Fuzziness automatic grading method based on color eyeground image

Abstract

The invention provides a fuzziness automatic grading method based on a color eyeground image. The method extracts three kinds of features of multi-threshold blood vessel pixel number, enhancement mean value and local standard variance to describe the degree of eyeground fuzziness, and based on the feature extraction result, a decision tree is designed to realize the final grading of the eyeground image. As other lesions, such as vitreous opacity, exudate, etc. will exist in the same eyeground image at the same time, the extraction of eyeground features is seriously affected and the final diagnosis of cataract is further affected, resulting in greater errors. In view of lesions such as vitreous opacity, the invention provides a method for eliminating vitreous opacity. The algorithm is also suitable for eliminating influences of exudates and can be applied to clinical medicine about interference of non-cataract lesions on cataract diagnosis, so as to improve the correct rate of diagnosis, to eliminate excessive errors and to provide the probability for large data screening and telemedicine. The method has a certain commercial value and an application value.

Description

technical field [0001] The invention belongs to the technical field of image classification, and in particular relates to an automatic blurring classification method based on a color fundus image. Background technique [0002] For a long time, slit-lamp microscopy and lens back-illuminated photographic examination have been widely used in the screening and diagnosis of cataract, and have always been the "gold standard" for cataract assessment. "Wisconsin cataract grading system" and "lens opacity classification system" are the main clinical evaluation methods for detecting cataract with slit lamp microscope. Although the above method provides a relatively accurate and professional evaluation system, its grading standards are relatively complicated, and different types of images need to be taken and compared with standard photos. The procedure is cumbersome, the efficiency is low, and a large number of experienced professionals are required. . Therefore, color fundus images...

AI Technical Summary

Problems solved by technology

At present, the existing algorithm research has achieved preliminary results in the diagnosis of cataract based on the fundus image, but there are still two major problems that have not been resolved: the first is that the existing algorithm does not deal with other lesions that appear in the fundus image. All lesions can be displayed in the obtained fundus image at the same time, and they will affect the correct extraction of features, thereby affecting the final grading result; another problem is that there is still a large error in the grading results obtained by the existing algorithm and the manual results. , such as a diagnosis with an error greater than level 1, which will seriously affect the effect of clinical auxiliary diagnosis

Images