A Survey on AI-Based Cornea Detection from Anterior Segment Images

AI and computer vision possess substantial potential to support and streamline corneal image analysis[cite: 3]. The optimal approach depends strictly on the clinical task: region localization, boundary segmentation, structural measurement, or disease screening[cite: 3]. While traditional methods offer simplicity, modern deep learning architectures provide superior pattern recognition capabilities at the cost of requiring massive labeled datasets and substantial compute resources[cite: 3]. Future research must prioritize multi-device validation, adaptive resource scheduling for edge deployments, uncertainty estimation, and the integration of explainable AI frameworks to facilitate real-world clinical adoption[cite: 3].

[1]  Hezekiah, J. D. K., & Chacko, S. (2020). “A Review on Cornea Imaging and Processing Techniques.” Current Medical Imaging, 16(3), 181-192. [2]  Tey, K. Y., Cheong, E. Z. K., & Ang, M. (2024). “Potential applications of artificial intelligence in image analysis in cornea diseases: a review.” Eye and Vision, 11, 10. [3]  Tahvildari, M., Singh, R. B., & Saeed, H. N. (2021). “Application of Artificial Intelligence in the Diagnosis and Management of Corneal Diseases.” Seminars in Ophthalmology, 36(8), 641-648. [4]  Nguyen, T., et al. (2024). “Artificial Intelligence in Corneal Diseases: A Narrative Review.” Contact Lens and Anterior Eye, 47(6), 102284. [5]  Bodmer, N. S., et al. (2024). “Deep Learning Models Used in the Diagnostic Workup of Keratoconus: A Systematic Review and Exploratory Meta-Analysis.” Cornea, 43(7), 916-931. [6]  Ronneberger, O., Fischer, P., & Brox, T. (2015). “U-Net: Convolutional Networks for Biomedical Image Segmentation.” MICCAI, 234-241.