Development And Clinical Application Of Corneal Mechanical Parameter Analysis System Based On Scheimpflug Camera

China is the country with the highest incidence of myopia in adolescents.The rate of refractive surgery in China is also very high.Refractive surgery performs quantitative thickness cutting of the cornea,destroying the integrity of the cornea,affecting postoperative intraocular pressure and corneal biomechanical properties,and may increase the risk of complications after refractive surgery^[1].Therefore,it is important to accurately examine the biomechanical properties of the cornea before and after refractive surgery.The Visual Corneal Biomechanical Analyzer（Corvis ST）is the best instrument for clinical commercial use of corneal biomechanical testing.The drawback is that the image resolution in the direction of pressing is low,and there is random highlight white spot noise,which all bring difficulties to the precise positioning of the corneal boundary.In terms of processing methods,the existing corneal image boundary localization algorithm for high-speed Scheimpflug cameras adopts a method directly related to image gray scale,which has weak anti-interference ability and weak robustness.At the same time,Corvis ST can obtain fewer parameters and cannot fully meet the needs of the rapid development of clinical and scientific research in related fields.In view of this,this paper first proposes a corneal boundary location algorithm based on U-Net.In the data preprocessing process,the simulated highlights are added to the training session to improve the network anti-interference ability;offline training+online fine-tuning mode is adopted.The displacement of the sinusoidal wave is introduced by the segmental affine transformation,and the subsequent fluctuation of the cornea under the pressure disturbance is simulated.In the post-processing,the dynamic deformation of the cornea is further fitted as a post-processing link by the sixth-order polynomial fitting.After obtaining high-precision corneal boundary positioning,we calculated more accurate and more corneal biomechanical parameters,and designed a Python-Flask-based online analysis system to help clinicians achieve large-scale data on corneal biomechanical properties.Automatic analysis.Finally,based on this software system,we applied the change of corneal biomechanical parameters based on CCT before and after refractive surgery,and found that CCT is linearly related to the amplitude of intraocular pressure change.In summary,the research on corneal biomechanical parameters is divided into three aspects:1)U-Net based corneal boundary accurate localization algorithm;2)Python-Flask based corneal biomechanical parameters online analysis software system.3)The changes of corneal biomechanical parameters before and after refractive surgery were explored.This work is expected to provide a powerful quantitative analysis tool for refractive surgery correction and corneal biomechanics research.