A Study Of EEG Characteristic Enhancement Through The Mental Arithmetic Task For Brain-Computer Interface

Event-related potential (ERP) is closely related to brain activity during the cognitive processe and it reflects the brain’s nerve electrophysiological changes. A brain-computer interface (BCI) system using ERP as electroencephalograph (EEG) feature carrier has the advantage of nondestructive collection, simple operation and a unique time resolution. However, due to the lack of spatial resolution, the signal-to-noise ratio is relative low, resulting in a low classification rate and a slow communication speed in BCI system, which becomes a bottleneck restricting the development of practical application.Mental arithmetic is an advanced cognitive processing activity that the multiple brain regions participate in. The type, contents and the size of mental arithmetic task have a major impact on evoked potentials in the frequency domain. Whether these characteristic components are able to increase the signal recognition rate of the brain-computer interface, thus improving the final performance of the system is the focus of this study.Based on the neuroscience and cognitive science, to solve the problems that in the traditional P300brain-computer interface system EEG signals provide less information and signal-to-noise ratio is low, a brain-computer interface paradigm under visual stimulus was proposed aiming to effectively activate the related brain areas and response signal while dealing with specific cognitive task (mental arithmetic task), so as to enhance the EEG feature.The research first designed and improved the traditional oddball experimental paradigm, introduced with mental arithmetic, including two different cognitive complexity experiments:the traditional counting task experiment with simple mental arithmetic and the more complicated experiment introduced with a mental arithmetic task. The collected EEG data was preprocessed, including removal of the interference, baseline correction, filtering and segmentation, followed by using coherent averaging method to extract signal features, then analyzed and compared the influences of different experimental paradigms on main components of event related potential.In the improved paradigm experiments the average increasing rate of P300 amplitude are6.83μV and73.94%; the brain activity from400ms is more severe and durable; besides, unlike traditional counting task, mental arithmetic task appear to have apparent activation at650ms. The results showed that the improved paradigm can better activate the related brain areas and enhanced the characteristics of signal; the evoked P300has a longer latency and larger amplitude.Then from a practical point of view, a P300paradigm using space character matrix as stimulus was designed. After preprocessing, the results further verified that the mental arithmetic task could enhance the EEG feature. Then the EEG signals were simply classified using support vector machine and compared the classification accuracy under the different input data templates or chosing the number of superimposed to determine the the best template and electrodes using as the input data of this experiment paradigm. The average classification accuracy was above85%.The results show that, through the introduction of more complex mental arithmetic cognitive tasks, the induced potentials signal characteristics were enhanced, also a higher classification rate was acquired. Applying mental arithmetic cognitive task in brain-computer interface technology helps to improve system performance and provides a possibility for the design and developement of the brain-computer interface system model.