Multi-model sugar network lesion automatic screening method based on convolutional neural network

Abstract

The invention relates to a multi-model sugar network lesion automatic screening method based on a convolutional neural network. The method comprises the following steps: obtaining a fundus image; screening out a normal eye fundus image in the eye fundus image by using a sugar net image quality inspection CNN classifier; obtaining the sugar net national standard grade of the lesion to which the sugar net image belongs in the normal fundus image through a sugar net grade classifier module; obtaining lesion positions and category information on the normal fundus images through a sugar net lesionarea detection module; fusing the sugar net national standard grade, the lesion position and the category information of the lesion to which the sugar net image belongs through a sugar net early screening grade classification fusion module to obtain a screening grade; and after the lesion position and the category information pass through a convolutional neural network category activation mappingmodule, obtaining the area of a bleeding position. According to the invention, early screening of the sugar network images can be accurately and automatically carried out, the early screening efficiency is improved, the workload of diagnosis personnel is reduced, and in addition, the defect of low error-tolerant rate of a single model is overcome.

Description

technical field [0001] The invention relates to a screening method, in particular to a convolutional neural network-based automatic screening method for multiple models of sugar net lesions. Background technique [0002] Diabetic retinopathy (hereinafter referred to as diabetic retinopathy) has become an important problem in ophthalmic diseases. It is one of the most common complications of diabetes and can lead to permanent blindness in severe cases. Early screening of diabetic reticulum plays a vital role in the control and treatment of diabetic reticulum lesions. The more mature screening method in the field of clinical medicine is to check the color digital fundus scan images. [0003] However, due to the time-consuming and labor-intensive manual screening process, which relies heavily on the professional ability of diagnosticians, it cannot meet the increasing demand of patients year by year, resulting in many patients not being diagnosed in time and delaying the best t...

AI Technical Summary

Problems solved by technology

[0006] However, in practical applications, the single neural network model has some defects in sugar net screening. First, the single neural network model cannot accurately eliminate a large number of low-quality drugs caused by objective factors such as medical equipment, doctor experience, and patient cooperation. fundus photos; on the other hand, a single neural network model that only has a classification effect cannot reveal the details of diabetic reticulum lesions for further analysis by doctors

Images