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

Medical image target area identification method, neural network model and application

A technology of neural network model and target area, which is applied in the identification of medical image target area, neural network model and application fields, can solve the problems of many hyperparameters, slow reasoning speed, and long iteration cycle, so as to improve calculation speed and accuracy The effect of improving the degree of robustness

Active Publication Date: 2021-10-19
ANHUI MEDICAL UNIV
View PDF11 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] However, the traditional segmentation network model has the defects of inaccurate target area segmentation, susceptible to interference from surrounding targets, and slow inference speed.
At the same time, there is also the need to consume a large amount of video memory and it is difficult to train the network model, and the training iteration cycle is long, and there are many hyperparameters, which makes adjustment difficult
In addition, as the depth of the network model deepens, its sensitivity to smaller target recognition will decrease.

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
  • Medical image target area identification method, neural network model and application
  • Medical image target area identification method, neural network model and application
  • Medical image target area identification method, neural network model and application

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0052] This embodiment discloses a method for identifying a target area of ​​an image. The image in this embodiment is introduced using an MRI image as an example, and the target area identified in this embodiment is introduced using a brain tumor in an MRI image as an example. The Dicom file includes MRI image weight information, and the target keyword is T2;

[0053] Include the following steps:

[0054] Step 1, traverse all Dicom files, use the open source tool PyDicom to read the sequence information of each Dicom file, and determine whether the sequence information contains the keyword T2, if the keyword T2 is included, then determine that the Dicom file is the target file, And use the PyDicom third-party library to read the matrix information in the Dicom file, and save the matrix information as an image in JPG format.

[0055] The format of the pictures taken by the nuclear magnetic resonance apparatus in the hospital is all Dicom format. The so-called Dicom format (D...

Embodiment 2

[0120] This embodiment discloses a neural network model based on a residual structure, including a shallow backbone network structure, a middle backbone network structure, a deep backbone network structure, an SPP layer, a first detection head, a second detection head, and a third detection head;

[0121] Each backbone network structure includes a residual structure, and the residual structure includes the basic unit formed by the first processing of the 1×1 Conv layer and the post-processing of the 3×3 Conv layer; in the same residual structure There is at least one basic unit; when the image to be recognized is processed by the shallow backbone network structure and input to the first detection head, it can output the target image with the first receptive field; when the image to be recognized is sequentially passed through the The shallow backbone network structure and the middle layer backbone network structure are processed and input to the second detection head, which can...

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 method for identifying a target area of a medical image, and the method comprises the following steps: reading information of a current Dicom file, and if the information comprises a target keyword, processing the Dicom file to obtain an image to be identified; inputting the to-be-recognized image into a neural network model based on a residual structure, and outputting the to-be-recognized image through different detection heads to obtain target images with different receptive fields, wherein the neural network model based on the residual structure comprises a backbone network structure, an SPP layer and a detection head. The invention further discloses a neural network model based on the residual structure. The network model established by the invention has the advantages of good non-linear expression capability, reduction of the parameter quantity of the network model, improvement of the calculation speed of the network model, improvement of the accuracy and robustness of the network model, good sensitivity to targets of different sizes, and accurate target identification.

Description

technical field [0001] The invention relates to the medical field, in particular to a recognition method, a neural network model and an application of a medical image target area. Background technique [0002] Currently, for image processing, especially for MRI image recognition, image segmentation techniques are mostly used. For example, a magnetic resonance image segmentation method, device, terminal equipment, and storage medium disclosed in patent application 201911243400.4, and an improved glioma segmentation method using cross-sequence nuclear magnetic resonance image generation disclosed in patent application 202011164826.3. [0003] However, the traditional segmentation network model has the defects of inaccurate target area segmentation, susceptible to interference from surrounding targets, and slow inference speed. At the same time, there is also the need to consume a large amount of video memory and it is difficult to train the network model, and the training ite...

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): G16H30/20G06K9/32G06K9/62G06N3/04G06N3/08
CPCG16H30/20G06N3/08G06N3/045G06F18/214
Inventor 单淳劼赵维佳梁振
Owner ANHUI MEDICAL 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