F. Ibrahim, N.A. Abu Osman, J. Usman and N.A. Kadri (Eds.): Biomed 06, IFMBE Proceedings 15, pp. 352-354, 2007 www.springerlink.com ' Springer-Verlag Berlin Heidelberg 2007 ---------------------------------------------------------------- IFMBE Proceedings Vol. 15 --------------------------------------------------------------- Assessment of Steady-State Visual Evoked Potential for Brain Computer Communication R. S. Leow 1 , F. Ibrahim 1 and M. Moghavvemi 2 1 Department of Biomedical Engineering, University of Malaya, Malaysia 2 Department of Electrical Engineering, University of Malaya, Malaysia Abstract—This paper describes the investigation of the steady-state visual evoked potential (SSVEP) response elicited by a flickering visual stimulus. Preliminary results from five subjects have shown that SSVEP have certain characteristics including good signal to noise ratio (SNR), minimal user train- ing and number of electrodes, despite constraints such as the need for a visual stimulating apparatus and possible induced of visual fatigue. Based on these promising results, SSVEP can be used as a tool for controlling signal for future brain computer interface (BCI) application. Keywords—brain computer interface (BCI), electroencephalo- graphy (EEG), steady-state visual evoked potential (SSVEP) I. INTRODUCTION A brain-computer interface (BCI) is a device that allows human to communicate with a computer by using brain signals. For individuals who have severe neuromuscular disorders, BCI can serve as an alternative method of com- munication or control so that the communication process does not have to depend on the brain s normal output path- ways of peripheral nerves and muscles [1]. The input to the BCI system can be obtained via invasive or non-invasive methods. The non-invasive electroencepha- lographic (EEG) signals acquired from human scalp are the more favorable method nowadays due to its simple and safer approach, although its signal to noise ratio (SNR) is less prominent [2]. By measuring and analyzing the EEG signals, specific features of brain activity can be translated into desired control commands. There are several types of EEG activities that can be util- ized as input features for BCI systems, e.g. slow cortical potentials [3], oscillatory EEG activity [4], P300 potential [5] and visual evoked potential [6]. The input feature is usually selected based on their effect on information trans- fer rate and reliability of the BCI system, besides consider- ing other factors such as its applicability for majority indi- viduals and training period required. Steady-state visual evoked potential (SSVEP) is a peri- odic response elicited in the brain by visual spatial attention on flickering stimulus at frequency of 6Hz and above [7]. SSVEPs have the same fundamental frequency as the stimu- lating frequency, but they also include higher and / or sub- harmonic frequencies under some situation [8]. SSVEPs are usually recorded from the occipital region of the scalp. Compared to other types of EEG features, SSVEP has better SNR. It is phase-locked to the triggering source. Therefore, simple frequency domain algorithms can be used to extract SSVEP signals [9]. For SSVEP-based BCI system, a flickering apparatus is needed to provide visual stimulus to the subject. Therefore, many of the applications are for subjects who have the abil- ity to control their eye movement but not by those with severe ocular motor impairment [7]. In this study, SSVEP is being investigated as an ap- proach for future BCI applications. The results are very promising. II. SUBJECTS AND METHODS A. Subjects A total of five voluntary healthy subjects consisted of three females and two males, aged between 22 and 24 were recruited in the study. All of them have normal or corrected- to-normal vision. Subjects were briefed about the overall procedure of the experiment, and required to sign a consent form. B. Experimental Procedure Subject was seated comfortably on a chair. The visual stimulus was placed 1.5 m in front of the subject, at about the same height as the subject s eyes. The visual stimulus was composed of a 2cm diameter red color light emitting diode (LED), with the flickering frequency controlled by a programmable microcontroller. Frequencies ranging from 7Hz to 35Hz were tested. Before the experiment, few minutes of spontaneous EEG data with subjects closing their eyes were recorded to col- lect some brain alpha activity. After that, subject was given a few minutes time to adapt to the flickering stimulus before the experiment started. During the experiment, subjects were required to maintain their gaze on the stimulus device. In each session of the experiment, the subject concentrated