Smooth pursuit detection in binocular eye-tracking data with automatic video-based performance evaluation.

Area of Science:

• Ophthalmology
• Neuroscience
• Computer Science

Background:

• Researchers increasingly use binocular eye-tracking for moving stimuli.
• Existing event-detection algorithms are often monocular and neglect smooth pursuit.
• Accurate discrimination of eye movements is crucial for data analysis.

Purpose of the Study:

• Develop a binocular eye-tracking algorithm to differentiate fixations and smooth pursuit movements.
• Evaluate the algorithm's performance using an automated video-based strategy.
• Enhance the analysis of eye movement data in research.

Main Methods:

• A clustering approach analyzing spatial and temporal aspects of binocular eye-tracking data.
• Evaluation using a novel video-based strategy with automatically detected moving objects.
• Comparison of binocular versus monocular algorithm performance.

Main Results:

• The binocular algorithm achieved 98% fixation detection on image stimuli (vs. 95% monocular).
• Both binocular and monocular algorithms detected approximately 40% of smooth pursuit movements in video stimuli.
• Binocular information proved advantageous for static stimuli without hindering dynamic movement detection.

Conclusions:

• Binocular eye-tracking data enhances fixation detection accuracy for static stimuli.
• The developed algorithm effectively discriminates fixations and smooth pursuit movements.
• Automated evaluation strategies reduce manual effort and increase data analysis capacity.