Small eye movements cannot be reliably measured by video-based P-CR eye-trackers.

Area of Science:

• Ophthalmology
• Neuroscience
• Biomedical Engineering

Background:

• The 'resolution' metric is proposed for evaluating eye-tracker suitability for microsaccade measurement, potentially surpassing the established root-mean-square of sample-to-sample distances (RMS-S2S).
• Existing research presents ambiguities regarding the resolution measure, including its calculation requirements and relationship with other precision metrics like RMS-S2S and standard deviation (STD).
• A separate metric with the same name quantifies amplitude measurement errors, adding to the confusion surrounding 'resolution' in eye-tracking.

Purpose of the Study:

• To introduce the Stepperbox, a novel mechanism for precisely rotating artificial eyes in small, controlled increments (from 1 arcmin upwards).
• To investigate the relationship between the proposed 'resolution' metric and established precision measures (RMS-S2S, STD) for eye-trackers.
• To assess the accuracy of various eye-trackers, particularly video-based (VOGs) and Dual Purkinje Imaging (DPI), in measuring small eye rotations and amplitudes.

Main Methods:

• Development and utilization of the Stepperbox to generate controlled artificial eye rotations starting from 1 arcmin.
• Testing 11 video-based eye-trackers (VOGs) and one Dual Purkinje Imaging (DPI) tracker using the Stepperbox.
• Calculating the 'resolution' metric and comparing it with RMS-S2S and STD precision measures for each tested eye-tracker.

Main Results:

• The 'resolution' metric demonstrated a significant correlation with RMS-S2S and a moderate correlation with STD across tested eye-trackers.
• Most video-based eye-trackers exhibited errors in measuring rotation amplitudes up to 2 degrees, with corneal reflection (CR) identified as a likely cause.
• The Dual Purkinje Imaging (DPI) tracker accurately measured rotation amplitudes down to 1 arcmin, outperforming video-based systems.

Conclusions:

• The Stepperbox provides a reliable method for evaluating eye-tracker performance in measuring small eye rotations.
• Video-based eye-trackers, particularly those relying on corneal reflection, may introduce significant errors in measuring small eye movements, potentially affecting previous research findings.
• The DPI tracker shows superior accuracy for measuring small eye rotations, suggesting its suitability for research requiring high precision in microsaccade and vergence measurements.