ANALYZING STEADY-STATE VISUAL EVOKED POTENTIALS FOR EFFECTIVE USER RESPONSE DETECTION FOR BRAIN-COMPUTER INTERFACES

摘要

This article evaluates an M-order Adaptive Kalman filter analysis on Steady-State Visual Evoked Potentials (SSVEPs). This model is based on finding the original brain source signals from their combined observed EEG signals. At each time step, observed brain signals are filtered according to their ideal reference signals measured from 10, 11, 12 and 13 Hz LED stimuli. SSVEP response detection is based on maximum Signal to Noise Ratio (SNR) of the brain source signals. In each test, the average system accuracy is calculated with and without overlapped time-windows along with system Information Transfer Rate (ITR). The overall system accuracy and ITR are showing promising level of SSVEP detection for future online BCI systems.