Improving the visual stimuli for brain-computer interfaces

Abstract

Visual stimuli are traditionally used to operate Brain-Computer Interface (BCI) systems employing the Steady State Visual Evoked Potentials (SSVEP) technique. In this technique, when a BCI user’s visual system is exposed to a flickering stimulus, the brain will generate oscillations with matching frequencies. Such signals will allow the user to control external devices such as computers, games, robots or drones. In non-invasive BCIs, the brain signals available are very weak, usually in the range of microvolts, thus various signal processing methods must be employed in order to facilitate successful system operations. It has been recognised that despite great efforts made by the research community to improve brain signal processing algorithms, improvements can be made at the very source – the brain itself.

A variety of novel testing scenarios have been conceptualised, designed and realised with the sole purpose of formulating and creating flickering graphic that would address this problem by eliciting the strongest, artefacts free, and minimal noise, brain signals. All parameters were individually tested, to evaluate and determine their impact on the brain signals. Among the parameters chosen for evaluation are flicker sizes, patterns, displayed backgrounds, and colours. Specifically, great attention has been put to test the impact of colours. Such determination for the colour flicker testing was directly derived from the fact that the human retina is constructed of photosensitive cells detecting precisely the same primary colours, which are found in digital displays used to present said flickers in the form of pixels. The introduction of colour flickers, not only provided better results, also filled a gap in research: that is, the use of colours in visual stimulation. In order to achieve the proposed experimental scenarios several hardware and software tools have been used. A bespoke iOS application capable of generating large variety of colour and black-white flickers has been used to elicit brain signals during experiments. Nearly two thousand of raw brain signals have been recorded and analysed. In the final stage of the research, the determined best performing flicker parameters have been used to generate flickers for an off-line BCI application. The resulting flickers delivered strong and clean brain signals, which validated the success of this research.