To establish a comprehensive framework for the classification, annotation, and quality control of dry eye imaging datasets for AI applications.
Approach:
Expert Consensus: An international group of ophthalmologists, imaging specialists, and AI researchers developed guidelines for creating high-quality datasets to support AI tools for dry eye diagnosis and management.
Key Findings:
Dry eye disease is a common ocular surface disorder with rising prevalence.
Lack of consistent standards for image annotation and classification hinders AI development.
High-quality data annotation is essential for robust AI model development.
The consensus outlines recommendations for five major imaging modalities used in dry eye assessment.
Quality assurance measures are critical for ensuring reliable AI outcomes.
Interpretation:
The consensus highlights the need for standardized datasets and quality assurance in AI applications for dry eye.
Limitations:
Variable image quality across institutions.
Absence of universally accepted annotation standards.
Limited algorithm generalizability due to single-center training datasets.
Barriers to multi-center data sharing.
Conclusion:
The consensus calls for the creation of large, diverse, multi-center image repositories and the adoption of standard operating procedures for image acquisition and annotation.
