Analysis And Diagnosis Of Dry Eye Based On Corneal Topography System

As a common ocular surface disease,dry eye disease(DED)is ubiquitous worldwide,which is mainly characterized in the loss of tear film homeostasis,and may be accompanied by various degrees of symptoms such as visual fatigue and corneal inflammation.Today,there are several dry eye diagnostic techniques.Among them,computer-aided corneal topography systems are widely used in the screening and early diagnosis of dry eye disease because of their advantages,such as high precision,low cost,simple structure,and low sensitivity to noise.The focus of this study is the assessment of tear film stability in the core mechanism of dry eye.Depending on the corneal topography system,the image processing algorithm was applied for feature extraction,parameter calculation and dynamic measurement of tear film stability.Finally,Tear film quality evaluation was achieved and an objective basis was provided for the clinical diagnosis of dry eye.The research contents mainly include the following three aspects:1.Tear film image preprocessing,including blur detection and location of the ring pattern area.Aiming at the inherent characteristics of tear film images,a blur detection method was proposed based on Sobel edge operator,which was used to remove the blur images that hinder feature extraction.The target image,with weakened background and eyelashes and prominent ring pattern,was obtained by the morphological open operation performed on the grayscale image.In order to further reduce the influence of the background and eyelashes on the post-image processing,a band-pass filter was applied to the target image according to the ring pattern frequency.At last,the ring pattern area was determined by local Otsu method binarization and morphological close operation.2.Ring pattern deformation detection.The relationship between the local structure change of the ring pattern and the quality of the tear film was analyzed.According to the texture features of the ring pattern,a deformation detection method(GLCM)based on gray-level co-occurrence matrix was proposed for local tear film rapture judgment.Common texture structure analysis methods were compared and verified in the ring pattern deformation detection.3.Dry eye analysis and diagnosis.The gray image circular pattern region was transformed from the cartesian coordinate system to the polar coordinate system,and divided into several non-overlapping sub-blocks.Block feature was extracted by the GLCM method.Through the block feature,the tear film quality index of the whole image was obtained.The characteristic parameters were fitted to determine the tear film breakup area and tear film breakup time.The experiment dealt with 54 tear film video samples.Diagnostic results were close to those of ophthalmologists,providing an objective basis for the analysis and diagnosis of dry eye.