The invention discloses a diabetic retinopathy staging prediction method

Abstract

The invention discloses a stage prediction method for diabetic retinopathy. The method comprises the following steps: collecting fundus color images of a diabetic patient; Inputting the fundus color image into a trained simple staging network model to obtain sugar net staging result characteristics; Inputting the fundus color image into a trained focus segmentation network model to obtain sugar net focus segmentation result characteristics; Combining the sugar net staging result characteristics with the sugar net focus segmentation result characteristics to obtain segmentation staging combination characteristics; And predicting the diabetic retinopathy staging category by combining the segmentation staging combination characteristics according to the pre-judged optimal staging classifiersand the priorities of the optimal classifiers. According to the method, a plurality of classifiers are adopted for fitting of segmentation and staging combination characteristics to sugar net stagingcategories, the advantages of different classifiers are fully utilized, and a more accurate and robust staging result is obtained through combination.

Description

technical field [0001] The invention relates to a stage prediction method for diabetic retinopathy, which belongs to the technical field of image processing and analysis. Background technique [0002] The staging of diabetic retinopathy (diabetic retinopathy for short) images based on color fundus photos has important clinical significance. The current sugar reticulum staging method has limitations in the following aspects: manual-based methods usually rely on the clinical experience of doctors, and for the same fundus image, different doctors may give different staging diagnosis results, so there are some limitations to the level of doctors. Higher requirements; longer time-consuming, the unequal number of patients and doctors causes patients to spend a lot of time waiting for doctors to diagnose. At the same time, the current automatic classification method for diabetic retinopathy based on color fundus photos cannot effectively obtain and utilize various lesion informati...

AI Technical Summary

Problems solved by technology

The current sugar reticulum staging method has limitations in the following aspects: manual-based methods usually rely on the clinical experience of doctors, and for the same fundus image, different doctors may give different staging diagnosis results, so there are some limitations to the level of doctors. Higher requirements; longer time-consuming, the unequal number of patients and doctors causes patients to spend a lot of time waiting for the doctor's diagnosis

At the same time, the current automatic classification method for diabetic retinopathy based on fundus color photos cannot effectively obtain and utilize various lesion information of diabetic retinopathy, and the accuracy and robustness of the algorithm are not ideal

Images