使用深度卷积神经网络和强化学习的肺脏微分同胚配准
Diffeomorphic Lung Registration Using Deep CNNs and Reinforced Learning.
作者信息
Onieva Jorge Onieva, Marti-Fuster Berta, de la Puente María Pedrero, José Estépar Raúl San
机构信息
Applied Chest Imaging Laboratory, Department of Radiology, Brigham and Women's Hospital, Harvard Medical School, Boston, MA, USA.
出版信息
Image Anal Mov Organ Breast Thorac Images (2018). 2018 Sep;11040:284-294. doi: 10.1007/978-3-030-00946-5_28. Epub 2018 Sep 12.
Abstract
Image registration is a well-known problem in the field of medical imaging. In this paper, we focus on the registration of chest inspiratory and expiratory computed tomography (CT) scans from the same patient. Our method recovers the diffeomorphic elastic displacement vector field (DVF) by jointly regressing the direct and the inverse transformation. Our architecture is based on the RegNet network but we implement a reinforced learning strategy that can accommodate a large training dataset. Our results show that our method performs with a lower estimation error for the same number of epochs than the RegNet approach.
摘要
图像配准是医学成像领域一个广为人知的问题。在本文中，我们专注于同一患者的胸部吸气和呼气计算机断层扫描（CT）的配准。我们的方法通过联合回归正向和反向变换来恢复微分同胚弹性位移矢量场（DVF）。我们的架构基于RegNet网络，但我们实施了一种强化学习策略，该策略可以适应大型训练数据集。我们的结果表明，在相同的训练轮数下，我们的方法比RegNet方法具有更低的估计误差。
相似文献
1
Diffeomorphic Lung Registration Using Deep CNNs and Reinforced Learning.使用深度卷积神经网络和强化学习的肺脏微分同胚配准
Image Anal Mov Organ Breast Thorac Images (2018). 2018 Sep;11040:284-294. doi: 10.1007/978-3-030-00946-5_28. Epub 2018 Sep 12.
2
LDVoxelMorph: A precise loss function and cascaded architecture for unsupervised diffeomorphic large displacement registration.LDVoxelMorph：一种用于无监督的仿射大位移配准的精确损失函数和级联架构。
Med Phys. 2022 Apr;49(4):2427-2441. doi: 10.1002/mp.15515. Epub 2022 Feb 17.
3
An adaptive motion regularization technique to support sliding motion in deformable image registration.一种自适应运动正则化技术，用于支持变形图像配准中的滑动运动。
Med Phys. 2018 Feb;45(2):735-747. doi: 10.1002/mp.12734. Epub 2018 Jan 12.
4
The same modality medical image registration with large deformation and clinical application based on adaptive diffeomorphic multi-resolution demons.基于自适应微分同胚多分辨率恶魔算法的大变形同模态医学图像配准及其临床应用
BMC Med Imaging. 2018 Aug 9;18(1):21. doi: 10.1186/s12880-018-0267-3.
5
A fast and scalable method for quality assurance of deformable image registration on lung CT scans using convolutional neural networks.使用卷积神经网络快速且可扩展的肺 CT 扫描图像配准质量保证方法。
Med Phys. 2020 Jan;47(1):99-109. doi: 10.1002/mp.13890. Epub 2019 Nov 19.
6
MO-F-BRA-04: Voxel-Based Statistical Analysis of Deformable Image Registration Error via a Finite Element Method.MO-F-BRA-04：基于体素的有限元法可变形图像配准误差统计分析
Med Phys. 2012 Jun;39(6Part21):3875. doi: 10.1118/1.4735823.
7
Adaptive Diffeomorphic Multiresolution Demons and Their Application to Same Modality Medical Image Registration with Large Deformation.自适应微分同胚多分辨率恶魔算法及其在大变形同模态医学图像配准中的应用
Int J Biomed Imaging. 2018 May 16;2018:7314612. doi: 10.1155/2018/7314612. eCollection 2018.
8
Combining MRF-based deformable registration and deep binary 3D-CNN descriptors for large lung motion estimation in COPD patients.基于 MRF 的可变形配准和深度学习的二进制 3D-CNN 描述符相结合，用于 COPD 患者的大肺运动估计。
Int J Comput Assist Radiol Surg. 2019 Jan;14(1):43-52. doi: 10.1007/s11548-018-1888-2. Epub 2018 Nov 14.
9
A neural network approach for fast, automated quantification of DIR performance.一种用于快速、自动量化 DIR 性能的神经网络方法。
Med Phys. 2017 Aug;44(8):4126-4138. doi: 10.1002/mp.12321. Epub 2017 Jul 17.
10
A deep learning framework for automatic detection of arbitrarily shaped fiducial markers in intrafraction fluoroscopic images.一种用于在分次透视图像中自动检测任意形状基准标记的深度学习框架。
Med Phys. 2019 May;46(5):2286-2297. doi: 10.1002/mp.13519. Epub 2019 Apr 15.
引用本文的文献
1
Artificial general intelligence for radiation oncology.用于放射肿瘤学的通用人工智能。
Meta Radiol. 2023 Nov;1(3). doi: 10.1016/j.metrad.2023.100045. Epub 2023 Nov 24.
2
A review of deep learning-based deformable medical image registration.基于深度学习的可变形医学图像配准综述。
Front Oncol. 2022 Dec 7;12:1047215. doi: 10.3389/fonc.2022.1047215. eCollection 2022.
3
Artificial Intelligence in Radiation Therapy.放射治疗中的人工智能
IEEE Trans Radiat Plasma Med Sci. 2022 Feb;6(2):158-181. doi: 10.1109/TRPMS.2021.3107454. Epub 2021 Aug 24.
4
CNN-based Deformable Registration Facilitates Fast and Accurate Air Trapping Measurements at Inspiratory and Expiratory CT.基于卷积神经网络的可变形配准有助于在吸气和呼气CT上快速准确地测量空气潴留。
Radiol Artif Intell. 2021 Nov 10;4(1):e210211. doi: 10.1148/ryai.2021210211. eCollection 2022 Jan.
5
Artificial intelligence in functional imaging of the lung.人工智能在肺部功能成像中的应用。
Br J Radiol. 2022 Apr 1;95(1132):20210527. doi: 10.1259/bjr.20210527. Epub 2021 Dec 10.
6
A review of deep learning-based three-dimensional medical image registration methods.基于深度学习的三维医学图像配准方法综述。
Quant Imaging Med Surg. 2021 Dec;11(12):4895-4916. doi: 10.21037/qims-21-175.
7
Deep learning in medical image registration: a review.深度学习在医学图像配准中的应用：综述。
Phys Med Biol. 2020 Oct 22;65(20):20TR01. doi: 10.1088/1361-6560/ab843e.
本文引用的文献
1
A survey on deep learning in medical image analysis.深度学习在医学图像分析中的应用研究综述。
Med Image Anal. 2017 Dec;42:60-88. doi: 10.1016/j.media.2017.07.005. Epub 2017 Jul 26.
2
A CNN Regression Approach for Real-Time 2D/3D Registration.一种用于实时 2D/3D 配准的 CNN 回归方法。
IEEE Trans Med Imaging. 2016 May;35(5):1352-1363. doi: 10.1109/TMI.2016.2521800. Epub 2016 Jan 26.
3
A reference dataset for deformable image registration spatial accuracy evaluation using the COPDgene study archive.使用 COPDgene 研究档案进行可变形图像配准空间准确性评估的参考数据集。
Phys Med Biol. 2013 May 7;58(9):2861-77. doi: 10.1088/0031-9155/58/9/2861. Epub 2013 Apr 10.
4
Evaluation of registration methods on thoracic CT: the EMPIRE10 challenge.胸部 CT 配准方法评估：EMPIRE10 挑战赛。
IEEE Trans Med Imaging. 2011 Nov;30(11):1901-20. doi: 10.1109/TMI.2011.2158349. Epub 2011 May 31.
5
Lung extraction, lobe segmentation and hierarchical region assessment for quantitative analysis on high resolution computed tomography images.用于高分辨率计算机断层扫描图像定量分析的肺提取、肺叶分割和分层区域评估
Med Image Comput Comput Assist Interv. 2009;12(Pt 2):690-8. doi: 10.1007/978-3-642-04271-3_84.
6
Genetic epidemiology of COPD (COPDGene) study design.COPD（COPDGene）遗传流行病学研究设计。
COPD. 2010 Feb;7(1):32-43. doi: 10.3109/15412550903499522.
7
Symmetric diffeomorphic image registration with cross-correlation: evaluating automated labeling of elderly and neurodegenerative brain.基于互相关的对称微分同胚图像配准：评估老年人和神经退行性脑部的自动标记
Med Image Anal. 2008 Feb;12(1):26-41. doi: 10.1016/j.media.2007.06.004. Epub 2007 Jun 23.
8
Nonrigid registration using free-form deformations: application to breast MR images.基于自由形式变形的非刚性配准：在乳腺磁共振图像中的应用。
IEEE Trans Med Imaging. 1999 Aug;18(8):712-21. doi: 10.1109/42.796284.
Zotero 插件