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

Multi-task relationship learning method for centrum positioning, identification and segmentation in nuclear magnetic resonance imaging

A technology of nuclear magnetic resonance image and learning method, applied in the field of medical image processing, can solve the problem of ignoring the close connection between tasks

Active Publication Date: 2020-05-22
SHANDONG UNIV
View PDF7 Cites 7 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, previous methods ignore the close connection between tasks

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
  • Multi-task relationship learning method for centrum positioning, identification and segmentation in nuclear magnetic resonance imaging
  • Multi-task relationship learning method for centrum positioning, identification and segmentation in nuclear magnetic resonance imaging
  • Multi-task relationship learning method for centrum positioning, identification and segmentation in nuclear magnetic resonance imaging

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0107] A multi-task relational learning method for vertebral body localization, recognition and segmentation in MRI images, such as figure 1 shown, including the following steps:

[0108] (1) Image preprocessing

[0109] By preprocessing the NMR images and semantic segmentation labels, the final data structure meets the requirements for input and loss function calculation of the multi-task relational learning network model; in this embodiment, the NMR images refer to MR lumbar spine images ;

[0110] (2) Build a multi-task relational learning network model

[0111] like figure 2 As shown, the multi-task relational learning network model includes Seg-Loc network, XOR operation and discriminant network;

[0112] As a generator for adversarial training, the Seg-Loc network uses the task mutual attention module to learn the relationship between semantic localization and semantic segmentation end-to-end through network parameter learning, and outputs semantic localization resu...

Embodiment 2

[0127] According to a kind of multi-task relational learning method that is used for the vertebral body location, identification and segmentation in the magnetic resonance image described in embodiment 1, its difference is:

[0128] In step (1), comprise steps as follows:

[0129] The original MRI image faces some challenges, such as the weak edge information of the vertebral body; the strong noise leads to uneven gray scale of the vertebral body imaging; the diversity of resolution leads to different sizes of vertebrae in the data set; and the generated MRI spine image contains With different degrees of lesions, each image contains different numbers of vertebral bodies. After statistics, the number of MRI images containing 6 vertebral bodies (S1-L5), 7 vertebral bodies (S1-T12), and 8 vertebral bodies (S1-T11) are nearly equal.

[0130] A. First adjust all MRI images to 512*512;

[0131] B. Use ITK-SNAP software to label all MRI images with vertebral body segmentation label...

Embodiment 3

[0141] According to a kind of multi-task relational learning method that is used for the vertebra body location, identification and segmentation in the nuclear magnetic resonance image described in embodiment 2, its difference is:

[0142] In step (2), the Seg-Loc network is the generative network of the proposed multi-task relational learning network. like image 3 As shown, the structure of the Seg-Loc network is an encoder-decoder network. The encoder-decoder network includes an encoder, two decoders, and two task mutual attention modules. The two decoders share an encoder. There are two encoders between the two decoders. A task mutual attention module;

[0143] The two decoders output semantic positioning results and semantic segmentation results respectively; the task mutual attention module is used to learn the relationship between semantic positioning and semantic segmentation;

[0144] The encoder includes convolutional layer, LSTM, hole convolution group, batch norm...

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 relates to a multi-task relationship learning method for centrum positioning, identification and segmentation in nuclear magnetic resonance imaging. According to the method, a relationship among multiple tasks is fully utilized based on deep learning, and challenges caused by intervertebral similarity and image quality are greatly improved. For automatic analysis of the spine, an effective multi-task learning framework is provided. The framework can be easily popularized to the application of other images, and a universal framework is provided for effectively solving the three tasks of image positioning, identification and segmentation.

Description

technical field [0001] The invention relates to a multi-task relational learning method for vertebral body positioning, recognition and segmentation in nuclear magnetic resonance imaging, and belongs to the technical field of medical image processing. Background technique [0002] In the context of computer-assisted spinal surgery, it is important to know exactly the shape of individual vertebral bodies, e.g. for spinal biopsy, implant or pedicle screw insertion, etc. But in most cases, it is not only required to obtain the shape of the vertebral body through accurate segmentation, but also to locate and identify the vertebral body. Automatic segmentation, localization, and labeling of vertebral bodies in computed tomography (CT) or magnetic resonance imaging (MRI) spine imaging has become an important tool for clinical tasks, including pathological diagnosis, surgical planning, and postoperative evaluation. Specific applications such as fracture detection, tumor detection....

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): G06T7/00G06T7/11G06T7/66G06T7/73G06K9/62G06N3/04
CPCG06T7/0012G06T7/11G06T7/66G06T7/73G06T2207/30012G06N3/045G06F18/253Y02A90/30
Inventor 李玉军张冉冉刘治张文真李邦军
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