Visual cup and optic disc segmentation method based on depth level set learning

Abstract

The invention discloses an optic cup and optic disk segmentation method based on depth level set learning, comprising the following steps: S1, inputting an eye fundus image into a U-shaped neural network, the U-shaped neural network outputting a level set function value of each pixel in the image, a corresponding zero level set being a contour of a target object; s2, using a level set loss function to calculate an error between the output of the U-shaped neural network and a real label and update parameters of the U-shaped neural network, and introducing a contour length constraint and a region consistency constraint into the loss function, so that the neural network can obtain a level set meeting priori knowledge; and S3, training the two U-shaped neural networks to segment the optic disc and the optic cup respectively. The high-precision optic cup and optic disc segmentation method is realized, and the method can be effectively applied to automatic diagnosis of retinal diseases.

Description

technical field [0001] The invention belongs to the technical field of medical image segmentation, and in particular relates to an optic cup and optic disc segmentation method based on deep level set learning. Background technique [0002] In clinical medicine, doctors often diagnose retinal diseases by evaluating structural changes in the optic nerve head region (Optic Nerve Head, ONH). A commonly used method for evaluating ONH is the vertical cup-to-disk ratio, which is the ratio of the vertical cup diameter to the vertical optic disc diameter. Since manual diagnosis is time-consuming and requires well-trained physicians, automatic optic cup and optic disc segmentation techniques have attracted extensive research. In the traditional level set method, the contour of the target object in the image is indirectly expressed by the zero level set of the high-dimensional surface (that is, the level set function), and then the evolution of the level set function can be used to si...

AI Technical Summary

Problems solved by technology

For the segmentation task of the optic cup and optic disc, although a general segmentation network can also be used, it cannot make full use of the prior knowledge of the shape and structure of the optic cup or optic disc.

Images