Unlock instant, AI-driven research and patent intelligence for your innovation.

Image denoising method in low-dose CT projection domain based on improved U-net

A CT image, low-dose technology, applied in the field of image processing, can solve problems such as inaccurate noise recognition and loss of image details, and achieve the effects of improving image quality, improving structure, and preventing important structural information from being saved.

Pending Publication Date: 2021-02-05
PLA STRATEGIC SUPPORT FORCE INFORMATION ENG UNIV PLA SSF IEU
View PDF5 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] Aiming at the problem of inaccurate noise recognition and loss of image details existing in the existing end-to-end image domain processing methods, the present invention provides a method that can quickly and accurately complete low-dose CT image noise removal in the projection domain by using a deep network

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
  • Image denoising method in low-dose CT projection domain based on improved U-net
  • Image denoising method in low-dose CT projection domain based on improved U-net
  • Image denoising method in low-dose CT projection domain based on improved U-net

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0036] An embodiment of the present invention provides an improved U-net-based image denoising method in the low-dose CT projection domain, including:

[0037] S101: Based on the noise characteristics of the CT image projection domain, create matched normal dose and low dose projection image datasets;

[0038] S102: Use the inception module to replace the convolutional layer part of the U-net network encoder;

[0039] S103: further optimize, train and test the U-net network structure;

[0040] S104: Input the low-dose CT projection data that needs to be de-noised into the trained U-net network, obtain the de-noised projection data, and perform filter back-projection reconstruction on the de-noised projection data.

[0041] Considering that the source of noise is located in the projection domain, it may be easier to study low-dose CT in the region closest to the noise, and the streak artifact appears as scattered noise in the projection domain. The low-dose CT image denoising...

Embodiment 2

[0044]On the basis of the above embodiment, the embodiment of the present invention provides another method for image denoising in the low-dose CT projection domain based on the improved U-net, including the following steps:

[0045] S201 : generating fan beam geometric projection data by using the Siddon ray driving method from the CT image of normal dose;

[0046] S202: When the light source is a monochromatic X-ray, approximate the noise in the low-dose CT projection by the Poisson noise shown in formula (1):

[0047] Z i ~Poisson{Z 0i exp(-s i )+r i}i=1,2...I(1)

[0048] Among them, Z i is the number of incident photons along the i-th X-ray path, Z 0i is the initial incident intensity of X-ray, r i is the background electronic noise along the i-th X-ray path, s i is the line integral of the attenuation coefficient;

[0049] As an implementation, Z 0i is evenly set to 10 5 .

[0050] S203: Add the noise obtained by the simulation in step S202 to the projection d...

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 image denoising method in a low-dose CT projection domain based on improved U-net. The method comprises the following steps of step 1, making matched normal dose and low doseprojection image data sets based on noise characteristics of a CT image projection domain; step 2, replacing a convolutional layer part of the U-net network encoder with an inception module; step 3,further optimizing, training and testing the U-net network structure; and step 4, inputting the low-dose CT projection data to be denoised into the trained U-net network to obtain denoised projectiondata, and performing filtering back projection reconstruction on the denoised projection data. According to the method, noise removal of the projection image is completed by mainly utilizing the distribution characteristics of the noise characteristics of the projection domain image and the advantages of the U-net network in the aspect of processing a small data set, and the noise is removed by learning an end-to-end mapping from a low-dose CT sinogram containing the noise to a corresponding noiseless normal-dose CT sinogram through the network.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to an improved U-net-based image denoising method in the low-dose CT projection domain. Background technique [0002] X-ray computed tomography (CT) has developed rapidly and is widely used in various fields of medical diagnosis. It has a good advantage in diagnosing small lesions inside the human body, so it has an irreplaceable role in many scanning instruments. Under low-dose conditions, the projection data will be destroyed due to the decrease in the number of photons received by the detector, resulting in serious noise and artifacts in the reconstructed CT image, which affects the judgment of the radiologist. In order to produce high-quality images that meet the diagnostic needs while ensuring patient safety, many research methods have been proposed related to the problem of low-dose CT. [0003] In traditional methods, low-dose CT denoising methods can be roughly di...

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): G06T5/00G06N3/04G06N3/08
CPCG06N3/08G06T2207/10081G06T2207/20081G06T2207/20084G06T2207/30004G06N3/045G06T5/70
Inventor 李磊宋晓芙韩玉张军政席晓琦闫镔朱林林孙艳敏刘梦楠
Owner PLA STRATEGIC SUPPORT FORCE INFORMATION ENG UNIV PLA SSF IEU
Features
  • R&D
  • Intellectual Property
  • Life Sciences
  • Materials
  • Tech Scout
Why Patsnap Eureka
  • Unparalleled Data Quality
  • Higher Quality Content
  • 60% Fewer Hallucinations
Social media
Patsnap Eureka Blog
Learn More

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

© 2025 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com