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

Low-dose PET data three-dimensional iterative updating reconstruction method based on deep learning

A deep learning and iterative update technology, applied in the field of medical imaging, can solve the problems of inability to use effective information, image artifacts and quantitative error performance bottlenecks, effective high-frequency information cannot be completely preserved, etc., to achieve the suppression of insufficient generalization ability, The effect of less time consumption and low computational complexity

Active Publication Date: 2021-01-22
ZHEJIANG LAB +1
View PDF6 Cites 6 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Since the reconstruction process of low-dose images is seriously affected by noise, the effective high-frequency information in the original PET data cannot be completely preserved, and these effective information cannot be used only by post-processing the reconstructed images, resulting in insufficient generalization ability of deep learning. The resulting image artifacts and quantitative errors have become the performance bottleneck of this type of technical route

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
  • Low-dose PET data three-dimensional iterative updating reconstruction method based on deep learning
  • Low-dose PET data three-dimensional iterative updating reconstruction method based on deep learning
  • Low-dose PET data three-dimensional iterative updating reconstruction method based on deep learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0025] The present invention provides a method for three-dimensional iterative update and reconstruction of low-dose PET data based on deep learning, which specifically includes the following steps:

[0026] Such as figure 1 It is a flow chart of the forward and reverse mapping network training method based on low-dose and standard-dose PET raw data:

[0027] (1) The low-dose sinogram and standard-dose sinogram are respectively subjected to attenuation, randomization, scatter correction, back projection and normalization to obtain the 3D representation of low-dose PET data in the image domain and standard dose PET data in 3D representation in the image domain ; The back projection is to back-project the sinogram to the image domain to obtain a highly blurred PET image laminogram; the highly blurred PET image laminogram has the following relationship with the PET reconstruction image:

[0028]

[0029] in, and Respectively represent a pixel on the 3D PET reconstructi...

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 low-dose PET data three-dimensional iterative update reconstruction method based on deep learning, which starts from low-dose PET original data, and comprises the steps of utilizing a three-dimensional deep neural network to fit forward and reverse mapping between original data back projection image and a PET reconstructed image; learning standard dose priori knowledge from the training sample and applying the standard dose priori knowledge to iteratively update the standard dose laminogram and the PET three-dimensional reconstructed image so as to obtain the PET reconstructed image which is lower in noise, higher in resolution and free of artifacts compared with a traditional reconstruction method and image domain noise reduction processing.

Description

technical field [0001] The invention belongs to the technical field of medical imaging, and in particular relates to a three-dimensional iterative update and reconstruction method for low-dose PET data based on deep learning. Background technique [0002] Positron Emission Tomography (PET) is a medical image that can provide biochemical and quantitative information in vivo. Because it can be imaged at the molecular level, and can simultaneously reflect human biological function metabolism information and morphological anatomical structure information, it makes It has important applications in oncology, cardiology, neurology, and psychiatric diseases. The PET imaging process includes injecting a radioactive tracer into the patient before scanning. The tracer decays to generate positrons when participating in physiological metabolism in the human body. The positrons and electrons in adjacent tissues have an annihilation effect, resulting in 511keV photon pairs that move invers...

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): G06T17/00G06N3/04G06N3/08
CPCG06T17/00G06N3/08G06T2200/08G06T2210/41G06N3/045
Inventor 杨宝朱闻韬周龙叶宏伟
Owner ZHEJIANG LAB