Bayesian Fully Convolutional Networks for Brain Image Registration
The purpose of medical image registration is to find geometric transformations that align two medical images so that the corresponding voxels on two images are spatially consistent. Nonrigid medical image registration is a key step in medical image processing, such as image comparison, data fusion,...
Main Authors: | , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Hindawi Limited
2021-01-01
|
Series: | Journal of Healthcare Engineering |
Online Access: | http://dx.doi.org/10.1155/2021/5528160 |
id |
doaj-4d91ab0c4d8a4aa1bc98b5d863a6ff14 |
---|---|
record_format |
Article |
spelling |
doaj-4d91ab0c4d8a4aa1bc98b5d863a6ff142021-08-09T00:01:22ZengHindawi LimitedJournal of Healthcare Engineering2040-23092021-01-01202110.1155/2021/5528160Bayesian Fully Convolutional Networks for Brain Image RegistrationKunpeng Cui0Panpan Fu1Yinghao Li2Yusong Lin3School of Information EngineeringSchool of SoftwareSchool of SoftwareCollaborative Innovation Center for Internet HealthcareThe purpose of medical image registration is to find geometric transformations that align two medical images so that the corresponding voxels on two images are spatially consistent. Nonrigid medical image registration is a key step in medical image processing, such as image comparison, data fusion, target recognition, and pathological change analysis. Existing registration methods only consider registration accuracy but largely neglect the uncertainty of registration results. In this work, a method based on the Bayesian fully convolutional neural network is proposed for nonrigid medical image registration. The proposed method can generate a geometric uncertainty map to calculate the uncertainty of registration results. This uncertainty can be interpreted as a confidence interval, which is essential for judging whether the source data are abnormal. Moreover, the proposed method introduces group normalization, which is conducive to the network convergence of the Bayesian neural network. Some representative learning-based image registration methods are compared with the proposed method on different image datasets. Experimental results show that the registration accuracy of the proposed method is better than that of the methods, and its antifolding performance is comparable to that of fast image registration and VoxelMorph. Furthermore, the proposed method can evaluate the uncertainty of registration results.http://dx.doi.org/10.1155/2021/5528160 |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Kunpeng Cui Panpan Fu Yinghao Li Yusong Lin |
spellingShingle |
Kunpeng Cui Panpan Fu Yinghao Li Yusong Lin Bayesian Fully Convolutional Networks for Brain Image Registration Journal of Healthcare Engineering |
author_facet |
Kunpeng Cui Panpan Fu Yinghao Li Yusong Lin |
author_sort |
Kunpeng Cui |
title |
Bayesian Fully Convolutional Networks for Brain Image Registration |
title_short |
Bayesian Fully Convolutional Networks for Brain Image Registration |
title_full |
Bayesian Fully Convolutional Networks for Brain Image Registration |
title_fullStr |
Bayesian Fully Convolutional Networks for Brain Image Registration |
title_full_unstemmed |
Bayesian Fully Convolutional Networks for Brain Image Registration |
title_sort |
bayesian fully convolutional networks for brain image registration |
publisher |
Hindawi Limited |
series |
Journal of Healthcare Engineering |
issn |
2040-2309 |
publishDate |
2021-01-01 |
description |
The purpose of medical image registration is to find geometric transformations that align two medical images so that the corresponding voxels on two images are spatially consistent. Nonrigid medical image registration is a key step in medical image processing, such as image comparison, data fusion, target recognition, and pathological change analysis. Existing registration methods only consider registration accuracy but largely neglect the uncertainty of registration results. In this work, a method based on the Bayesian fully convolutional neural network is proposed for nonrigid medical image registration. The proposed method can generate a geometric uncertainty map to calculate the uncertainty of registration results. This uncertainty can be interpreted as a confidence interval, which is essential for judging whether the source data are abnormal. Moreover, the proposed method introduces group normalization, which is conducive to the network convergence of the Bayesian neural network. Some representative learning-based image registration methods are compared with the proposed method on different image datasets. Experimental results show that the registration accuracy of the proposed method is better than that of the methods, and its antifolding performance is comparable to that of fast image registration and VoxelMorph. Furthermore, the proposed method can evaluate the uncertainty of registration results. |
url |
http://dx.doi.org/10.1155/2021/5528160 |
work_keys_str_mv |
AT kunpengcui bayesianfullyconvolutionalnetworksforbrainimageregistration AT panpanfu bayesianfullyconvolutionalnetworksforbrainimageregistration AT yinghaoli bayesianfullyconvolutionalnetworksforbrainimageregistration AT yusonglin bayesianfullyconvolutionalnetworksforbrainimageregistration |
_version_ |
1721215400847343616 |