Detecção automática da camada epitelial da córnea a partir de imagens de Scheimpflug

Abstract

Introduction: The cornea is responsible for receiving light rays and is the most external part of the eye. It is histologically composed of 5 layers, which from the outermost to the inner part are arranged: epithelium, Bowman’s membrane, stroma, Descemet’s membrane, and the endothelium. The epithelium is the most exposed layer of the cornea. Some diseases affect the cornea causing alterations in the corneal layers. The investigation of these diseases is of great relevance to clinical practice and allows the detection, often early, of these diseases. There is also equipment that helps in the diagnosis of these diseases, providing important information. Using techniques to extract new information regarding the epithelium for these devices further contributes to a more accurate diagnosis. The epithelium is the first layer that is impacted by diseases such as keratoconus. Objective: Identify the epithelium in an automated way in the images obtained by the Scheimpflug camera. Methodology: The proposed methodology analyzes 279 exams of normal corneas obtained through the captures by the Scheimpflug camera, applying the classic methods of detection of edges existing in the literature to identify, isolate, validate and analyze information from the epithelium. Results: The Canny, Zerocross, and Log algorithms were able to detect the epithelium in the total measure, the lowest averages found in both thicknesses were with the log and zerocross methods with their variations, which had 79.74 µm, 79.85 µm and 80.38 µm in thickness and 65.91 µm, 66.08 µm and 67.25 µm in thickness by Euclidean distance. However, zerocross had the lowest number of defective images, and log had more than 50% of the base images with problems. In the central measure, the lowest average found in both thicknesses was also with the log and zerocross methods with their variations, which had 75.50 µm, 75.58 µm and 75.75 µm in the thickness and 61.61 µm, 61.70 µm and 62.04 µm in the thickness by the Euclidean distance. Conclusion:We were able to perform the identification of the epithelium in an automated way with the images of the Scheimpflug camera with the edge detection methods Canny, Log, and Zerocross and we were able to have better use of the images using Zerocross with Threshold: 0.0003.