Extraction of Eye Redness for Standardized Ocular Surface Photography

Color photography is the basis to evaluate ocular surface imaging biomarkers such as eye redness. Different grading scales are hereby used clinically to examine the severity of eye redness ranging from a white to a red eye. Currently used imaging and grading is time consuming and subjective. In this work we propose a baseline pipeline to assess the ocular redness based on standardized images of the ocular surface. Images were acquired using a novel ocular surface photography system, specifically tailored for standardized imaging in terms of lighting, focus and position. The pipeline comprises three major steps in extracting the eye redness: (i) defining a region-of-interest in the image of the ocular surface, (ii) detection of scleral tissue by tiling the high-resolution images and subsequent classification of the tiles and (iii) quantification of ocular redness based on image features. The pipeline was evaluated on a data set containing external eye images of healthy subjects and showed promising results on the detection of scleral tiles, which can subsequently be used for eye redness extraction. The performance and the simplicity of the approach makes the baseline pipeline a suitable candidate for further development and translating the concept to clinical patient data.