Brain image registration method based on DGAN

An image registration and brain technology, applied in the field of medical image processing, can solve the problems of artificial registration misjudgment, missed diagnosis, low timeliness, etc., and achieve the effect of strong generalization ability and fast brain image registration

Inactive Publication Date: 2021-11-30
HENAN UNIVERSITY OF TECHNOLOGY
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] In order to solve the problem of misjudgment and missed diagnosis in manual registration in medical image registration, and the timeliness of traditional registration methods is too low, the present invention provides a method for automatically and accurately registering brain images using generative confrontation networks

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
  • Brain image registration method based on DGAN
  • Brain image registration method based on DGAN
  • Brain image registration method based on DGAN

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0016] In order to verify the performance of the present invention in GAN-based brain image registration, we selected the 2018 BraTS public dataset for training, verification and testing.

[0017] Step 1: Preprocessing the brain image data, using Pycharm software, using image rotation, translation transformation for image enhancement, image denoising, and contrast enhancement methods for image normalization.

[0018] Step 2: Train the GAN network in Pycharm software, batch_size is 1, learning_rate is 2e-4, Adama optimizer is used, kinetic energy is 0.5 for optimization, training iterations are 20,000 times, the training set and verification set are divided by 11:1, intermittent Train the above two processes, adjust the network parameters until the network converges, and the training ends.

[0019] Step 3: Use the test set of the 2018 BraTS dataset to test the GAN network, and the experimental results are evaluated by the coincidence coefficient (Dice).

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 existing brain image registration technology has defects in registration precision and speed, and is poor in coping capacity for large deformation in an image to be registered. In order to realize efficient and rapid registration of brain images and facilitate treatment of doctors, the invention provides a brain image registration method (D-GAN) based on a generative adversarial network. A convolutional neural network and denoising and feature enhancement are added in front of the generated model. By adding denoising and feature enhancement, the definition of the image can be higher, and more accurate processing of the image is facilitated. In the generation model, extraction of multi-scale features of data is realized through a U-shaped convolutional network, a discrimination network inputs a sample probability graph obtained by the generation model and a sample truth value label into a deep convolutional network together, a real sample is endowed with a relatively high label, and a generated sample is endowed with a relatively low label. Network parameter weights are adjusted through mutual counterbalance of the generation model and the discrimination model, and the whole network is converged when the generation model and the discrimination model are balanced. According to the invention, the generalization ability of the model can be enhanced by automatically learning the mapping relation between the same data set. According to the invention, accurate and rapid registration can be realized.

Description

technical field [0001] The invention belongs to the technical field of medical image processing, and in particular relates to a method for brain image registration. Background technique [0002] Medical imaging technology provides modern clinical medicine with X-ray imaging, CT computed tomography, US ultrasound imaging, MRI magnetic resonance imaging, FMRI functional magnetic resonance imaging, MRA magnetic resonance angiography, DSA digital subtraction angiography, SPECT single photon emission Tomography imaging, PET positron emission tomography imaging, DF digital decane light contrast imaging and other image information FML with different functions and forms. In clinical diagnosis and treatment, doctors often need to obtain comprehensive and useful information from images obtained in different modalities or in different periods, W for disease diagnosis, treatment and surgical follow-up. Although various imaging technologies have been developed rapidly, and the imaging e...

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): G06T7/00G06T7/33G06T5/00G06N3/04G06N3/08
CPCG06T7/0012G06T7/33G06T5/002G06N3/04G06N3/08G06T2207/20081G06T2207/20084G06T2207/20132G06T2207/30016
Inventor 张鑫李冰洁杨铁军赵祥
Owner HENAN UNIVERSITY OF TECHNOLOGY
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products