A Frequency-Domain Pattern Recognition Model for Motor Imagery-Based Brain-Computer Interface

Abstract

Brain-computer interface (BCI) is an appropriate technique for totally paralyzed people with a healthy brain. BCI based motor imagery (MI) is a common approach and widely used in neuroscience, rehabilitation engineering, as well as wheelchair control. In a BCI based wheelchair control system the procedure of pattern recognition in term of preprocessing, feature extraction, and classification plays a significant role in system performance. Otherwise, the recognition errors can lead to the wrong command that will put the user in unsafe conditions. The main objectives of this study are to develop a generic pattern recognition model-based EEG –MI Brain-computer interfaces for wheelchair steering control. In term of preprocessing, signal filtering, and segmentation, multiple time window was used for de-noising and finding the MI feedback. In term of feature extraction, five statistical features namely (mean, median, min, max, and standard deviation) were used for extracting signal features in the frequency domain. In term of feature classification, seven machine learning were used towards finding the single and hybrid classifier for the generic model. For validation, EEG data from BCI Competition dataset (Graz University) were used to validate the developed generic pattern recognition model. The obtained result of this study as the following: (1) from the preprocessing perspective it was seen that the two-second time window is optimal for extracting MI signal feedback. (2) statistical features are seen have a good efficiency for extracting EEG-MI features in the frequency domain. (3) Classification using (MLP-LR) is perfect in a frequency domain based generic pattern recognition model. Finally, it can be concluded that the generic pattern recognition model-based hybrid classifier is efficient and can be deployed in a real-time EEG-MI based wheelchair control system.