||Collaborative brain computer interface (cBCI) is an emerging track in BCI research to make a control decision through the engagement of electroencephalogram (EEG) from multiple individuals concurrently. Leveraging multi-individual EEG signal enables to boost signal-to-noise ratio and improve the BCI performance with respect to a conventional single-individual BCI. Most cBCI works were done by employing multiple laboratory-oriented EEG-sensing devices with wet electrodes, which led to cost-inefficient, time-consuming and labor-intensive setup. This work thus developed a cost-efficient multi-individual EEG acquisition platform, featuring dry electrodes and wireless transmission.|
The proposed multi-individual EEG system is composed of a behavioral stimulation platform, portable EEG acquisition devices, and a data receiving platform. The core EEG device was thoroughly verified by the aspects of power consumption, frequency responses, synchronization of EEG and event, and multi-individual EEG acquisition. At this stage, the system implemented a scenario for three individuals and verified the practicality through a 10-day auditory oddball paradigm. Event related potential (ERP) analysis was then performed to derive attention-related EEG responses across individuals and days.
The results showed that the device could be functional for 3.5 hours with a 9-V battery (power consumption was about 250 mW), and its actual frequency response of the designed bandpass filter (0.6 ~ 56.5 Hz) differed from the ideal simulation with an acceptable range less than 2.5 dB at cut-off frequencies. As for the EEG-event synchronization outcome derived in a 30-min test of 1800 events per second, the averaged error of signal devices was 2.8 ± 1.86 ms with a maximum of 15 ms (occurrence was only 0.05%), whereas the cross-device counterpart led to an averaged error 1.44 ± 2 ms, with a maximum of 16 ms (0.05%). Furthermore, the ERP results showed that most single-day EEG sessions exhibited a prominent P300 amplitude for target sounds against non-target sounds (p<0.01). The cross-individual and cross-day ERP results consistently exhibited N100 and P300 components as well. The above outcomes evidently demonstrated the integrity of the multi-individual EEG acquisition system developed in the present work. In the feature, the developed system not only facilitates multi-individual brain research as well as practical cBCI applications, such as monitoring students’ attention in class, social activities, and neurogaming.
Keywords: collaborative brain computer interface, multi-individual electroencephalogram acquisition, event-related potential, EEG signal processing, analog front end.