Anisotropy-Aware Cellular Automaton Framework for AI-Powered Pterygium Screening.

Area of Science:

• Ophthalmology
• Artificial Intelligence
• Medical Imaging

Background:

• Pterygium diagnosis often relies on expert annotation, limiting accessibility in primary care.
• Current AI diagnostic methods for pterygium require extensive, professionally annotated datasets.

Purpose of the Study:

• To develop a weakly-labeled, positive-sample-free AI method for pterygium auxiliary diagnosis.
• To create an effective screening tool for primary healthcare institutions and self-examination.

Main Methods:

• Constructed an ocular surface anatomic atlas using semantic segmentation.
• Employed an improved cellular automaton method to simulate pterygium growth patterns.
• Generated high-quality training samples without manual annotation.

Main Results:

• The AI model achieved accurate identification of pterygium lesions.
• Model performance surpassed conventional training approaches.
• Overcame the dependency on professionally annotated data.

Conclusions:

• Presented a pathomimetic computational framework for automatic generation of labeled pterygium cases.
• Facilitated an auxiliary-diagnostic AI model for early pterygium screening.
• Demonstrated suitability for resource-limited primary healthcare settings to reduce vision impairment risk.