Automatic retinal vessel segmentation method for glaucoma

Abstract

The present invention proposes an automatic retinal vessel segmentation method for clinical diagnosis of glaucoma. By fusing five models of matching filter, neural network, multi-scale line detection, scale-space analysis and morphology that depend on different image processing techniques, the optic disc is eliminated. and the effects of bright areas such as exudates. At the same time, the present invention does not require a large amount of data to establish a retinal vessel segmentation model, which greatly reduces the amount and complexity of data to be processed, is easy to implement, and can effectively improve the efficiency of retinal vessel segmentation. The present invention also uses the region growing method and gradient information to iteratively grow the background and blood vessel regions on the basis of the multimodal fusion results. The segmentation results have better continuity and smoothness, and can retain more details of retinal blood vessels and more Complete visual retinal vascular network, thereby effectively assisting ophthalmologists in diagnosing diseases and reducing the burden on ophthalmologists.

Description

technical field [0001] The invention relates to the fields of digital medical image processing and analysis and intelligent medical care, in particular to an automatic retinal blood vessel segmentation method for glaucoma clinical diagnosis. Background technique [0002] Glaucoma is the world's leading irreversible blinding eye disease, known as the "silent thief of sight". It is estimated that by 2020, the number of glaucoma patients worldwide will rise to 79.6 million. At present, the prevalence of open-angle glaucoma in people over 40 years old in my country has reached 2.6%, accounting for 2 / 3 of the total number of glaucoma patients, and the blindness rate is 15% to 30%, which is much higher than the average level of 8% in economically developed countries. Early detection, early diagnosis, and early treatment are very important to inhibit the development of glaucoma. Color fundus images can directly observe retinal vascular lesions and other lesions such as exudates a...

AI Technical Summary

Problems solved by technology

Compared with the traditional method, although this method can obtain better segmentation accuracy and robustness; however, deep learning is a data-driven model that requires massive levels of data for guarantee, which will seriously affect the accuracy of retinal vessel segmentation. efficiency, which limits its application in clinical practice

Images