Research And Development Of A Portable Auditory Brain-computer Interface

Brain-computer interface (BCI) is a human-computer interaction technology whichdoes not rely upon the motor nerves and muscle systems. Research on the practical BCIsystems has been of increasing interest in recent years. BCIs are applied mostly in theclinical context and on the consumer products, thus its portability is in great need in reallife applications. Auditory BCI uses the earphone to present the auditory stimulus,which is more preferable for a portable BCI. Based on the technology of wirelesselectroencephalography (EEG) and smart phone, a portable auditory BCI was designedand implemented in this study, with the paradigm using auditory spatial attention andactive mental response.To explore the possibility of portable online application, quantitative comparisonwas made between two auditory BCI paradigms. Both paradigms employ the activemental task and the difference between them is using spatial attention or not. Auditorystimulus with distinct spatial location in virtual sound field were synthesized byemploying the head related transfer functions (HRTFs), then presented via the stereoearphone. The spatial and non-spatial paradigms were compared in four aspects: eventrelated potential (ERP) spatio-temporal pattern, average ERP waveform, single ERPwaveform and the target detection performance. The comparison shows that the spatialparadigm has better performance than non-spatial one. Only two EEG channels wereused in the BCI classification for improving the portability of the whole system.A wireless electroencephalography (EEG) amplifier and a smart phone for dataanalyzing were used as the hardware of the portable auditory BCI. The BCI softwarewas designed and implemented according to the requirement of the aforementionedspatial paradigm. BCI classification was performed using the EEG from electrode Czand Pz by calculating the area underneath the event-related potential waveforms. For the5subjects participating in the online BCI experiment of phone dialing, an averageonline accuracy of77±4%was achieved.As our proposed system is highly portable with promising performances, it can bepotentially used both in clinical applications for disabled patients and as a new computer interface for healthy people.