Automatic Noise Removal and Phase Synchronization for MI EEG Analysis

1. In this book, a novel brain-computer interface (BCI) system is proposed to analyze motor imagery (MI) electroencephalogram (EEG) signals. 2. After eliminating EOG artifacts automatically and extracting features by wavelet-based phase synchronization approach, support vector machine (SVM) is adopted for the classification of single-trial left and right MI data. 3. The EOG artifacts are automatically removed by means of modified independent component analysis (ICA). 4. The features are extracted from wavelet data by phase synchronization, and then classified by the SVM. 5. Compared with the results without EOG artifact removal, spectral band and AR model features, the proposed system achieves satisfactory results in BCI applications.