EyeLoop: An Open-Source System for High-Speed, Closed-Loop Eye-Tracking.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Computer Vision

Background:

• Eye-tracking is crucial for studying nervous system dynamics and neuropathology.
• Current eye-tracking systems are often expensive, proprietary, and lack real-time analysis capabilities.
• Limitations hinder integration with neural activity monitoring and closed-loop experimental designs.

Purpose of the Study:

• To develop an open-source, cost-effective, and high-performance eye-tracking solution.
• To enable real-time analysis and facilitate closed-loop experimental designs.
• To make advanced eye-tracking technology accessible to a wider range of research facilities.

Main Methods:

• Developed EyeLoop, a Python-based open-source eye-tracker.
• Utilized a highly efficient vectorized pupil detection method for uninterrupted tracking.
• Integrated custom functions via code modules for flexible application.
• Achieved high accuracy comparable to established eye-tracking modules like DeepLabCut.

Main Results:

• EyeLoop provides uninterrupted, high-accuracy eye tracking at over 1,000 frames per second on consumer hardware.
• Demonstrated utility in both open-loop experiments and biomedical disease identification.
• Successfully tracked eyes across multiple species, including rodents, humans, and non-human primates.
• Achieved performance on par with commercial solutions like DeepLabCut.

Conclusions:

• EyeLoop offers a low-cost, accessible alternative for high-speed eye-tracking.
• The open-source nature and ease of integration lower barriers for researchers.
• Facilitates advanced research in neuroscience and biomedical applications by enabling real-time analysis and integration with neural recordings.