Noise-robust fixation detection in eye movement data: Identification by two-means clustering (I2MC)
Roy S Hessels1,2, Diederick C Niehorster3,4, Chantal Kemner5,6,7
1Department of Experimental Psychology, Helmholtz Institute, Utrecht University, Utrecht, The Netherlands. royhessels@gmail.com.
Behavior Research Methods
|November 2, 2016
Summary
A new algorithm, identification by two-means clustering (I2MC), improves eye-tracking data quality in infants and children. This method is robust to noise and data loss, offering a reliable solution for developmental research.
Area of Science:
- Developmental Psychology
- Cognitive Science
- Neuroscience
Background:
- Eye-tracking research in infants and children is growing but faces challenges with data quality due to noise and data loss.
- Existing fixation detection algorithms are not optimized for the unique data characteristics of young participants.
- Manual correction of fixation data is time-consuming and subjective, highlighting the need for automated solutions.
Purpose of the Study:
- To introduce and evaluate a novel fixation detection algorithm, identification by two-means clustering (I2MC), specifically designed for noisy eye-tracking data from infants and children.
- To assess the robustness of I2MC compared to existing state-of-the-art algorithms under varying noise and data loss conditions.
- To provide an automated, offline solution for improving the reliability of eye-tracking data analysis in developmental studies.
Main Methods:
- Development of the identification by two-means clustering (I2MC) algorithm, engineered to handle high noise levels and data loss.
- Comparative evaluation of I2MC against seven other leading event detection algorithms.
- Testing the algorithm's performance on eye-tracking data from infants, children, and potentially patient groups using static stimuli.
Main Results:
- The I2MC algorithm demonstrated superior robustness to high noise and data loss compared to seven other state-of-the-art algorithms.
- I2MC provides automatic, offline processing suitable for remote and tower-mounted eye-trackers.
- The algorithm's performance is consistent across various noise and data loss levels, making it reliable for diverse research scenarios.
Conclusions:
- The I2MC algorithm offers a significant advancement for eye-tracking research in infants and children by enhancing data quality and reliability.
- Its robustness to noise and data loss makes it a valuable tool for developmental, clinical, and longitudinal studies.
- I2MC facilitates more accurate and efficient analysis of eye movement data in populations with challenging data quality.
