A rapidly growing number of studies demonstrate the potential benefits of Brain-Computer Interfaces (BCIs) in a wide range of use cases including education, workplace safety, and mental health. Recent technological advances have allowed a growing number of wearable BCIs appear both as research prototypes and on the market. The goal of this project was to develop and evaluate a BCI platform that integrates both electrical and metabolic sensing in a wearable glasses form-factor.
Our system combines data from three different physiological modalities: four channels of functional Near-Infrared Spectroscopy (fNIRS), three Electroencephalography (EEG) channels and two Electrooculography (EOG) channels. We compared different preprocessing and cognitive load classification strategies using data collected in a 14 subject mental arithmetic study. An SVM produced an average accuracy of 79% using EEG+EOG+fNIRS features compared with 62% and 63% for EEG and fNIRS respectively.