A novel approach of independent brain-computer interface based on SSVEP

Abstract

Over the past ten years, Brain Computer Interfaces (BCIs) based on Steady- State Visual Evoked Potentials (SSVEP) have attracted the attention of many researchers due to the promissory results and the high accuracy rates achieved. This type of BCI provides to people with severe neuromotor difficulties the possibility to communicate with the world around them using visual attention modulation to blinking lights at a given frequency. This thesis aiming at developing a new approach of Independent BCI, in which users are not required to perform neuromuscular tasks to select visual targets, a feature that distinguishes it from traditional SSVEP-BCIs. Thus, people with severe motor disabilities as Amyotrophic Lateral Sclerosis (ALS) have a new alternative channel to communicate with the world around them using brain signals. Several contributions were done in this thesis, such as: improvement of the feature extractor called Multivariate Synchronization Index (MSI) for detecting evoked potentials; development of a new method for detecting evoked potentials through correlating multidimensional models (tensors); a first study on the influence of colored stimuli in SSVEPs detection using LEDs; the development of the concept of Compressive sensing applied to SSVEPs; and, finally, the development of a novel independent BCI under an approach named Figure-Ground Perception (FGP)