A Practical Chinese Auxiliary Communication System Based On Brain-computer Interface

Brain-computer interface(BCI)system based on steady-state visual evoked potential(SSVEP)signal has been widely studied due to its relatively high signal-to-noise ratio,high information transmission rate,and less training.Text input is an important application of BCI,which allows patients with motor and speech impairments to use neural signals to communicate with the outside world.Most of the existing BCI text input systems only support alphabetic script such as English,and Chinese as a logographic script is usually inefficient for inputting using a BCI system due to its complex input process.To solve this problem,this paper starting from the recognition algorithm and the human-machine interaction design aspect,developed a practical Chinese auxiliary communication system based on the SSVEP-BCI.At the level of SSVEP recognition algorithm,this study proposed an asynchronous training-free SSVEP-BCI detection algorithm for non-equal prior probability scenarios.The algorithm is based on spatio-temporal equalization multi-window technology(STE-MW),introduces the maximum a posteriori criterion(MAP),and makes full use of prior information to improve the performance of asynchronous training-free BCI systems.The online free control experiment of 10-target simulated vehicle control with 17 healthy subjects showed that the classification accuracy of the proposed algorithm was 99.21±0.15%,and the occurrence of false alarms was only 0.19 times/min.On this basis,this paper expanded the number of candidate target of the proposed algorithm to 40,and applied it to the Chinese input system based on the Shuang-pin scheme.This paper used statistical methods to obtain the prior probability distribution of the initials and finals targets,and dynamically adjusted the prior probability distribution of the finals target to achieve the effect of dynamically reducing the number of candidate targets.The online experiment showed that the classification accuracy of the proposed algorithm achieved 98.16%±0.32%,and the practical bit rate(PBR)achieved 80.22±0.78 bits/min.At the level of human-machine interaction design,this paper first fully optimized the Chinese input stimulus paradigm and the sequence of keystrokes,and strived to minimize the distance of the user’s head movement during use.In order to further improve the input efficiency of Chinese,the intelligent association function of the input method was introduced into the Chinese input system of BCI.On the basis of Chinese input,the input function of English and numbers was added,so that users can convey information to the outside world more completely.Finally,this paper integrated the above contents,and formed a practical Chinese auxiliary communication system that is efficient,interactive-friendly and user-targeted.Free input experiment showed that the system achieved an average Chinese character input rate of 10.93±0.3 8 s/word without pretraining.