EEG enhancement and context-awareness for robust SSVEP-based brain-computer interfaces

点击翻译按钮获取中文摘要

In Canada, hundreds of thousands of individuals are not able to communicate with their loved ones usingconventional methods such as speech or gestures. This can be due to severe disabilities, locked-in syndrome,neurodegenerative diseases such as amyotrophic lateral sclerosis (ALS), or autism spectrum disorders.Cognixion aims at giving these individuals a voice by tapping into their brain waves and converting uniqueneural signatures into pre-defined messages. More specifically, they utilize so-called steady-state visuallyevoked potentials (SSVEP) measured from electroencephalography (EEG). Their goal is to provide users witha low-cost, portable solution that can accompany them from home to the clinic without hassles. Mobility andreliance on low-cost EEG devices, however, significantly reduce signal quality and hamper systemperformance. This project aims to fill this gap by proposing three key innovations: (1) a new automated EEGenhancement algorithm, (2) a new EEG signal quality metric, and (3) the use of the metric to extract betterfeatures and to devise quality-aware machine learning algorithms. When combined, increased robustness toEEG artefacts will be possible, thus reducing system false alarms and allowing users to communicate moreeffectively, become more independent, and ultimately, improve their quality of life.