A Browser-Driven cVEP-based BCI Web Speller

By utilizing brain signals, the Brain-Computer Interface (BCI) enables non-muscular communication. In recent decades, BCI systems —particularly speller interfaces— have offered a variety of graphical user interfaces (GUIs). Many attempts had been made to improve the system's user-friendliness and speller speed. In this paper we present a web-based BCI speller based on Code-Modulated Visual Evoked Potentials (cVEPs), which can be accessed at https://bci-Iab.hochschule-rhein-waal.de/en/cvepspeller/. As a result, this web speller is now available to BCI researchers worldwide free of charge, and can be used with variety of own classifier applications. This web speller was successfully tested with the majority of modern web browsers. In the three-step web speller, each character can be selected by navigating through three distinct web interface screens. In this study the web-based cVEP speller was tested and compared to our “local speller” (incorporating the GUI and the signal processing in one application) by seven subjects, who were asked to spell the words “BCI_LAB” and “KLEVE”. All subjects were able to perform the spelling tasks, with a mean accuracy of 95.02% and an average Information Transfer Rate (ITR) of 42.55 bits/min, compared to our “local speller” with a mean accuracy of 93.81 % and an average ITR of 48.58 bits/min, respectively. The results showed similar values, confirming the suitability of the suggested web speller for the representation of the cVEP stimuli.