Blood vessel segmentation method, device and apparatus for retinal image

Abstract

The invention discloses a blood vessel segmentation method, device and apparatus for a retina image. The method comprises the steps of obtaining a target retina image, inputting the target retina image into a pre-trained full convolutional network model, determining a blood vessel segmentation image of the target retina image based on an output result of the full convolutional network model; wherein a decoding network of the full convolutional network model comprises a densely connected convolutional network, and training the full convolutional network model in advance based on a historical retina image and a known blood vessel segmentation image of the historical retina image. The dense connection convolutional network in the full convolutional network model can ensure the context relationship of the image features between different layers. The problem of gradient disappearance or gradient explosion in the full convolutional network model is effectively relieved, so that the featureloss of image features in the transmission process can be effectively reduced, and the accuracy of blood vessel segmentation for the retinal image is improved.

Description

technical field [0001] The present application relates to the technical field of image segmentation, in particular to a blood vessel segmentation method, device and equipment for retinal images. Background technique [0002] In recent years, with the rapid development of artificial intelligence, computer-aided diagnosis technology has gradually developed to a certain extent. Among them, computer-aided diagnosis technology refers to the use of medical image processing and other technologies to assist imaging doctors to quickly and accurately find lesions and improve the efficiency of diagnosis. When computer-aided diagnosis technology is used to detect retinal vascular lesions, it is usually necessary to segment blood vessels in retinal images. [0003] At present, when segmenting blood vessels in retinal images, image features in retinal images are usually automatically extracted based on deep learning methods, and then blood vessels in retinal images are segmented based on...

AI Technical Summary

Problems solved by technology

However, based on the existing deep learning methods, it is impossible to accurately segment the blood vessels in the retinal image where the blood vessel contrast is not obvious, and the relatively small blood vessel branches in the retinal image, so that the segmentation results for the blood vessels in the retinal image are not very good. ideal

Images