DTW Based SSVEP Recognition Method For A Portable Brain-computer Interface System

The adoption of brain computer interface(BCI)has grown tremendously in various fields ranging from healthcare to entertainment.BCI is used to create a means of communication between the human brain and an external device to enable humans express their intentions directly by the brain.Steady state visual evoked potential(SSVEP)based BCI system has a great potential be adopted by users for everyday use.This is because it requires little or no subject training and has a high information transfer rate(ITR).However,most SSVEP based BCI systems employ the use of cumbersome stimulus display device and costly EEG recording device.This leads to a non-portable BCI system which cannot be easily moved from one place to another.This work solves this problem by incorporating the use of an Emotiv EPOC headset,which is a low cost and portable EEG recording device,and a smart mobile phone,which is used for presenting the stimulus,in the design of a portable BCI system.Considering the problem of traditional stimuli which cause irritation to the eye.this work adopted an approach based on changes in the gray values for the stimulus design.The stimulus produced using this approach is user friendly.Compared to traditional expensive EEG recording devices with high sampling frequency.EEG signals recorded by Emotiv EPOC headset are usually of low quality and contain little details majorly due to its low sampling frequency(128 Hz).Therefore,the problem of recognizing SSVEP response in the portable brain computer interface(BCI) system is critical.Currently,canonical correlation analysis(CCA)is widely used to recognize SSVEP response.To develop a better approach for recognizing SSVEP from a data with low sampling frequency,in this work,an alternative approach based on dynamic time warping(DTW)is proposed for recognizing SSVEP response.The approach can be effectively used to analyze EEG data recorded by the Emotiv EPOC device.Basically,the proposed DTW-based approach recognizes SSVEP response by measuring the similarity between an EEG signal and a set of templates.Therefore,the type of the template used,may influence the performance of the DTW-based approach.To select the best template for the proposed approach,several templates were constructed and examined.These templates include artificial reference signals,discrete wavelet transform template(DWTT),multiset template(MT)and individual template(IT).The results showed that the DWTT performed better than all other templates.The performance of the proposed approach was compared with that of the standard CCA.The experiment shows that our proposed approach obtained better result than the standard CCA approach.DTW obtained an overall accuracy of 83.63%while CCA obtained an accuracy of 65.71%,showing a promising future for DTW-based approach to be used in the SSVEP based BCI applications.