Research On Mixed Brain-Computer Interface Based On Imitating Reading And SSVEP

Brain-computer interface is a kind of information transmission channel which does not depend on normal neuromuscular channels.Its purpose is to provide a new way for patients with motor dysfunction to communicate with the outside world.In the past two decades,braincomputer interface technology has been improving with the efforts of many scholars and researchers,but it needs to be further improved in order to meet the needs of practical application.Most of the brain-computer interface systems currently studied analyze a single electroencephalogram characteristic component,which can recognize the simple behavior consciousness of the human brain.However,it is difficult to identify complex thinking task with single electroencephalogram feature component.In recent years,hybrid brain-computer interface,which is composed of multiple electroencephalogram signals,has become an important direction in the field of brain-computer interface research.The brain-computer interface system of "simulated reading" was put forward by the key Laboratory of Cognitive Science State people’s Commission of Central South University for nationalities.The experimental paradigm of simulated reading is to use a string of target and non-target characters to induce event-related potentials.Then the target character is identified by identifying the characteristics of the event-related potential.The input of a character in a two-page brain-controlled character spelling system designed on the basis of the simulated reading experiment paradigm requires the evoked and recognition of two event-related potentials,which leads to a long input time for a single character.In order to solve the above problems,this paper attempts to add a new electroencephalogram pattern to the brain-computer interface system of analog reading,and design a hybrid brain-computer interface system which combines the two electroencephalogram patterns.In this study,a hybrid brain-computer interface system(named IS-BCI system)based on analog reading and steady-state visual evoked potentials was designed with steady state visual evoked potential.The details are as follows:(1)The design of a hybrid brain-computer interface stimulus paradigm(named IS paradigm)based on analog reading and steady state visual evoked potential.The is paradigm combines two electroencephalogram patterns to produce two forms of stimulation.One is the target and non-target stimuli of the simulated reading-induced paradigm,the other is the scintillation stimulation under the steady state visual evoked potential-induced paradigm.The stimulus interface for the IS paradigm is set to a 7-row,7-column key arrangement,the key consists of two parts: the upper and lower parts.In the top half arethe names of the characters represented by the key,and the lower part is a small window,which is used to display stimulating characters.The simulated reading evoked paradigm uses a string composed of a string of target and non-target characters to trigger eventrelated potentials.The evoked normal form of steady-state visual evoked potentials is realized by exchanging two strings of different colors to achieve a fixed frequency of flickering.Both stimuli appear at the same time in the small window that shows the stimulus.Determining the column where the target character is located by detecting the position of the N2P3 induced by the target stimulation,judging the row where the target character is located by the target frequency identification.The position of the target character can be determined by determining the position of the row and column of the target character.(2)Pattern recognition and result analysis.There are two evoked electroencephalogram features in the mixed electroencephalogram signals collected under the IS paradigm,in which independent component analysis is carried out in the time domain,N2P3 features are extracted,and support vector machines are selected to classify the extracted features.In the frequency domain,the signal is first filtered in space to remove the influence of independent frequency,and then the typical correlation analysis is selected to complete the recognition and classification of the frequency feature.The effects of row-row spacing and time window length on the recognition accuracy of steady-state visual evoked potentials are also analyzed.The optimal row-row spacing and time window length are selected to improve the character recognition accuracy of IS-BCI system.The result shows that the recognition time of single character in analog reading brain-computer interface doublepage character spelling system is 5.7s,and the recognition time of single character in ISBCI character spelling system is only 3.45 s.That is,the integration of steady-state visual evoked potentials does not degrade the performance of the analog reading brain-computer interface system,and the information transmission rate of the newly designed IS-BCI system is nearly doubled compared with the simulated reading brain-computer interface character spelling system.