Evaluation of the Target Positioning in a SSVEP-BCI

The use of applications and systems controlled by Brain Computer Interfaces (BCI) has been growing in the recent years. Steady-State Visual Evoked Potentials (SSVEPs)-BCI systems are a promising field which consists in using brain response to any visual stimuli flickering at a certain frequency as input to an interface. The aim of this study is to evaluate the best positioning of visual targets, between cross and square arrangements, flickering at different frequencies (6.66, 8.57, 10, 11.9 and 15 Hz) to be used in a SSVEP-BCI system. Four participants were positioned, individually, in front of a LCD screen to perform the experiment. The Welch’s Power Spectral Density (PSD) was extracted from the signal and used to select 9 optimal electroencephalogram (EEG) channels (O1, O2, Oz, PO5, PO6, PO4, PO3, PO7 and PO8) through a Z-Score statistical analysis. A Fast Fourier Transform (FFT) was performed as signal feature extraction and used as input for a classifier. Using a One-Against-All (OAA) Support Vector Machine (SVM) classifier, a 96.66% accuracy was obtained for the cross interface and an 85.88% accuracy for the square layout. Thus, we concluded that the best configuration is the cross arrangement. In addition, the best classification was obtained for the stimulus at 15 Hz.