Investigating brain-computer interaction in web-based systems

Abstract

Breakthroughs in technology have positively impacted the lives of individuals who su er from highly-restrictive physical disabilities, by allowing them to interact with machines through the use of brain signals, eye gazes and other techniques. This study focuses on investigating the use of Brain Computer Interfaces (BCIs), particularly those related to the eld of Steady State Visual Evoked Potentials (SSVEPs) for web interaction. An SSVEP is a neuronal response which takes place in the visual cortex of the brain whenever a person deliberately focuses attention on a visual stimulus which

ickers at a frequency of 6 Hertz or greater. The state of the art presents specialized desktop-based technologies, which are conventionally used to perform the rendering of stimuli, such as Psychtoolbox (for use with Matlab or GNU Octave). This study aims to determine whether SSVEP stimuli can be e ectively generated within a standard web environment, providing web developers with the possibility of developing lightweight and portable web-based SSVEP stimulation applications. To reach these aims, this study focused on the main technologies commonly available in major browsers, namely, CSS, JavaScript and WebGL. Three SSVEP stimuli-generators were developed with each of these technologies in order to be able to collate empirical evidence through a series of lab studies. Several performance metrics were used to determine how these di erent technologies performed when changing experimental variables, including hardware pro les, browser engines and operating systems, as well as stimulator frequency ranges and number of on-screen stimuli. Results showed that out of the three web technologies, WebGL attained the best overall performance in terms of accuracy and stability of the generated stimuli. This technology also outperformed the state of the art technology (Psychtoolbox, Matlab on Ubuntu), which is an encouraging outcome. In the second phase of the study, a proof of concept web-based SSVEP BCI system was implemented and evaluated with human subjects, from which several observations and insights were obtained.