A comparison of MSI and CCA for SSVEP classification

The Brain Computer Interface (BCI) system allows those with dyskinesia to control the device using only brain signals. At present, Steady State Visual Evoked Potential (SSVEP) is highly valued because of its high accuracy of analysis and lack of training. How to effectively identify the SSVEP signal frequency is the key issue of SSVEP-BCI, and it is related to the advantages and disadvantages of the BCI system. In this paper, multivariate synchronization index and canonical correlation analysis methods are used to compare SSVEP signal classification, and the effects of the two methods on the SSVEP signal classification effect are discussed in terms of data length, number of leads, lead position, and number of harmonics of the reference signal. Six subjects participated in the experiment. The experimental results show that the multivariate synchronization index method performs better than the typical correlation analysis method when the time window is small and the data length is small. For SSVEP signal analysis, the accuracy of the lead position is the most fundamental factor affecting the frequency analysis algorithm.