| The concept of BCI was first proposed in 1970 s and has been developed for several decades. The original aim is to assistant the disabled or severe neuromuscular damaged people who can not live independently to re-join their normal life. Current the BCI concept has been developed and applied to many other fields, such as in the field of rehabilitation medicine, entertainment, special operation and biological identification, etc..In the use of single EEG features as the system input traditional BCIs do face some shortcomings. For example, classic P300 Speller system needs to run all rows and columns stimulus to finally recognize a character. Problems result a low transfer rate and poor execution efficiency and so on.The aim of developing a hybrid brain-computer interface is to apply multi-feature of EEG signals or combine a BCI with another type of interface so to utilize more information and achieve more efficiency in multitasking than traditional BCI which is based on the single EEG feature. So this paper is proposed a simultaneous hybrid BCI based on SSVEP and MVEP.In this paper, a novel hybrid BCI combined SSVEP and MVEP simultaneously is proposed. In a 3 by 3 character spelling matrix, vertical white band in columns of characters are flicked at a set of fixed frequency rate to induce SSVEP and a random moving bar in rows is to induce MVEP. 8 subjects participated in the hybrid paradigm and the MVEP-only paradigm. The signal feature analysis is applied in time domain, frequency domain and spatial domain. Results for the proposed hybrid paradigm show that after multiple superimposed average, characteristic component of MVEP--N200 induced at parietotemporal lobes, and there is significant latency of wave N200 between hybrid paradigm and MVEP-only paradigm, P<0.05. SSVEP can be detected at occipital lobe by periodogram, the amplitudes of occipital leads decreased with the increased distance from Oz channel. An interesting and special EEG feature here we named SSVEP blocking phenomenon that two of the subjects’ SSVEP frequency do not remain constant and uninterrupted as time changes, is found by the STFT analysis at Oz channel. Furthermore, this special feature is confirmed by the CCA analysis at the occipital region that when the movement appears, the correlation coefficient of ρ for induced EEG signals in occipital region and the stimulus frequency is less than the movement doesn’t appear, and the evidence is considered that a sudden movement affects and blocks SSVEP response signals. However this feature varies individually and further studies are needed to analyze more on the cause. Finally, support vector machine(SVM)is designed to classify the EEG signals and the classification accuracy, and the transfer rate are calculated separately. Results indicate that the hybrid paradigm is better than MVEP-only paradigm in term of the classification accuracy and the transfer rate. For the final performance, the average classification accuracy of hybrid paradigm can be improved by more than 3% when the target is present for 2 times. When the target stimulus is presented and enhanced for just 3 times, the character classification accuracy of the proposed hybrid paradigm is able to achieve more than 90%. The highest ITR of hybrid paradigm is up to 40bit/min.The proposed simultaneous hybrid BCI paradigm is able to induce two EEG signals simultaneously and the goal character determined by a row and a column stimulus which the subject gazed can be detected through the feature extraction and the signal classification. Thanks to the combination of dual EEG features, the system performance is improved in terms of reducing running time so that can improve the system processing speed and reduce the visual fatigue. In conclusion, the proposed simultaneous hybrid BCI paradigm provides a theory basis and a technical support to create a more practical BCI system. |
PDF Full Text Request