Analysis of the Effect of Unified Diagnosis Framework for Automated Nuclear Cataract Grading Algorithm

Abstract

In this paper, we discuss the effectiveness and acceptability of unified diagnosis framework for automated nuclear cataract grading algorithm (UDFA) based on the mobile phone slit lamp. We obtain the effectiveness analysis data from June 2020 to December 2020 from Shenyang Eyerobo Company, three senior ophthalmologist classify 13816 eye images into three-way classification. The unified diagnosis framework for automated nuclear cataract grading algorithm was used to re-predict the results of eye images collected from September to December 2020 and the eye images retaken from the same eye under the same conditions to evaluate the consistency of the retest results of the algorithm. The sensitivity and specificity of the cataract deep learning algorithm were collected from 206 patients randomly selected from ophthalmic hospitals by 2 skilled operators using the mobile phone slit lamp to collect eye images. And we tested the algorithm in the north of Shenyang in Liaoning province for three months, to extract the rural doctors semi-structured interviews and its customers, to discuss their feelings after using. The mobile phone slit lamp is highly effective in screening cataract, and this new screening model has been accepted and recognized by users.