Related Experiment Video
Updated: Aug 9, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
2.9K
AES-CSFS: an automatic evaluation system for corneal sodium fluorescein staining based on deep learning
Shaopan Wang1,2,3, Jiezhou He1, Xin He2,3,4
1Institute of Artificial Intelligence, Xiamen University, Xiamen, China.
Therapeutic Advances in Chronic Disease
|February 17, 2023
Summary
This study introduces an AI system for objective corneal fluorescein staining assessment, improving diagnostic accuracy for ocular surface diseases. The AI system demonstrated superior performance compared to junior ophthalmologists.
Area of Science:
- Ophthalmology
- Medical Imaging
- Artificial Intelligence
Background:
- Corneal fluorescein sodium staining is crucial for diagnosing ocular surface diseases.
- Current diagnostic methods rely heavily on subjective ophthalmologist interpretation, leading to variability.
Purpose of the Study:
- To develop a deep learning-based artificial intelligence (AI) system for quantitative assessment of corneal fluorescein staining.
- To accurately measure the size of corneal epithelial defects using AI.
Main Methods:
- Proposed an AI system integrating two segmentation models and one classification model for corneal image analysis.
- Evaluated the AI system's performance against manual labeling and ophthalmologist assessments.
Main Results:
- Achieved high accuracy in segmenting corneal boundaries (DSC 0.98) and epithelial defects (DSC 0.97).
- The AI system outperformed leading algorithms in segmentation and classification tasks, with 91.2% accuracy.
- AI system performance surpassed that of junior ophthalmologists in evaluating corneal staining.
Conclusions:
- The developed AI system provides a reliable automated method for corneal fluorescein staining assessment.
- This AI tool can reduce diagnostic errors stemming from subjective judgment and limited expertise.
