Assessing ocular bulbar redness: a comparison of methods.

Area of Science:

• Ophthalmology
• Medical Imaging
• Biomedical Engineering

Background:

• Ocular bulbar redness is a key indicator in diagnosing and monitoring various eye conditions, including dry eye disease.
• Current methods for assessing bulbar redness include subjective grading by clinicians and automated systems, which may have limitations in precision and variability.
• Quantifying redness objectively is crucial for consistent clinical evaluation and research.

Purpose of the Study:

• To evaluate the clinical applicability of a novel image processing technique for quantifying ocular bulbar redness using relative Red-channel activity (Red-value).
• To compare the performance and variability of this Red-value method against an established automated grading system (Oculus Keratograph 5M, R-scan) and subjective clinician grading (IER scale).

Main Methods:

• Digital photographs of the nasal bulbar conjunctiva from dry eye patients (n=25) were analyzed using image processing to determine Red-values.
• Red-values were compared with subjective bulbar redness grades assigned by six clinicians using the IER scale.
• Agreement and variability between the Red-value method, clinician grades (IER), and the automated R-scan system were assessed using geometric coefficient of variation (gCoV) and Bland-Altman analyses.

Main Results:

• The Red-value method demonstrated a strong linear relationship (R² = 0.99) with the automated R-scan system and exhibited the lowest variability (gCoV = 0.97%).
• The IER scale showed a linear relationship with Red-value (R² = 0.99) but had a floor effect, failing to differentiate redness below a threshold (1.75 units or 33.0% Red-value).
• The R-scan and IER scales showed similar intra-observer variability, while the R-scan tended to report lower absolute redness scores compared to IER.

Conclusions:

• Image processing using relative Red-channel activity provides a highly reliable and least variable method for quantifying ocular bulbar redness in a clinical sample.
• The established R-scan and IER grading systems exhibit comparable intra-observer variability, but the Red-value method offers superior precision.
• The strong correlation between the R-scan and Red-value suggests that the R-scan's automated scores could potentially be derived or validated using similar image processing principles.