Tracking gaze position from EEG: Exploring the possibility of an EEG-based virtual eye-tracker

Area of Science:

• Neuroscience and Biomedical Engineering
• Signal Processing and Machine Learning

Background:

• Ocular artifacts in electroencephalogram (EEG) signals traditionally impede research.
• Blind Source Separation (BSS) methods like Independent Component Analysis (ICA) and Second-Order Blind Identification (SOBI) are crucial for improving neural signal quality.
• A novel method combining SOBI and Discriminant and Similarity (DANS) identification effectively extracts eye movement components from EEG with high accuracy (>95% localization).

Purpose of the Study:

• To introduce a proof-of-concept for an EEG-based virtual eye-tracker (EEG-VET).
• To demonstrate the feasibility of tracking gaze position using only EEG signals.
• To facilitate the study of neural mechanisms underlying cognition during natural eye movements.

Main Methods:

• Development of the EEG-VET system, integrating SOBI for signal component separation.
• Utilizing a DANS algorithm for automated identification of ocular components within EEG data.
• Employing a linear model to translate identified ocular components into precise gaze positions.

Main Results:

• The EEG-VET prototype achieved high accuracy (0.920° best, 1.008° ± 0.357° average) and precision (1.510° best, 2.348° ± 0.580° average) in visual angle measurements.
• The system successfully tracked eye movements using EEG data alone across 18 participants.
• Ocular components were localized within ocular structures with a high goodness of fit (>95%).

Conclusions:

• This study presents a novel approach for co-registering eye movement and neural signals from a single EEG recording.
• The EEG-VET enhances the ease of studying neural mechanisms related to natural cognition and free eye movement.
• EEG-derived SOBI components show potential for building predictive models for gaze tracking.