Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Retina hemangioma image segmentation method based on GAN (Generative Adversarial Network)

A retinal blood vessel and image segmentation technology, applied in image analysis, biological neural network model, image enhancement, etc., can solve the problem of missing details of blood vessel branches, useless information, complex distribution of thick and thin blood vessels, and inability to take into account the integrity and accuracy of hemangioma. , to achieve the effect of reducing segmentation results, accurate segmentation details, and improving segmentation efficiency

Pending Publication Date: 2018-09-14
SHANDONG UNIV
View PDF4 Cites 24 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

For retinal hemangioma images, due to the mixed distribution of blood vessels of different thicknesses in such images, traditional segmentation techniques often cannot take into account the integrity and accuracy of hemangiomas, that is, some small details of blood vessel branches may be missing or contain redundant useless information

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Retina hemangioma image segmentation method based on GAN (Generative Adversarial Network)
  • Retina hemangioma image segmentation method based on GAN (Generative Adversarial Network)
  • Retina hemangioma image segmentation method based on GAN (Generative Adversarial Network)

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0042] This embodiment discloses a retinal hemangioma segmentation method based on generating an adversarial network, such as figure 1 shown, including the following steps:

[0043] (1) Obtain training data

[0044] First, the fundus image of the retina is collected. The fundus image provides sufficient information and is widely used in disease diagnosis. The fundus image is directly obtained by the fundus camera. The fundus image is preprocessed, and the resolution of the image is unified to N×N, and the hemangioma in the image is manually and accurately segmented, and the segmented data is used for subsequent training.

[0045] (2) Construct Generative Adversarial Network

[0046] Construct generator G(x, the generator network includes at least a plurality of downsampling network layers, upsampling network layers with the same number of downsampling network layers. Wherein, the downsampling network layer includes a downsampling layer, a volume product layer and an activat...

Embodiment 2

[0060] The purpose of this embodiment is to provide a computing device.

[0061] In order to achieve the above object, the present invention adopts the following technical scheme:

[0062] A device for segmenting images of retinal hemangiomas based on generating an adversarial network, comprising a memory, a processor, and a computer program stored on the memory and operable on the processor, when the processor executes the program, it realizes:

[0063] Construct generation confrontation network structure, described generation confrontation network comprises generator and discriminator;

[0064] Using artificially segmented images of hemangioma and original fundus retinal images as training data, iteratively trains the generated confrontation network to obtain an optimal generator;

[0065] The hemangioma image is segmented based on the optimal generator for the fundus retinal image to be segmented.

Embodiment 3

[0067] The purpose of this embodiment is to provide a computer-readable storage medium.

[0068] In order to achieve the above object, the present invention adopts the following technical scheme:

[0069] A computer-readable storage medium having stored thereon a computer program which, when executed by a processor, performs:

[0070] Construct generation confrontation network structure, described generation confrontation network comprises generator and discriminator;

[0071] Using artificially segmented images of hemangioma and original fundus retinal images as training data, iteratively trains the generated confrontation network to obtain an optimal generator;

[0072] The hemangioma image is segmented based on the optimal generator for the fundus retinal image to be segmented.

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a retina hemangioma image segmentation method based on a GAN (Generative Adversarial Network), and the method comprises the steps: constructing a GAN structure, wherein the GANstructure comprises a generator and a discriminator, the generator comprises a plurality of downsampling layers and an equal number of upsampling layers, and the discriminator comprises a plurality of downsampling layers; taking a hemangioma artificial segmented image and an original fundus retina image as training data, performing iterative training of the GAN, and obtaining an optimal generator; performing hemangioma image segmentation of a to-be-segmented fundus retina image based on the optimal generator. Compared with a conventional retina hemangioma image segmentation method based on animage processing algorithm, the method can achieve the more accurate and clear segmentation of the image, reduces a false positive segmentation result, and is more precise in image segmentation details.

Description

technical field [0001] The invention relates to the field of medical image processing, in particular to the technical fields of generating confrontational networks and retinal hemangioma segmentation. Background technique [0002] Retinal hemangiomas are part of hemangiomatosis, which can be solitary, sporadic, noninherited, or autosomal dominant. As the tumor grows and expands, it can impair vision and, in more severe cases, lead to glaucoma, uveitis, traction retinal detachment, concurrent cataracts, or atrophy of the globe leading to complete loss of vision. Doctors can diagnose related lesions by observing the retinal fundus images of patients. Therefore, periodic examination and diagnosis of patients in advance can allow patients to prevent related diseases in advance. Generally, hemangiomas are detected by ophthalmologists, but it takes a lot of time; using computer image processing technology to segment fundus images of retinal hemangiomas can effectively improve the...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06T7/11G06N3/04G06N3/08
CPCG06N3/084G06T7/11G06T2207/20081G06T2207/20084G06T2207/10004G06T2207/30096G06T2207/30101G06T2207/30041G06N3/045
Inventor 李建文刘森泽刘治肖晓燕曹艳坤
Owner SHANDONG UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com