SD-OCT denoising method based on unsupervised adversarial neural network

A SD-OCT, EDI-OCT technology, applied in neural learning methods, biological neural network models, image data processing, etc., can solve the problems of a large number of label samples, poor SD-OCT denoising effect, etc., to remove image noise. and the effect of bar artifacts

Pending Publication Date: 2019-11-08
NANJING UNIV OF TECH
View PDF4 Cites 7 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to provide a SD-OCT denoising method based on an unsupervised generation confrontation network, which mainly solves the problem that the existing SD-OCT denoising effect is not good and requires a large number of label samples

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
  • SD-OCT denoising method based on unsupervised adversarial neural network
  • SD-OCT denoising method based on unsupervised adversarial neural network
  • SD-OCT denoising method based on unsupervised adversarial neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0032] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0033] The present invention comprises the steps:

[0034] Step 1: Collect N SD-OCT images and M EDI-OCT images, M EDI-OCT images are the denoising images of M SD-OCT images respectively, N>M; extract N SD-OCT images respectively Retinal anatomical structure area, register the EDI-OCT image with its corresponding SD-OCT image, and find the retinal anatomical structure area in the EDI-OCT image;

[0035] Step 2: Take the retinal anatomical structure area in N SD-OCT images as image data samples, and M EDI-OCT images as sample labels to construct an image data sample set;

[0036] Step 3: Design a recurrent generative confrontation network with global structure and local structure constraints;

[0037] Step 4: Use the image data sample set to train the recurrent generative confrontation network to obtain the SD-OCT denoising model sensitive to structura...

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 provides an SD-OCT denoising method based on an unsupervised adversarial neural network, and the method achieves the generation of an image with EDI-OCT quality from SD-OCT through the unsupervised learning of the migration from an SD-OCT image domain to an EDI-OCT image domain, thereby achieving the purpose of denoising. The global structure loss and the local structure loss are added to the model, so that the structure information and the local details of the image can be effectively maintained. Compared with an existing image noise reduction algorithm, the method has the advantages that image noise and strip artifacts are effectively removed from the processed image of the model, local details of the image are well reserved, and the processed image is closer to the original image. Meanwhile, the parameter index of the model prediction image is higher than other algorithm processing results. Therefore, no matter from subjective visual effect or objective quality evaluation, the processing result of the model provided by the invention is superior to the processing results of other algorithms, which indicates that the model provided by the invention is feasible and effective in the aspect of SD-OCT image denoising.

Description

technical field [0001] The invention belongs to the technical field of SD-OCT denoising processing, and in particular relates to an SD-OCT denoising method based on an unsupervised confrontational neural network. Background technique [0002] Frequency-domain optical coherence tomography is a non-invasive, non-ionizing optical imaging mode, which has the advantages of high resolution, non-contact, non-invasive, and fast scanning speed, and has been widely used in the diagnosis and measurement of clinical ophthalmology. However, due to the limitation of low-coherence interferometry and imaging equipment, the image quality of SD-OCT is often affected by speckle noise, which will cover up subtle image features and affect the follow-up processing of SD-OCT and the diagnosis of diseases. Suppressing speckle noise is a key issue in the field of OCT imaging. [0003] The current traditional denoising methods for SD-OCT images can be divided into two categories: methods based on th...

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
Patent Type & Authority Applications(China)
IPC IPC(8): G06T5/00G06N3/08
CPCG06T5/002G06N3/08G06T2207/10101G06T2207/20081G06T2207/20084G06T2207/30041
Inventor 吴梦麟蔡鑫鑫徐晓瑀
Owner NANJING UNIV OF TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap