An SSVEP-based BCI with LEDs visual stimuli using dynamic window CCA algorithm

Abstract

The improvements of the visual stimulus paradigm and the target recognition algorithm in the (steady-state visual evoked potential) SSVEP-based (brain computer interface) BCI have attracted widespread attention among BCI researchers. The LEDs visual stimulus paradigm was adopted in this paper, which is more portable and lighter than the widely used computer screen visual stimulus paradigm. In terms of algorithms, the ‘dynamic window’ strategy was introduced in canonical correlation analysis (CCA). The dynamic window CCA can automatically find the optimal window length of data in the recognition process, and this could increase the information transfer rate (ITR) of the system. An offline experiment with LEDs visual stimulus paradigm was designed to evaluate the performance of dynamic window CCA. The results showed that the LEDs visual stimulus paradigm is feasible in the SSVEP-based BCI, and the dynamic window CCA got the higher ITR compared with CCA. In this paper, an online experiment was also conducted using dynamic window CCA. The online SSVEP system used the LEDs visual stimulus paradigm to control an artificial hand. The results showed that the dynamic window CCA is well suited for online systems.