Full-automatic retinal blood vessel analysis method and system suitable for intelligent portable equipment

Abstract

The invention provides a full-automatic retinal blood vessel analysis method and system suitable for intelligent portable equipment. The invention improves a saliency feature method by adding a preprocessing step which comprises performing homogenization on a fundus image and removing the central light reflection treatment of the blood vessel, thereby realizing rapid segmentation of the retinal blood vessel. The method has the advantages of good stability, simple operation and high speed. The invention realizes accurate analysis of various retinal blood vessels and realizes the combination with intelligent portable equipment by optimizing algorithm complexity and operation speed, so computer-aided diagnosis can provide systemic blood vessel analysis with strong applicability, stable calculation and high precision in retinal image analysis.

Description

technical field [0001] The invention belongs to the field of computer-aided diagnosis, and in particular relates to a fully automatic retinal blood vessel analysis method and system suitable for intelligent portable devices. Background technique [0002] There are more than 400 million diabetic patients in the world. Organized, regular and large-scale fundus image screening for diabetic patients, even high-risk groups of diabetes, is a method commonly used in European and American countries. The "Guidelines for the Prevention and Treatment of Type 2 Diabetes in China" also pointed out that once a diabetic patient is diagnosed, he or she should undergo a fundus examination at least once a year; once a diabetic patient develops symptoms of severe retinopathy, a fundus reexamination should be performed every 2 to 4 months . At present, doctors rely on doctors to screen color images one by one. This process has a large workload, low efficiency, and high subjectivity. Therefore,...

AI Technical Summary

Problems solved by technology

However, due to the influence of shooting conditions, individual differences and lesions, fundus images have problems such as uneven exposure, susceptibility to noise interference, and large differences between images, and the accuracy of the method is low.

In addition, retinal blood vessel analysis includes multiple tasks such as blood vessel segmentation, measurement, and distinction between arteries and veins. However, the existing technology can only achieve a single content and cannot achieve systematic analysis, which is very limited in application.

On the other hand, in order to facilitate the use in remote areas, grass-roots hospitals, families, etc., an easy-to-obtain platform is needed. Most of the existing methods use large-scale computers or workstations as platforms, which have high requirements for computing hardware, and lack friendly, user-friendly interface

[0004] The traditional graph theory-based blood vessel segmentation method uses a three-dimensional graph theory model, which requires high calculations and is difficult to transplant to portable devices, which greatly limits the scope of use of its method

Images