Online Hybrid Brain-Computer Interface Technique Serially Combining Multiple Physiological Signals

Brain-computer interface (BCI) systems allow users to convey informationbetween the brain and peripheral unit or outward environment independent ofperipheral nerves and muscular pathways. A hybrid BCI combined with at least oneother system or human-machine interact device to help people send information andcould provide higher information transfer rate than classical BCIs. Moreover, hBCIscould solve to some extent the BCI illiteracy. Thus hBCI had become a new idea ofBCI design which had attracted much attention and received great compliments inrecent years.In this thesis, the hBCI included three kinds of brain-computer orhuman-machine interact paradigm such as P300based BCI, SSVEP based BCI andocclusion evoked scalp-myoelectricity (OES-ME) based HMI with independentrhythm. Research firstly constructed independent subsystems of the three paradigms,and tested each of them with online experiments. Results showed that the highestinformation transfer rate reached24.62bits/min with P300based BCI and34.29bits/min with SSVEP based BCI. Subjects could operate the OES-ME based HMIwell with an average time of one second in single-step operation and an averageaccuracy up to92.9%in instructions output. Based on the simple BCI and HCIsystems, this thesis successfully constructed a serial asynchronous hybrid BCI systemcombining multiple physiologic features. This system included four modules: thevisual stimulation part, the signal recording part, the user interface and signalprocessing part, and the control part. The visual stimulation part provided the P300speller screen and SSVEP flicker stimulation. The signal recoding part was able torecord the EEG and receive the triggers in real time. The user interface and signalprocessing part would send commands to the control part after reading and processingthe EEG data and the triggers. The control part received the commands, transferredthem to actions, such as cursor motion and click, and character input, and providedthe auditory feedback. The proposed system had the functions of typing letters,moving cursor and mouse-click, and the switch among them. Thus it could help theusers browse the webpage.5subjects were involved in this study. They all finishedthe mentioned functions in the online tests. Results showed that the practical operation had no more than7operations compared to the theoretical operation. Thesubjects achieved an accuracy above90%after training, which met the requirement inadvance.This thesis realized a hybrid brain-computer interface combining threeparadigms, P300-speller, SSVEP-based BCI and OESME-HCI, which was compatible.Moreover, it could improve the BCI performance and provide technical support forclinic application.