance. It also has the following advantages: a) it does not require additional electrooculogram/electromyogram channels, long data segments or a large number of EEG channels, b) it allows real-time processing, and c) it reduces signal distortion.

Adaptive classification in a self-paced hybrid brain-computer interface system.

As the characteristics of EEG signals change over time, updating the classifier of a brain computer interface, BCI, (over time) would improve the performance of the system. Developing an adaptive classifier for a self-paced BCI however is not easy because the user's intention (and therefore the true labels of the EEG signals) are not known during the operation of the system. For certain applications, it may be possible to predict the labels of some of the EEG segments using some information about the user's state (e.g., the error potentials or gaze information). This study proposes a method that adaptively updates the classifier of a self-paced BCI in a supervised or semi-supervised manner, using those EEG segments whose labels can be predicted. We employ the eye position information obtained from an eye-tracker to predict the EEG labels. This eye-tracker is also used along with a self-paced BCI to form a hybrid BCI system. The results obtained from seven individuals show that the proposed algorithm outperforms the non-adaptive and other unsupervised adaptive classifiers. It achieves a true positive rate of 49.7% and lowers the number of false positives significantly to only 2.2 FPs/minute.

Toward development of a two-state brain-computer interface based on mental tasks.

A recently collected EEG dataset is analyzed and processed in order to evaluate the performance of a previously designed brain-computer interface (BCI) system. The EEG signals are collected from 29 channels distributed over the scalp. Four subjects completed three sessions each by performing four different mental tasks during each session. The BCI is designed in such a way that only one of the mental tasks can activate it. One important advantage of this BCI is its simplicity, since autoregressive modeling and quadratic discriminant analysis are used for feature extraction and classification, respectively. The autoregressive order which yields the best overall performance is obtained during a fivefold nested cross-validation process. The results are promising as the false positive rates are zero while the true positive rates are sufficiently high (67.26% average).

The development of a sensory substitution system for the sexual rehabilitation of men with chronic spinal cord injury.

INTRODUCTION: Sexual health is often severely impacted after spinal cord injury (SCI). Current research has primarily addressed male erection and fertility, when in fact pleasure and orgasm are top priorities for functional recovery. Sensory substitution technology operates by communicating input from a lost sensory pathway to another intact sensory modality. It was hypothesized that through training and neuroplasticity, mapped tongue sensations would be interpreted as sensory perceptions arising from insensate genitalia, and improve the sexual experience. AIM: To report the development of a sensory substitution system for the sexual rehabilitation of men with chronic SCI. METHODS: Subjects performed sexual self-stimulation while using a novel sensory substitution device that mapped the stroking motion of the hand to a congruous flow of electrocutaneous sensations on the tongue. MAIN OUTCOME MEASURES: Three questionnaires, along with structured interviews, were used to rate the perceived sexual sensations following each training session. RESULTS: Subjects completed 20 sessions over approximately 8 weeks of training. Each subject reported an increased level of sexual pleasure soon after training with the device. Each subject also reported specific perceptions of cutaneous-like sensations below their lesion that matched their hand motion. Later sessions, while remaining pleasurable and interesting, were inconsistent, and no subject reported an orgasmic feeling during a session. The subjects were all interested in continuing training with the device at home, if possible, in the future. CONCLUSIONS: This study is the first to show that sensory substitution is a possible therapeutic avenue for sexual rehabilitation in people lacking normal genital sexual sensations. However more research, for instance on frequency and duration of training, is needed in order to induce functional lasting neuroplasticity. In the near term, SCI rehabilitation should more fully address sexuality and the role of neuroplasticity for promoting the maximal potential for sexual pleasure and orgasm.

A self-paced BCI using stationary wavelet packets.

The stationary wavelet packet analysis is exploited for the first time in the design of a self-paced BCI based on mental tasks. The BCI system is custom designed to achieve a zero false positive rate, as false activations highly restricts the applications of BCIs in real life. The EEG signals of four subjects performing five different mental tasks are used as the dataset. The stationary wavelet packets decompose the signal into eight components. The features used are the autoregressive coefficients obtained by applying autoregressive modeling on the resultant wavelet components. Classification is a two-stage process. The first stage is based on quadratic discriminant analysis which is extremely fast. The second stage is a simple majority voting classifier. During model selection, which is performed via 5-folded cross-validation, the combination of decomposed components and the autoregressive model order that yield the best performance are selected. Results show enhancements in the overall performance for three subjects comparing to our previously designed BCI.

Plausibility assessment of a 2-state self-paced mental task-based BCI using the no-control performance analysis.

The feasibility of having a self-paced brain-computer interface (BCI) based on mental tasks is investigated. The EEG signals of four subjects performing five mental tasks each are used in the design of a 2-state self-paced BCI. The output of the BCI should only be activated when the subject performs a specific mental task and should remain inactive otherwise. For each subject and each task, the feature coefficient and the classifier that yield the best performance are selected, using the autoregressive coefficients as the features. The classifier with a zero false positive rate and the highest true positive rate is selected as the best classifier. The classifiers tested include: linear discriminant analysis, quadratic discriminant analysis, Mahalanobis discriminant analysis, support vector machine, and radial basis function neural network. The results show that: (1) some classifiers obtained the desired zero false positive rate; (2) the linear discriminant analysis classifier does not yield acceptable performance; (3) the quadratic discriminant analysis classifier outperforms the Mahalanobis discriminant analysis classifier and performs almost as well as the radial basis function neural network; and (4) the support vector machine classifier has the highest true positive rates but unfortunately has nonzero false positive rates in most cases.

Performance of a self-paced brain computer interface on data contaminated with eye-movement artifacts and on data recorded in a subsequent session.

The performance of a specific self-paced BCI (SBCI) is investigated using two different datasets to determine its suitability for using online: (1) data contaminated with large-amplitude eye movements, and (2) data recorded in a session subsequent to the original sessions used to design the system. No part of the data was rejected in the subsequent session. Therefore, this dataset can be regarded as a "pseudo-online" test set. The SBCI under investigation uses features extracted from three specific neurological phenomena. Each of these neurological phenomena belongs to a different frequency band. Since many prominent artifacts are either of mostly low-frequency (e.g., eye movements) or mostly high-frequency nature (e.g., muscle movements), it is expected that the system shows a fairly robust performance over artifact-contaminated data. Analysis of the data of four participants using epochs contaminated with large-amplitude eye-movement artifacts shows that the system's performance deteriorates only slightly. Furthermore, the system's performance during the session subsequent to the original sessions remained largely the same as in the original sessions for three out of the four participants. This moderate drop in performance can be considered tolerable, since allowing artifact-contaminated data to be used as inputs makes the system available for users at ALL times.

Quantitative evaluation of ocular artifact removal methods based on real and estimated EOG signals.

We propose a novel metric for quantitatively evaluating ocular artifact (OA) removal methods on real electroencephalogram (EEG) data. For real EEG, existing metrics measure the amount of artifact removed. Our metric measures how much a given method is likely to distort the underlying EEG. The new metric was used to evaluate two existing OA removal algorithms that use the electro-oculogram (EOG) as a reference signal. The combination of a previous metric and our new metric showed there is a trade-off between how well an algorithm removes OAs and how likely it is to distort the underlying EEG. These algorithms require a reference EOG signal, yet for certain applications (e.g., a brain computer interface or BCI) it is preferable or necessary to avoid attaching electrodes around the eyes. We thus also used various combinations of up to 55 channels of EEG to estimate the EOG reference. The metric was again used to compare the use of estimated vs. measured EOG. Our initial results showed that using EOG estimated from as few as 4 EEG electrodes increased the likelihood of distorting the EEG from 14% to 19% and from 21% to 23% for the two algorithms. For some applications (e.g., BCI), the slight reduction in performance may be acceptable in order to avoid using EOG electrodes.

Evaluating the performance of a self-paced BCI with a new movement and using a more engaging environment.

In previous studies, we proposed a self-paced brain computer interface (SBCI) system that employed three neurological phenomena to identify intentional control (IC) commands from the no control (NC) states of EEG signals. We showed that this SBCI system achieved a good performance that was better than those of other EEG-based SBCI systems. In this paper, we carry out a new study to show that this system can be generalized. Specifically, we show that it can also achieve good performance when 1) a new type of movement is used (hand extension vs. the finger flexion this system was designed for), and 2) NC data are recorded in an engaging environment. A more reliable artifact monitoring system is also added to the system to rule out not only the effects of eye blinks but also the frontalis muscles when controlling the system. Using the data from five participants it is shown that the system obtains good performance compared to other EEG-based SBCI systems.

Robust Common Spatial Patterns for EEG signal preprocessing.

The Common Spatial Patterns (CSP) algorithm finds spatial filters that are useful in discriminating different classes of electroencephalogram (EEG) signals such as those corresponding to different types of motor activities. This algorithm is however, sensitive to outliers because it involves the estimation of covariance matrices. Classical sample covariance estimates are easily affected even if a single outlier exists. To improve the CSP algorithm's robustness against outliers, this paper first investigates how multivariate outliers affect the performance of the CSP algorithm. We then propose a modified version of the algorithm whereby the classical covariance estimates are replaced by the robust covariance estimates obtained using Minimum Covariance Determinant (MCD) estimator. Median Absolute Deviation (MAD) is also used to robustly estimate the variance of the projected EEG signals. The results show that the proposed algorithm is able to reduce the influence of the outliers. When an average of 2.5% outliers is introduced, the average drop in the accuracy is 9.21% for the CSP algorithm and 0.72% for the proposed algorithm.

Self-paced BCI using multiple SWT-based classifiers.

The presence of false activations inhibits the use of existing self-paced BCIs in real life applications. We present a new design method for a self-paced BCI that yielded 0% false activations using the data of two subjects. This system obtains templates/shapes of the movement related finger flexion patterns. To obtain the templates, the intentional control data are decomposed into 5 levels using the stationary wavelet transform. Then, ensemble averaging is done. These templates are used to train 5 radial basis function neural networks. This is followed by a majority voting classifier.

Effect of eye-blinks on a self-paced brain interface design.

OBJECTIVE: To test the performance of an EEG-based self-paced brain interface when data contaminated with eye-blink artefacts are included in the evaluation. METHODS: Two different designs of a self-paced brain interface (the low frequency-asynchronous switch design, LF-ASD) are evaluated and compared using offline data from eight subjects. The true positive rates of the two designs are compared for three cases: (a) data containing eye-blink artefacts are excluded from the input; (b) all data, including eye-blinks, are included as input but the output decisions are inactivated during eye-blink artefacts; (c) all the data, including eye-blinks, are included as input and the output decisions are reported in all times including during eye-blink artefacts. RESULTS: The true positive rates of one design of the LF-ASD (LF-ASD-V5) for case (c) and of another design (LF-ASD-V4) for case (b) are 40.5% and 42.4%, respectively, for false positive rates of 1%. CONCLUSIONS: The true positive rates of LF-ASD-V5 when eye-blinks are included in the analysis deteriorate slightly compared to when the output during eye-blink artefacts is inactivated in LF-ASD-V4. SIGNIFICANCE: LF-ASD-V5 allows the device to be functional at all times and can handle artefacts better than LF-ASD-V4. If a slight decrease in true positive rates is acceptable, no further devices are needed to record the electro-oculogram (EOG) for detecting eye-blinks.

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?

A comparative study on generating training-data for self-paced brain interfaces.

Direct brain interface (BI) systems provide an alternative communication and control solution for individuals with severe motor disabilities, bypassing impaired interface pathways. Most BI systems are aimed to be operated by individuals with severe disabilities. With these individuals, there is no observable indicator of their intent to control or communicate with the BI system. In contrast, able-bodied subjects can perform the desired physical movements such as finger flexion and one can observe the movement as the indicator of intent. Since no external knowledge of intention is available for individuals with severe motor disabilities, generating the data for system training is problematic. This paper introduces three methods for generating training-data for self-paced BI systems and compares their performances with four alternative methods of training-data generation. Results of the offline analysis on the electroencephalogram data of eight subjects during self-paced BI experiments show that two of the proposed methods increase true positive rates (at fixed false positive rate of 2%) over that of the four alternative methods from 50.8%-58.4% to about 62% which corresponds to 3.6%-11.2% improvement.

A self-paced brain interface system that uses movement related potentials and changes in the power of brain rhythms.

Movement execution results in the simultaneous generation of movement-related potentials (MRP) as well as changes in the power of Mu and Beta rhythms. This paper proposes a new self-paced multi-channel BI that combines features extracted from MRPs and from changes in the power of Mu and Beta rhythms. We developed a new algorithm to classify the high-dimensional feature space. It uses a two-stage multiple-classifier system (MCS). First, an MCS classifies each neurological phenomenon separately using the information extracted from specific EEG channels (EEG channels are selected by a genetic algorithm). In the second stage, another MCS combines the outputs of MCSs developed in the first stage. Analysis of the data of four able-bodied subjects shows the superior performance of the proposed algorithm compared with a scheme where the features were all combined in a single feature vector and then classified.

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.

Towards development of a 3-state self-paced brain-computer interface.

Most existing brain-computer interfaces (BCIs) detect specific mental activity in a so-called synchronous paradigm. Unlike synchronous systems which are operational at specific system-defined periods, self-paced (asynchronous) interfaces have the advantage of being operational at all times. The low-frequency asynchronous switch design (LF-ASD) is a 2-state self-paced BCI that detects the presence of a specific finger movement in the ongoing EEG. Recent evaluations of the 2-state LF-ASD show an average true positive rate of 41% at the fixed false positive rate of 1%. This paper proposes two designs for a 3-state self-paced BCI that is capable of handling idle brain state. The two proposed designs aim at detecting right- and left-hand extensions from the ongoing EEG. They are formed of two consecutive detectors. The first detects the presence of a right- or a left-hand movement and the second classifies the detected movement as a right or a left one. In an offline analysis of the EEG data collected from four able-bodied individuals, the 3-state brain-computer interface shows a comparable performance with a 2-state system and significant performance improvement if used as a 2-state BCI, that is, in detecting the presence of a right- or a left-hand movement (regardless of the type of movement). It has an average true positive rate of 37.5% and 42.8% (at false positives rate of 1%) in detecting right- and left-hand extensions, respectively, in the context of a 3-state self-paced BCI and average detection rate of 58.1% (at false positive rate of 1%) in the context of a 2-state self-paced BCI.

Automatic user customization for improving the performance of a self-paced brain interface system.

Customizing the parameter values of brain interface (BI) systems by a human expert has the advantage of being fast and computationally efficient. However, as the number of users and EEG channels grows, this process becomes increasingly time consuming and exhausting. Manual customization also introduces inaccuracies in the estimation of the parameter values. In this paper, the performance of a self-paced BI system whose design parameter values were automatically user customized using a genetic algorithm (GA) is studied. The GA automatically estimates the shapes of movement-related potentials (MRPs), whose features are then extracted to drive the BI. Offline analysis of the data of eight subjects revealed that automatic user customization improved the true positive (TP) rate of the system by an average of 6.68% over that whose customization was carried out by a human expert, i.e., by visually inspecting the MRP templates. On average, the best improvement in the TP rate (an average of 9.82%) was achieved for four individuals with spinal cord injury. In this case, the visual estimation of the parameter values of the MRP templates was very difficult because of the highly noisy nature of the EEG signals. For four able-bodied subjects, for which the MRP templates were less noisy, the automatic user customization led to an average improvement of 3.58% in the TP rate. The results also show that the inter-subject variability of the TP rate is also reduced compared to the case when user customization is carried out by a human expert. These findings provide some primary evidence that automatic user customization leads to beneficial results in the design of a self-paced BI for individuals with spinal cord injury.