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

Super-resolution reconstruction algorithm for medical imaging

A technology for super-resolution reconstruction and medical imaging, which is applied in computing, image data processing, graphics and image conversion, etc. It can solve problems such as difficult network training, waste of computing resources, and reduced network performance, so as to speed up convergence, reduce parameters, Effects that improve image quality

Pending Publication Date: 2020-01-21
TIANJIN UNIV
View PDF0 Cites 12 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

As the depth increases, the network generates more and more redundant information, which wastes a lot of computing resources and even degrades network performance.
[0003] The depth of the convolutional neural network has a great influence on the performance of the network, but an overly deep network is difficult to train
The shallow features of the network contain rich high-frequency information, which is very important for the reconstruction of image details, but these high-frequency information are treated equally among channels, which hinders the expressive ability of the 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
  • Super-resolution reconstruction algorithm for medical imaging
  • Super-resolution reconstruction algorithm for medical imaging
  • Super-resolution reconstruction algorithm for medical imaging

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0030] The present invention will be described in further detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention, not to limit the present invention.

[0031] Such as Figure 1-4 As shown, the super-resolution reconstruction algorithm for medical imaging of the present invention is realized by using the built global attention network to perform the following operations in sequence, and the steps are as follows:

[0032] S1. Use shallow feature extraction module to extract shallow feature F shallow (x):

[0033] f shallow (x) = Conv 3×3 (x)

[0034] S2. Using the deep feature extraction module to extract deep features F deep (x):

[0035] f deep (x)=BLOCK(x)

[0036] =ReLu(InstanceNorm(Conv 3×3 (x)))

[0037] Among them, Conv 3×3 () is a convolutional layer, InstanceNorm() is an instance normalization layer, and ReLu is a ...

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 super-resolution reconstruction algorithm for medical imaging. The super-resolution reconstruction algorithm comprises the following steps: extracting shallow features containing high-frequency information and deep features containing abstract information of a target medical image by utilizing a global attention network; after the image size is up-sampled and amplified, carrying out global attention processing, and obtaining global features; and on the basis of the global features, carrying out image reconstruction, and outputting a reconstructed image. According to the method, important feature components of the input image are extracted through a double-attention mechanism, the image quality of super-resolution reconstruction is improved, image details which areextremely critical to medical analysis are recovered, neural network parameters are reduced, and the convergence speed of the neural network is increased.

Description

technical field [0001] The invention relates to the technical field of computer vision, in particular to a super-resolution reconstruction algorithm for medical imaging. Background technique [0002] Deep learning-based super-resolution reconstruction techniques such as convolutional neural networks, residual learning, and dense connections are widely used in super-resolution reconstruction work, such as three-layer convolutional neural networks for single-image super-resolution reconstruction tasks (CNN), called SRCNN, optimizes the feature extraction, nonlinear mapping, and image reconstruction stages in an end-to-end manner. However, limited by the number of convolutional layers, this method extracts limited features, and the restoration quality still needs to be improved. On the basis of SRCNN, someone proposed a deeper network structure - VDSR, with a depth of 20 layers, and introduced residual learning and gradient clipping to reduce the difficulty of training. As th...

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): G06T3/40G06N3/04
CPCG06T3/4076G06N3/045
Inventor 雷志春李泽田
Owner TIANJIN UNIV