No abstract

Robotic-assisted rehabilitation system based on Brain-Computer Interface (BCI) is an applicable solution for stroke survivors with a poorly functioning hemiparetic arm. The key technique for rehabilitation system is the feature extraction of Motor Imagery Electroencephalography (MI-EEG), which is a nonlinear time-varying and nonstationary signal with remarkable time-frequency characteristic. Though a few people have made efforts to explore the nonlinear nature from the perspective of manifold learning, they hardly take into full account both time-frequency feature and nonlinear nature. In this paper, a novel feature extraction method is proposed based on the Locally Linear Embedding (LLE) algorithm and DWT. The multiscale multiresolution analysis is implemented for MI-EEG by DWT. LLE is applied to the approximation components to extract the nonlinear features, and the statistics of the detail components are calculated to obtain the time-frequency features. Then, the two features are combined serially. A backpropagation neural network is optimized by genetic algorithm and employed as a classifier to evaluate the effectiveness of the proposed method. The experiment results of 10-fold cross validation on a public BCI Competition dataset show that the nonlinear features visually display obvious clustering distribution and the fused features improve the classification accuracy and stability. This paper successfully achieves application of manifold learning in BCI.

No abstract

No abstract

In order to get high classification accuracy, feature combination is commonly used in analysis of motor imagery electroencephalography EEG signal, including the nonlinear analysis and traditional time-frequency analysis. In this paper, Sample entropy(SampEn) was computed and represented as the nonlinear feature of motor imagery EEG signal for it can quantify the probability of new information appeared in time series. In addition, orthogonal empirical mode decomposition (OEMD) was also employed to extract the average energy of selected intrinsic mode functions(IMF) as the time-frequency feature for motor imagery EEG signal. Based on a public dataset, many experiments were conducted. Slide window was used to select the best time period for a better performance in feature extraction, and cross validation of 10 folds was applied in all the classification procedure. The highest recognition rate using SampEn and OEMD is respectively 86.07% and 83.21% classified by incremental support vector machine (ISVM) respectively. However, the highest classification rate of combined features is 86.79% by using ISVM which is a little higher than that of SampEn. Big linear correlation between SampEn and energy of IMF explains why the classification accuracy by combining the two types of features is not as high as expected.