Eye fundus image judgment method and equipment

Abstract

The invention provides an eye fundus image judgment method and eye fundus image judgment equipment. The eye fundus image judgment method comprises the following steps: acquiring a to-be-judged eye fundus image of a user; Obtaining a reference fundus image associated with the fundus image to be judged, wherein the reference fundus image is a fundus image with a known state being normal; Generatinga difference image according to the to-be-judged fundus image and the reference fundus image; And taking the differential image as input data of a machine learning model, enabling the machine learningmodel to output a judgment result of the to-be-judged fundus image, and training the machine learning model by using the training differential image and corresponding label data.

Description

technical field [0001] The invention relates to the field of medical image processing, in particular to a fundus image judgment method and equipment. Background technique [0002] With the rapid development of image acquisition and storage technology and deep learning, deep learning has penetrated into various fields of society. In the medical field, due to the serious shortage of medical resources in my country, the detection and screening of medical images through deep learning and other cutting-edge technologies has become a very popular research field. The number of diabetic patients in China has exceeded 130 million, among which the number of diabetic retinopathy patients is about 30 million. Diabetic retinopathy has a very high probability of blindness after the progression of diabetic retinopathy. If screening, diagnosis and treatment of diabetic retinopathy are carried out at an early stage, the risk of blindness of patients can be reduced by 94.4%. [0003] The e...

AI Technical Summary

Problems solved by technology

[0004] Because there may be differences in the fundus images of each person in a normal state, the fundus images of the same disease vary greatly. However, for some less obvious disease features, such as mild sugar reticulum, small drusen, mild arteriosclerosis, etc. , the characteristics of these diseases will be suppressed due to the diversity of fundus images

This makes the existing fundus image-based detection schemes less effective in detecting some tiny feature points.

Images