Related Experiment Video
Updated: Jul 16, 2026
Quantification of Vascular Parameters in Whole Mount Retinas of Mice with Non-Proliferative and Proliferative Retinopathies
Published on: March 12, 2022
Conjunctival Vascular Metrics Using Automated Vessel Detection from Slit Lamp Images for Hyperemia Severity
Damon Wong1,2,3, Yvonne Ng2, Leila Sara Eppenberger1
1Singapore Eye Research Institute, Singapore National Eye Centre, Singapore 168751, Singapore.
Insights
Automated deep learning analysis of conjunctival vessels shows promise for objective conjunctival hyperemia (redness) assessment. Vessel density and fractal dimension correlate well with manual grading, offering potential for more consistent clinical evaluation.
Area of Science:
- Ophthalmology
- Medical Imaging
- Artificial Intelligence
Background:
- Conjunctival hyperemia (redness) is common but subjectively graded, leading to inconsistencies.
- Objective assessment methods are needed to improve clinical evaluation and diagnosis.
- Deep learning offers potential for automated, quantitative analysis of ocular vascularity.
Purpose of the Study:
- To evaluate vascular metrics derived from automated deep learning vessel detection.
- To compare these automated metrics with manual grading of conjunctival hyperemia severity.
- To assess the potential for objective measurement of conjunctival hyperemia.
Main Methods:
- Slit lamp images from 139 glaucoma patients were analyzed.
- Manual grading of conjunctival hyperemia was performed using the Efron Grading Scheme by two independent graders.
- An automated deep learning pipeline detected conjunctival vessels, calculating vessel density, fractal dimension, and tortuosity.
- Vascular metrics were compared against manual Efron grades.
Main Results:
- Manual grading showed good inter-grader consistency but significant differences in moderate grades.
- Automated vessel density strongly correlated with manual Efron grades (Spearman's rho = 0.78, p < 0.001).
- Fractal dimension also showed significant association with manual grading (Spearman's rho = 0.55, p < 0.001).
- Vessel tortuosity showed poor agreement with manual grades.
Conclusions:
- Automated vessel density and fractal dimension derived from deep learning show significant correlation with manual conjunctival hyperemia grading.
- These quantitative vascular metrics offer a potential pathway towards more objective and reproducible assessment of conjunctival hyperemia severity.
- Deep learning-based analysis may improve clinical evaluation and diagnostic consistency for ocular surface conditions.
Abstract:
Background/Objectives: Conjunctival hyperemia is a common clinical finding in clinical practice; however there are significant differences between graders. Vessel detection using deep-learning approaches could enable more objective measures. We aimed to evaluate vascular metrics derived from automated vessel detection and compare these metrics with manual severity gradings. Methods: Slit lamp images from 139 glaucoma patients were included. Images from 103 participants were used as the primary development dataset and the remaining as a validation subset. The images were independently graded by two graders for conjunctival hyperemia using the Efron Grading Scheme. Conjunctival vessels were detected using an automated vessel detection pipeline based on semi-supervised learning. Vessel density, fractal dimension and tortuosity were calculated and compared with the manual Efron grades. Results: Grading of conjunctival hyperemia between the two graders were consistent (Spearman's rho: 0.79; ICC: 0.79 [95%CI: 0.72-0.84]) but showed significant differences with a higher proportion of differences in the moderate grades. Of the vascular metrics, vessel density showed significant associations with the individual Efron grading and against the mean Efron grading (0.78, p < 0.001). Fractal dimension was significantly associated with the mean Efron grading (0.55, p < 0.001). Agreements were similar in the subset (vessel density, 0.80, p < 0.001; fractal dimension 0.62, p < 0.001). Vessel tortuosity showed lower agreements (<0.23). Conclusions: Vessel density and fractal dimension showed significant associations with manual Efron gradings. These metrics could be potentially used to enable more objective and interpretable measures of conjunctival hyperemia severity.
