Motor Imagery Electroencephalogram Analysis Using Adaptive Neural-Fuzzy Classification

In this study, an adaptive neural-fuzzy analysis system is proposed for single-trial classification of motor imagery (MI) electroencephalogram (EEG) data. Associated with enhanced active segment selection and wavelet-fractal features, adaptive fuzzy neural network (AFNN) is used for the recognition of left and right MI data. In addition to continuous wavelet transform (CWT) and Student’s two-sample t-statistics, the 2D anisotropic Gaussian filter is proposed to further enhance the selection of active segments in the time-scale domain. Multiresolution fractal features are then extracted from wavelet data by using modified fractal dimension. Finally, fractal features are discriminated by AFNN clustering. The system is tested on two publicly available EEG datasets and compared with several popular approaches. The experimental results demonstrate that enhanced active segment selection can greatly improve the performance and AFNN clustering has a splendid potential in the applications of brain-computer interface (BCI) work.