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

A CT image reconstruction method based on convolutional neural network

A convolutional neural network and CT image technology, which is applied in the field of CT image reconstruction based on convolutional neural network, can solve the problems of long time, failure to meet the real-time imaging requirements of CT, and image resolution reduction.

Active Publication Date: 2021-06-15
SOUTHERN MEDICAL UNIVERSITY
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] Disadvantages and causes of the iterative reconstruction method based on the statistical model: when reconstructing a CT image of the same size, the time spent by the iterative reconstruction method based on the statistical model is much longer than that of the traditional analytical reconstruction method, which cannot meet the clinical needs of CT. Real-time imaging requirements, the reason is that the iterative reconstruction method based on the statistical model requires dozens or even hundreds of iterations to solve the objective function, resulting in a significant increase in image reconstruction time
[0005] Disadvantages and causes of the analytical reconstruction method based on projection data filtering: The traditional analytical reconstruction method based on projection data filtering will inevitably lead to the loss of original image detail information during the projection data denoising process, resulting in the loss of the corresponding CT image resolution drop

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
  • A CT image reconstruction method based on convolutional neural network
  • A CT image reconstruction method based on convolutional neural network
  • A CT image reconstruction method based on convolutional neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0043] like Figure 1-3 As shown, a CT image reconstruction method based on convolutional neural network includes the following steps:

[0044] A1. For the original chord diagram data I K Perform back-projection operation to obtain back-projected image data I' K .

[0045] In step A1, the back-projection operation is specifically performed on the original chord diagram data I by the CT scanner. K Perform geometric imaging processing.

[0046] A2. Back-projection image data I' K Perform normalization processing to obtain normalized back-projection image data P K .

[0047] The method steps of normalization processing in step A2 are as follows:

[0048] T1. Calculate back-projection image data I' K mean X I’ and variance S I’ .

[0049] T2. Calculate the normalized back-projection image data P according to formula (1). K .

[0050] P K =(I' K -X I’ ) / S I’ ······Formula 1).

[0051] A3. Normalize the back-projection image P K Convolutional neural network filter...

Embodiment 2

[0067] A CT image reconstruction method based on convolutional neural network, other features are the same as those in Embodiment 1, except that the nonlinear activation function δ( ) is a sigmoid function, which is calculated according to formula (4):

[0068] δ(x)=1 / (1+e -x ), e is a natural base, ······ formula (4).

[0069] It should be noted that the nonlinear activation function δ(·) can select the type of nonlinear activation function according to the objective function of the convolutional neural network training and the optimization algorithm.

[0070] The processing method is simple in operation and convenient in processing, and can greatly reduce image noise and artifacts while maintaining the resolution of the original image, and finally achieve high-quality reconstruction of the CT image.

Embodiment 3

[0072] A CT image reconstruction method based on a convolutional neural network, other features are the same as in Embodiment 1, except that the nonlinear activation function δ( ) is a hyperbolic tangent function, and is calculated according to formula (5):

[0073] δ(x)=(e x -e -x ) / (e x +e -x ), where e is a natural base, ······ formula (5).

[0074] It should be noted that the nonlinear activation function δ(·) can select the type of nonlinear activation function according to the objective function of the convolutional neural network training and the optimization algorithm.

[0075] The processing method is simple in operation and convenient in processing, and can greatly reduce image noise and artifacts while maintaining the resolution of the original image, and finally achieve high-quality reconstruction of the CT image.

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

A CT image reconstruction method based on convolutional neural network, including the original chord data I K Perform the back projection operation to obtain the back projection image data I' K , for the backprojection image data I’ K Perform normalization processing to obtain normalized back-projection image data P K , the normalized backprojection image P K Convolutional neural network filtering is performed through a convolutional neural network to generate an image to be processed P’ K , the image to be processed P’ K Perform denormalization processing to obtain the final reconstructed image P final . There is no need for special design of image filtering, and the learning of image filtering is automatically completed through the training of the convolutional neural network model. The reconstruction method of the present invention is simple to operate, convenient to process, and can greatly reduce image noise and artifacts, while better Maintain the resolution of the original image, and finally achieve high-quality reconstruction of CT images.

Description

technical field [0001] The invention relates to the technical field of image reconstruction methods for medical images, in particular to a CT image reconstruction method based on a convolutional neural network. Background technique [0002] X-ray CT has been widely used in clinical medical imaging diagnosis, but the high dose of X-ray radiation in CT scans has the risk of cancer. How to minimize the dose of X-rays has become a key technology in the field of medical CT imaging. [0003] Currently, reducing the tube current and scanning time during CT scanning is the easiest and most common way to achieve low-dose CT imaging. However, due to the reduction of tube current and scanning time, the projection data contains a lot of noise, and the quality of the reconstructed image based on the traditional filtered back projection method is seriously degraded, which is difficult to meet the needs of clinical diagnosis. In order to greatly reduce the X-ray radiation dose on the pre...

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 Patents(China)
IPC IPC(8): G06T11/00G06T5/00G06N3/04G06N3/08
CPCG06T11/006G06T11/008G06N3/08G06T2211/421G06T2207/10081G06T2207/20024G06T2207/20081G06T2207/20084G06N3/045G06T5/70
Inventor 马建华何基边兆英曾栋黄静
Owner SOUTHERN MEDICAL UNIVERSITY
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