V-133 | Improvements for Online Eye-Tracking in Neuroscience Research: Implementing Blink Detection, Head Movement Tracking, and Continuous Distance Estimation

Tools Development and Open Source Neuroscience
Author: Agustin Penas | Email: agustinpenas@gmail.com

Agustin Penas^1°2°3°, Gustavo Juantorena^1°2°3°,Juan Kamienkowski^1°2°3°

^1° Instituto de Investigación en Ciencias de la Computación (ICC)
^2° FCEyN
^3° UBA-CONICET

The use of online eye-tracking tools like WebGazer.js has the potential of revolutionizing data collection in neuroscience by enabling remote experiments. However, the accuracy and versatility of these tools can be limited by factors such as participant head movement and inconsistent distance to the screen. To address these challenges, we have implemented three significant enhancements: blink detection, head movement tracking, and an innovative method for estimating the subject’s distance from the screen.
The addition of blink detection improves the precision of eye-tracking data by identifying and filtering out blinks, which can introduce noise both in the calibration/validation phase or in the experiment itself. The head movement tracking is designed to detect significant shifts in the participant’s position, identifying instances where calibration may have been compromised. This allows researchers to maintain the integrity of the data by determining when recalibration is necessary. The new distance estimation feature offers a practical alternative to the traditional ‘virtual chin rest’ method. While not necessarily superior, our approach has the advantage of not demanding interaction from the participant and continuously estimating the distance, rather than performing a single 2 minute measurement. This ensures ongoing accuracy throughout the experiment.
Altogether, this is a step forward into massive eye-movement experiments and remote clinical assessments.