Spread spectrum techniques in BCI: A review of auditory and visual BCI-systems using continuous and binary noise tagged stimuli

Abstract

Brain computer interfaces (BCIs) are an essential tool for locked-in patients, providing a link with the outside world. However, the type of stimuli used in most brain computer interface system may have a detrimental influence on the information transfer rates (ITR) achieved. Steady state evoked potentials (SSEP) have been well studied and applied in EEG-based BCI-systems, reaching high ITRs, but the number of SSEP-stimuli that can be applied simultaneously is limited. These responses also suffer from noise due to spontaneous oscillations that occur in the brain.

In spread spectrum techniques, signals are distributed over a broader bandwidth in a pseudorandom fashion, making them much more robust against interference from noise or cross-stimulus interactions. Broadband noise signals have been successfully used in other neurological disciplines, but have remained a relatively neglected class of stimuli in BCI-systems.

Spread spectrum elicited evoked potentials offer a valuable extension of the palette of AEPs and VEPs available for BCI-systems, as their favourable auto- and cross-correlation characteristics provide good anti-interference properties, which make them especially beneficial in systems using multiple simultaneously presented stimuli, like speller-setups.

This thesis gives an introduction into spread spectrum techniques and the pseudorandom noise sequences used herein. The limited number of auditory and visual BCI-systems using continuous and binary noise tagged stimuli are reviewed and compared. Some of these systems reached information transfer rates of >100 bits/min. showing the potential of this approach.