深度磁共振图像重建：逆问题与神经网络相遇
Deep Magnetic Resonance Image Reconstruction: Inverse Problems Meet Neural Networks.
作者信息
Liang Dong, Cheng Jing, Ke Ziwen, Ying Leslie
机构信息
Paul C. Lauterbur Research Center for Biomedical Imaging.
Research Center for Medical AI, Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, in Shenzhen, Guangdong, China.
出版信息
IEEE Signal Process Mag. 2020 Jan;37(1):141-151. doi: 10.1109/MSP.2019.2950557. Epub 2020 Jan 20.
Abstract
Image reconstruction from undersampled k-space data has been playing an important role in fast MRI. Recently, deep learning has demonstrated tremendous success in various fields and has also shown potential in significantly accelerating MRI reconstruction with fewer measurements. This article provides an overview of the deep learning-based image reconstruction methods for MRI. Two types of deep learning-based approaches are reviewed: those based on unrolled algorithms and those which are not. The main structure of both approaches are explained, respectively. Several signal processing issues for maximizing the potential of deep reconstruction in fast MRI are discussed. The discussion may facilitate further development of the networks and the analysis of performance from a theoretical point of view.
摘要
从欠采样k空间数据进行图像重建在快速磁共振成像（MRI）中一直发挥着重要作用。近年来，深度学习在各个领域都取得了巨大成功，并且在以更少测量次数显著加速MRI重建方面也显示出潜力。本文概述了基于深度学习的MRI图像重建方法。回顾了两种基于深度学习的方法：基于展开算法的方法和非基于展开算法的方法。分别解释了这两种方法的主要结构。讨论了在快速MRI中最大化深度重建潜力的几个信号处理问题。该讨论可能有助于网络的进一步发展以及从理论角度对性能进行分析。
相似文献
1
Deep Magnetic Resonance Image Reconstruction: Inverse Problems Meet Neural Networks.深度磁共振图像重建：逆问题与神经网络相遇
IEEE Signal Process Mag. 2020 Jan;37(1):141-151. doi: 10.1109/MSP.2019.2950557. Epub 2020 Jan 20.
2
Deep supervised dictionary learning by algorithm unrolling-Application to fast 2D dynamic MR image reconstruction.基于算法展开的深度监督字典学习——在快速二维动态磁共振图像重建中的应用
Med Phys. 2023 May;50(5):2939-2960. doi: 10.1002/mp.16182. Epub 2023 Jan 17.
3
Deep, deep learning with BART.基于 BART 的深度深度学习。
Magn Reson Med. 2023 Feb;89(2):678-693. doi: 10.1002/mrm.29485. Epub 2022 Oct 18.
4
Adaptive convolutional neural networks for accelerating magnetic resonance imaging via k-space data interpolation.基于自适应卷积神经网络的磁共振成像 k 空间数据插值加速方法。
Med Image Anal. 2021 Aug;72:102098. doi: 10.1016/j.media.2021.102098. Epub 2021 May 16.
5
Deep-Learning Methods for Parallel Magnetic Resonance Imaging Reconstruction: A Survey of the Current Approaches, Trends, and Issues.用于并行磁共振成像重建的深度学习方法：当前方法、趋势及问题综述
IEEE Signal Process Mag. 2020 Jan;37(1):128-140. doi: 10.1109/MSP.2019.2950640. Epub 2020 Jan 20.
6
Emerging Trends in Fast MRI Using Deep-Learning Reconstruction on Undersampled k-Space Data: A Systematic Review.基于欠采样k空间数据使用深度学习重建的快速磁共振成像新趋势：一项系统综述
Bioengineering (Basel). 2023 Aug 26;10(9):1012. doi: 10.3390/bioengineering10091012.
7
A review and experimental evaluation of deep learning methods for MRI reconstruction.磁共振成像重建深度学习方法的综述与实验评估
J Mach Learn Biomed Imaging. 2022 Mar;1. Epub 2022 Mar 11.
8
Radial magnetic resonance image reconstruction with a deep unrolled projected fast iterative soft-thresholding network.基于深度解卷积投影快速迭代软阈值网络的磁共振径向重建。
Comput Biol Med. 2024 Jan;168:107707. doi: 10.1016/j.compbiomed.2023.107707. Epub 2023 Nov 19.
9
Scan-specific robust artificial-neural-networks for k-space interpolation (RAKI) reconstruction: Database-free deep learning for fast imaging.基于扫描特异性稳健人工神经网络的 K 空间插值（RAKI）重建：无数据库的深度学习实现快速成像。
Magn Reson Med. 2019 Jan;81(1):439-453. doi: 10.1002/mrm.27420. Epub 2018 Sep 18.
10
Physics-Driven Deep Learning for Computational Magnetic Resonance Imaging: Combining physics and machine learning for improved medical imaging.用于计算磁共振成像的物理驱动深度学习：结合物理与机器学习以改善医学成像。
IEEE Signal Process Mag. 2023 Jan;40(1):98-114. doi: 10.1109/msp.2022.3215288. Epub 2023 Jan 2.
引用本文的文献
1
A dynamic reconstruction and motion estimation framework for cardiorespiratory motion-resolved real-time volumetric MR imaging (DREME-MR).用于心肺运动分辨实时容积磁共振成像的动态重建与运动估计框架（DREME-MR）。
Phys Med Biol. 2025 Aug 8. doi: 10.1088/1361-6560/adf9b9.
2
Self-supervised learning for MRI reconstruction: a review and new perspective.用于磁共振成像重建的自监督学习：综述与新视角
MAGMA. 2025 Jun 26. doi: 10.1007/s10334-025-01274-y.
3
Super-resolution sodium MRI of human gliomas at 3T using physics-based generative artificial intelligence.使用基于物理的生成式人工智能在3T条件下对人脑胶质瘤进行超分辨率钠磁共振成像。
J Neurooncol. 2025 Jun 3. doi: 10.1007/s11060-025-05094-x.
4
Rosette Trajectory MRI Reconstruction with Vision Transformers.基于视觉Transformer的玫瑰花结轨迹MRI重建
Tomography. 2025 Apr 1;11(4):41. doi: 10.3390/tomography11040041.
5
6
Complex-valued neural networks to speed-up MR thermometry during hyperthermia using Fourier PD and PDUNet.使用傅里叶相位差和相位差U-Net的复值神经网络加速热疗期间的磁共振测温
Sci Rep. 2025 Apr 6;15(1):11765. doi: 10.1038/s41598-025-96071-x.
7
MR Spatiospectral Reconstruction Integrating Subspace Modeling and Self-Supervised Spatiotemporal Denoising.集成子空间建模和自监督时空去噪的磁共振空间光谱重建
IEEE Trans Med Imaging. 2025 Mar 28;PP. doi: 10.1109/TMI.2025.3555928.
8
Metrology for MRI: the field you've never heard of.磁共振成像计量学：你从未听说过的领域。
MAGMA. 2025 Mar 19. doi: 10.1007/s10334-025-01238-2.
9
UNCERTAINTY-GUIDED PHYSICS-DRIVEN DEEP LEARNING RECONSTRUCTION VIA CYCLIC MEASUREMENT CONSISTENCY.通过循环测量一致性实现不确定性引导的物理驱动深度学习重建
Proc IEEE Int Conf Acoust Speech Signal Process. 2024 Apr;2024:13441-13445. doi: 10.1109/icassp48485.2024.10447594. Epub 2024 Mar 18.
10
Optimized MR pulse sequence for high-resolution brain 3D-T1ρ mapping with weighted spin-lock acquisitions.用于高分辨率脑部三维T1ρ图谱绘制的加权自旋锁定采集优化磁共振脉冲序列。
Magn Reson Med. 2025 Apr;93(4):1458-1470. doi: 10.1002/mrm.30412. Epub 2024 Dec 22.
本文引用的文献
1
ACCELERATING MAGNETIC RESONANCE IMAGING VIA DEEP LEARNING.通过深度学习加速磁共振成像
Proc IEEE Int Symp Biomed Imaging. 2016 Apr;2016:514-517. doi: 10.1109/ISBI.2016.7493320. Epub 2016 Jun 16.
2
DIMENSION: Dynamic MR imaging with both k-space and spatial prior knowledge obtained via multi-supervised network training.维度：通过多监督网络训练获得的具有 k 空间和空间先验知识的动态磁共振成像。
NMR Biomed. 2022 Apr;35(4):e4131. doi: 10.1002/nbm.4131. Epub 2019 Sep 4.
3
Deep Proximal Unrolling: Algorithmic Framework, Convergence Analysis and Applications.深度近端展开：算法框架、收敛性分析及应用
IEEE Trans Image Process. 2019 May 2. doi: 10.1109/TIP.2019.2913536.
4
MANTIS: Model-Augmented Neural neTwork with Incoherent k-space Sampling for efficient MR parameter mapping.MANTIS：用于高效磁共振参数映射的基于模型增强的神经网络与非相干 k 空间采样。
Magn Reson Med. 2019 Jul;82(1):174-188. doi: 10.1002/mrm.27707. Epub 2019 Mar 12.
5
ADMM-CSNet: A Deep Learning Approach for Image Compressive Sensing.ADMM-CSNet：一种用于图像压缩感知的深度学习方法。
IEEE Trans Pattern Anal Mach Intell. 2020 Mar;42(3):521-538. doi: 10.1109/TPAMI.2018.2883941. Epub 2018 Nov 28.
6
Scan-specific robust artificial-neural-networks for k-space interpolation (RAKI) reconstruction: Database-free deep learning for fast imaging.基于扫描特异性稳健人工神经网络的 K 空间插值（RAKI）重建：无数据库的深度学习实现快速成像。
Magn Reson Med. 2019 Jan;81(1):439-453. doi: 10.1002/mrm.27420. Epub 2018 Sep 18.
7
Real-time cardiovascular MR with spatio-temporal artifact suppression using deep learning-proof of concept in congenital heart disease.基于深度学习的实时心血管磁共振时空伪影抑制技术：在先天性心脏病中的初步研究。
Magn Reson Med. 2019 Feb;81(2):1143-1156. doi: 10.1002/mrm.27480. Epub 2018 Sep 8.
8
MoDL: Model-Based Deep Learning Architecture for Inverse Problems.MoDL：基于模型的深度学习架构用于反问题。
IEEE Trans Med Imaging. 2019 Feb;38(2):394-405. doi: 10.1109/TMI.2018.2865356. Epub 2018 Aug 13.
9
Convolutional Recurrent Neural Networks for Dynamic MR Image Reconstruction.卷积循环神经网络在动态磁共振图像重建中的应用。
IEEE Trans Med Imaging. 2019 Jan;38(1):280-290. doi: 10.1109/TMI.2018.2863670. Epub 2018 Aug 6.
10
Deep Generative Adversarial Neural Networks for Compressive Sensing MRI.用于压缩感知 MRI 的深度生成对抗神经网络。
IEEE Trans Med Imaging. 2019 Jan;38(1):167-179. doi: 10.1109/TMI.2018.2858752. Epub 2018 Jul 23.
Zotero 插件