Assessment of Electrophysiological Responses during an Embodied Brain-Computer Interface training task in a Virtual Reality environment

Abstract

Brain-Computer Interfaces (BCIs) provide a non-muscular channel to stroke patients for assistive or restorative use. Motor Imagery (MI)-based BCIs can be complemented with Virtual Reality (VR) and haptic feedback to provide new, personalized rehabilitation protocols that may be suitable even for patients with severe motor impairments. The influence of those candidate protocols on human physiology has not yet been thoroughly examined. Thus, these candidate protocols are treated as experimental conditions. Two questions are addressed in this thesis: (1) How do physiological signals (Electroencephalography (EEG), Electrocardiography (ECG), Photoplethysmography (PPG), and respiration) vary across different experimental conditions? and (2) What underlying factors account for the observed variations in these physiological signals across different experimental conditions? To address these questions we analyzed the electrophysiological responses of 19 healthy subjects recruited to perform MI training through five different experimental conditions. These conditions incorporated various combinations of abstract vs. realistic feedback through the “NeuRow” virtual environment, Head-Mounted Display (HMD) vs. monitor, and with or without haptic feedback. The conditions were compared with actual motor execution data extracted from the same subjects. The results showed that all conditions require higher neurocognitive effort and may be more mentally demanding than motor execution. The condition of combined VR and haptics provides the highest brain activation and was chosen as the most suitable for a new, highly successful rehabilitation protocol.