Wavelet transform (WT) is an important tool to analyze the time-frequency structure of a signal. The WT relies on a prototype signal that is called the mother wavelet. However, there is no single universal wavelet that fits all signals. Thus, the selection of mother wavelet function might be challenging to represent the signal to achieve the optimum performance. There are some studies to determine the optimal mother wavelet for other biomedical signals, however, there exists no evaluation for Steady-State Visually-Evoked Potentials (SSVEP) signals that becomes very popular among signals manipulated for brain-computer interfaces (BCIs) recently. This study aims to explore if any, the mother wavelet that suits best to represent SSVEP signals for classification purposes in BCIs. In this study, three common wavelet-based features (variance, energy, and entropy) extracted from SSVEP signals for five distinct EEG frequency bands (delta, theta, alpha, beta, and gamma) were classified to determine three different user commands using six fundamental classifier algorithms. The study was repeated for six different commonly-used mother wavelet functions (Haar, Daubechies, Symlet, Coiflet, Biorthogonal, and Reverse Biorthogonal). The best discrimination was obtained as the accuracy of 100% with the average of 75.85%. Besides, Ensemble Learner gives the highest accuracies for half of the trials. Haar wavelet had the best performance in representing SSVEP signals among other all mother wavelets adopted in this study. Concomitantly, all three features of energy, variance, and entropy should be used together since none of these features had superior classifier performance alone.

No abstract

Brain-computer interface (BCI) system based on steady-state visual evoked potentials (SSVEP) have been acceleratingly used in different application areas from entertainment to rehabilitation, like clinical neuroscience, cognitive, and use of engineering researches. Of various electroencephalography paradigms, SSVEP-based BCI systems enable apoplectic people to communicate with outside world easily, due to their simple system structure, short or no training time, high temporal resolution, high information transfer rate, and affordable by comparing to other methods. SSVEP-based BCIs use multiple visual stimuli flickering at different frequencies to generate distinct commands. In this paper, we compared the classifier performances of combinations of binary commands flickering at seven different frequencies to determine which frequency pair gives the highest performance using temporal and spectral methods. For SSVEP frequency recognition, in total 25 temporal change characteristics of the signals and 15 frequency-based feature vectors extracted from the SSVEP signal. These feature vectors were applied to the input of seven well-known machine learning algorithms (Decision Tree, Discriminant Analysis, Logistic Regression, Naive Bayes, Support Vector Machines, Nearest Neighbour, and Ensemble Learning). In conclusion, we achieved 100% accuracy in 7.5 - 10 frequency pairs among these 2,520 distinct runs and we found that the most successful classifier is the Ensemble Learning classifier. The combination of these methods leads to an appropriate detailed and comparative analysis that represents the robustness and effectiveness of classical approaches.

Motor Imaginary (MI) electroencephalography (EEG) signals are obtained when a subject imagines a task without essentially applying it. The accurate decoding of MI EEG signals plays an important role in the design of brain-computer interface (BCI) systems due to the use of these signals in the rehabilitation process of paralyzed patients in recent studies. In this study, two different MI tasks were tried to be differentiated by extracting time-domain and frequency-domain features from 22 channel EEG signals and determining best combination of important and distinctive features based on statistical significance. MI EEG signals were supplied from BCI Competition IV Dataset-IIa. These features were differentiated using 25 different classification algorithms and 5-fold cross-validation method. The repeatability of the results was examined testing each algorithm 10 times. As a result, the highest average accuracy rate of 60.69% was calculated in the Quadratic Support Vector Machine (SVM) using all features and 62.52% in the Ensemble Subspace Discriminant (ESD) algorithm using only the selected features by the independent t-test. The results showed that the independent t-test based feature selection increased the performance in 20 classifiers, and decreased the performance in 5 classifiers. Also, the effectiveness of the feature selection method examined using the paired-sample t-test which is known as repeated measures t-test. The significance value, p-value was found as 0.04. Therefore, the independent t-test based feature selection method is an effective feature selection method and is providing the significant improvement in classifier performance.

In recent studies, in the field of Brain-Computer Interface (BCI), researchers have focused on Motor Imagery tasks. Motor Imagery-based electroencephalogram (EEG) signals provide the interaction and communication between the paralyzed patients and the outside world for moving and controlling external devices such as wheelchair and moving cursors. However, current approaches in the Motor Imagery-BCI system design require effective feature extraction methods and classification algorithms to acquire discriminative features from EEG signals due to the non-linear and non-stationary structure of EEG signals. This study investigates the effect of statistical significance-based feature selection on binary and multi-class Motor Imagery EEG signal classifications. In the feature extraction process performed 24 different time-domain features, 15 different frequency-domain features which are energy, variance, and entropy of Fourier transform within five EEG frequency subbands, 15 different time-frequency domain features which are energy, variance, and entropy of Wavelet transform based on five EEG frequency subbands, and 4 different Poincare plot-based non-linear parameters are extracted from each EEG channel. A total of 1,364 Motor Imagery EEG features are supplied from 22 channel EEG signals for each input EEG data. In the statistical significance-based feature selection process, the best one among all possible combinations of these features is tried to be determined using the independent t-test and one-way analysis of variance (ANOVA) test on binary and multi-class Motor Imagery EEG signal classifications, respectively. The whole extracted feature set and the feature set that contain statistically significant features only are classified in this study. We implemented 6 and 7 different classifiers in multi-class and binary (two-class) classification tasks, respectively. The classification process is evaluated using the five-fold cross-validation method, and each classification algorithm is tested 10 times. These repeated tests provide to check the repeatability of the results. The maximum of 61.86 and 47.36% for the two-class and four-class scenarios, respectively, are obtained with Ensemble Subspace Discriminant among all these classifiers using selected features including only statistically significant features. The results reveal that the introduced statistical significance-based feature selection approach improves the classifier performances by achieving higher classifier performances with fewer relevant components in Motor Imagery task classification. In conclusion, the main contribution of the presented study is two-fold evaluation of non-linear parameters as an alternative to the commonly used features and the prediction of multiple Motor Imagery tasks using statistically significant features.

No abstract

Eight distinct stimulation frequencies can be discriminated with the average of 36.1%.  Reverse Biorthogonal mother wavelet gives the maximum accuracy.  Ensemble learners give maximum accuracy among classifiers Figure A. Block diagram of the study Purpose: This study focuses on determining the stimulating frequency of a flashing image on the computer screen via electroencephalography (EEG) signals to investigate whether there is a correlation between stimulation frequency and brain activities. Theory and Methods: Some researches indicate an increase in classifier performances for brain-computer interfaces (BCIs) when objects are shown as blinking on the screen. The recorded EEG signals while applying a blinking image on a screen are called steady-state visually-evoked potentials (SSVEP). Most of the studies tried to discriminate which object is gazed while recording. We tried to determine the stimulating frequency from this dataset using an open-source SSVEP dataset. Wavelet features of variance, energy, and entropy were calculated from six different mother wavelets (Haar, Daubechies, Symlet, Coiflet, Biorthogonal, Reverse Biorthogonal). All features and ANOVA-selected features were applied to distinct classifiers (Decision Trees, Discriminant Analysis, Naive Bayes, Support Vector Machines, k-Nearest Neighbors, and Ensemble Learners). Results: The mother wavelet of Reverse Biorthogonal gives the maximum accuracy by using all classifiers. The ensemble learner determines the stimulating frequency correctly with a maximum accuracy of 37.9%. On the other hand, ANOVA-based feature selection decreases the classifier performances for all subjects. Conclusion: Results indicate that the stimulating frequency is embedded into SSVEP signals. We concluded that SSVEPbased BCI studies for object classification or decision making should be conducted carefully since not only the location of the blinking object but also its blinking frequency is the part of EEG signals. In conclusion, it will be possible to detect gazed object among a few pictures by showing the flickering objects at the different frequencies at the same time.

No abstract