Long-term stabilized iris tracking with unsupervised constraints on dynamic AS-OCT.

Area of Science:

• Ophthalmology
• Medical Imaging
• Computer Vision

Background:

• Primary angle-closure glaucoma (PACG) causes significant global blindness.
• PACG is often asymptomatic until irreversible vision loss occurs.
• Iris rigidity and pupillary reflex velocity are altered in severe PACG.

Purpose of the Study:

• To develop an automated method for quantifying dynamic iris motion from Anterior Segment Optical Coherence Tomography (AS-OCT) videos.
• To address challenges in AS-OCT video tracking, including image jitter, irregular motion, and limited datasets.
• To enable clinical assessment of iris dynamics for glaucoma diagnosis.

Main Methods:

• Proposed an unsupervised constraint-based jitter refinement tracking (CJRTrack) framework for AS-OCT video analysis.
• Utilized a point tracking algorithm for key region extraction in low-resolution images.
• Implemented a multi-frame differentiable registration network and a temporal topology constraint module for accurate, stabilized tracking.

Main Results:

• CJRTrack demonstrated superior accuracy and stability compared to existing tracking models on independent AS-OCT datasets.
• The framework successfully performed jitter-corrected quantification of iris motion.
• The quantified data was effectively used to classify neuropathic damage in primary angle closure patients.

Conclusions:

• CJRTrack provides a robust solution for automated dynamic iris motion quantification from AS-OCT videos.
• The developed method overcomes key technical challenges in analyzing high-resolution, dynamic ocular imaging.
• This approach shows significant clinical potential for diagnosing and managing primary angle-closure glaucoma.