Fundus retinal vessel segmentation system based on enhanced generative adversarial network

Abstract

The invention provides a fundus retina blood vessel segmentation system based on an enhanced generative adversarial network, belongs to the technical field of medical image processing, and aims to solve the problems of algorithm limitation, low actual imaging contrast, blood vessel breakage of a segmentation result and redundancy of blood vessel branch details in existing fundus retina blood vessel segmentation. The system includes: a retina image preprocessing unit which is used for carrying out enhancement processing on manually segmented sample labels and retina image comparison adaptive histogram equalization; a data enhancement unit which is used for carrying out enhancement processing on the image again by increasing the trained data volume; and an adversarial generative network model which is used for segmenting the retina data image processed by the data enhancement unit so as to extract more fine blood vessels and improve the accuracy of the segmentation effect. The system hasthe effects that the network training is greatly accelerated, fine blood vessel branch details can be segmented, the noise of the retinal image is effectively removed, and the blood vessel segmentation accuracy is improved.

Description

technical field [0001] The present invention relates to the technical field of medical image processing, in particular, to a fundus retinal blood vessel segmentation system based on enhanced confrontation generation network. Background technique [0002] Retinal microvessels are the only deep microvessels in the human body's vascular system that can be directly observed without trauma. Many diseases such as hypertension and diabetes will cause changes in the state or structure of retinal blood vessels. Information such as the number, branch, and width of retinal blood vessels can be used as the basis for diagnosis of retinal blood vessel-related diseases. Through the analysis of retinal blood vessels, it is of great clinical significance to make early diagnosis and prevention of some diseases. Therefore, retinal vessel segmentation technology is particularly important. [0003] Due to the uneven gray distribution of retinal images, complex vascular structure, low contrast ...

AI Technical Summary

Problems solved by technology

[0004] The above method can extract most of the retinal blood vessels. Due to the limitation of the algorithm or the low contrast of the actual imaging, the segmentation results of the algorithm often show the phenomenon of blood vessel breakage and the details of blood vessel branches are redundant.

Images