Absolute Eye Gaze Estimation With Biosensors in Hearing Aids

Area of Science:

• Audiology
• Biomedical Engineering
• Human-Computer Interaction

Background:

• Hearing impairment significantly impacts the ability to follow conversations, especially in multi-talker environments.
• Steering audio beamformers using eye gaze is a promising assistive technology for individuals with hearing loss.
• In-ear electrooculography (EarEOG) can track relative eye movements, and motion sensors can track head movements.

Purpose of the Study:

• To model and analyze electrooculography in the ear (EarEOG) signals during a simulated multi-talker listening task.
• To estimate absolute eye gaze using EarEOG and head motion data in hearing-impaired individuals.
• To assess the feasibility of this eye-gaze estimation technique in both fixed and free head conditions.

Main Methods:

• Developed a model for relative eye-gaze estimation incorporating saccades, fixations, head movement, and electrode drift.
• Experiment involved 11 hearing-impaired participants focusing on visual targets in simulated multi-talker scenarios.
• Collected EarEOG and motion sensor data in two setups: fixed head and free head movement.

Main Results:

• The developed model explained 90.5% of EarEOG variance with a fixed head and 82.6% with a free head.
• Absolute eye-gaze estimation was reliable when the head was fixed.
• Hardware limitations resulted in unreliable absolute eye-gaze estimation with free head movement.

Conclusions:

• This study demonstrates the potential of combining EarEOG and motion sensors for absolute eye-gaze estimation.
• The developed model shows high accuracy in explaining EarEOG variance.
• Further refinement is needed to overcome hardware challenges for reliable real-world application in dynamic head movement scenarios.