Diabetic retinopathy classification method and device based on a graph network

Abstract

The invention belongs to the technical field of medical image processing, and discloses a diabetic retinopathy classification method and device based on a graph network. The diabetic retinopathy classification method based on the graph network comprises the steps of acquiring a preset number of fundus color images and subjecting the fundus color images to normalization preprocessing to serve as normalization images; constructing a graph structure by using the normalized image, and carrying out feature extraction on the graph structure to obtain to-be-processed data; and constructing a convolutional neural network model, and carrying out convolutional neural network training on the to-be-processed data to obtain a diabetic retinopathy grade classification model. Graphic data are structured,and classification of fundus color images is realized through a graph structure analysis algorithm in combination with a convolutional neural network model and through automatic film reading of a computer.

Description

technical field [0001] The invention belongs to the technical field of medical image processing, and in particular relates to a graph network-based diabetic retinopathy classification method and device. Background technique [0002] Diabetic retinopathy is a fundus lesion with specific changes and is one of the most common and serious complications of diabetes. Regular fundus screening for diabetic patients and early treatment after discovery can greatly reduce the incidence of blindness. Regular fundus screening for diabetic patients can prevent the occurrence of vision loss caused by diabetes and greatly reduce the incidence of blindness. However, there are several problems in fundus screening at present: 1. In clinical practice, doctors need to analyze and process a large amount of data, and the judgment of manual film reading depends on the doctor's personal experience, which is time-consuming and labor-intensive; Third, although the traditional deep learning method fo...

AI Technical Summary

Problems solved by technology

However, there are several problems in fundus screening at present: 1. In clinical practice, doctors need to analyze and process a large amount of data, and the judgment of manual film reading depends on the doctor's personal experience, which is time-consuming and labor-intensive; Third, although the traditional deep learning method for discrimination and classification has achieved relatively good results, its own "black box" characteristics, and the algorithm principle is not clear, and it does not have interpretability, which greatly hinders the actual clinical application. application on

Images