Markerless monocular tracking system for guided external eye surgery

Area of Science:

• Ophthalmology
• Computer Vision
• Medical Imaging

Background:

• Accurate eye tracking is crucial for guiding ophthalmologists during external eye surgery.
• Traditional tracking methods often fail due to occlusion of the pupil and cornea.
• A novel approach is needed to overcome these limitations in real-time surgical scenarios.

Purpose of the Study:

• To develop and evaluate a markerless monocular tracking system for external eye surgery.
• To enable accurate eye tracking using only scleral textures, independent of pupil or corneal visibility.
• To assess the system's performance using established invariant point detection algorithms.

Main Methods:

• Implemented a markerless monocular tracking system using Scalable Invariant Feature Transforms (SIFT) and Speed Up Robust Features (SURF).
• Focused on detecting invariant points and correspondences solely from scleral textures.
• Tested the system on phantom eyes with 360° rotation and real eye images.

Main Results:

• The system demonstrated high accuracy, with tracking errors below 0.5% for spherical objects at 360° rotation.
• Both SIFT and SURF algorithms yielded comparable, excellent tracking performance.
• Promising results were observed on real eye images, even with occlusions from blood or surgical instruments.

Conclusions:

• The developed markerless tracking system offers a robust solution for ophthalmic surgery guidance.
• Utilizing scleral textures overcomes limitations of traditional methods, improving tracking reliability.
• The system shows significant potential for enhancing precision and safety in external eye surgeries.