Date Published:
2013 Jul 18
Abstract:
PURPOSE: To develop and validate a novel automated system to assess ocular redness (OR) in clinical images. METHODS: We developed a novel software that quantifies OR in digital images based on a mathematic algorithm using a centesimal continuous scoring scale. Subsequently, we conducted a study to validate the scores obtained with this system by correlating them with those obtained by two physicians using two image-based comparative subjective scales, the Efron and the Validated Bulbar Redness (VBR) grading scales. Additionally, we evaluated the level of clinical agreement between the Ocular Redness Index (ORI) score and the two image-based methods by means of the Bland-Altman analysis. Main outcome measures included correlation and level of agreement between the ORI score, Efron score, and the VBR score. RESULTS: One hundred and two clinical photographs of eyes with OR were evaluated. The ORI scores significantly correlated with the scores obtained by the two clinicians using the Efron (Observer 1, R=0.925, P<0.001; Observer 2, R=0.857, P<0.001), and VBR (Observer 1, R=0.830, P<0.001; Observer 2, R=0.821, P<0.001) scales. The Bland-Altman analysis revealed levels of disagreement of up to 30 and 27 units for the ORI-Efron and ORI-VBR score comparisons, respectively. CONCLUSIONS: The ORI provides an objective and continuous scale for evaluating ocular injection in an automated manner, and without need for a trained physician for scoring. The ORI may be used as a new alternative for objective OR evaluation in clinics and in clinical trials.
