DAGAN: Deep De-Aliasing Generative Adversarial Networks for Fast Compressed Sensing MRI Reconstruction

被引:795
作者
Yang, Guang [1 ]
Yu, Simiao [2 ]
Dong, Hao [2 ]
Slabaugh, Greg [3 ]
Dragotti, Pier Luigi [4 ]
Ye, Xujiong [5 ]
Liu, Fangde [2 ]
Arridge, Simon [6 ]
Keegan, Jennifer [1 ]
Guo, Yike [2 ]
Firmin, David [1 ]
机构
[1] Imperial Coll London, Natl Heart & Lung Inst, London SW3 6NP, England
[2] Imperial Coll London, Data Sci Inst, London SW7 2AZ, England
[3] City Univ London, Dept Comp Sci, London EC1V 0HB, England
[4] Imperial Coll London, EEE Dept, London SW3 6NP, England
[5] Univ Lincoln, Sch Comp Sci, Lincoln LN6 7TS, England
[6] UCL, CMIC, London WC1E 6BT, England
基金
英国工程与自然科学研究理事会;
关键词
Compressed sensing; magnetic resonance imaging (MRI); fast MRI; deep learning; generative adversarial networks (GAN); de-aliasing; inverse problems; IMAGE-RECONSTRUCTION; ALGORITHMS; TIME; SPARSITY;
D O I
10.1109/TMI.2017.2785879
中图分类号
TP39 [计算机的应用];
学科分类号
081203 ; 0835 ;
摘要
Compressed sensing magnetic resonance imaging (CS-MRI) enables fast acquisition, which is highly desirable for numerous clinical applications. This can not only reduce the scanning cost and ease patient burden, but also potentially reduce motion artefacts and the effect of contrast washout, thus yielding better image quality. Different from parallel imaging-based fast MRI, which utilizes multiple coils to simultaneously receive MR signals, CS-MRI breaks the Nyquist-Shannon sampling barrier to reconstruct MRI images with much less required raw data. This paper provides a deep learning-based strategy for reconstruction of CS-MRI, and bridges a substantial gap between conventional non-learning methods working only on data from a single image, and prior knowledge from large training data sets. In particular, a novel conditional Generative Adversarial Networks-based model (DAGAN)-based model is proposed to reconstruct CS-MRI. In our DAGAN architecture, we have designed a refinement learning method to stabilize our U-Net based generator, which provides an end-to-end network to reduce aliasing artefacts. To better preserve texture and edges in the reconstruction, we have coupled the adversarial loss with an innovative content loss. In addition, we incorporate frequency-domain information to enforce similarity in both the image and frequency domains. We have performed comprehensive comparison studies with both conventional CS-MRI reconstruction methods and newly investigated deep learning approaches. Compared with these methods, our DAGAN method provides superior reconstruction with preserved perceptual image details. Furthermore, each image is reconstructed in about 5 ms, which is suitable for real-time processing.
引用
收藏
页码:1310 / 1321
页数:12
相关论文
共 50 条
[41]   Deep Natural Image Reconstruction from Human Brain Activity Based on Conditional Progressively Growing Generative Adversarial Networks [J].
Huang, Wei ;
Yan, Hongmei ;
Wang, Chong ;
Yang, Xiaoqing ;
Li, Jiyi ;
Zuo, Zhentao ;
Zhang, Jiang ;
Chen, Huafu .
NEUROSCIENCE BULLETIN, 2021, 37 (03) :369-379
[42]   Enhancing plantar pressure distribution reconstruction with conditional generative adversarial networks from multi-region foot pressure sensing [J].
Chan, Hsiao-Lung ;
Liang, Jing-Rong ;
Chang, Ya-Ju ;
Chen, Rou-Shayn ;
Kuo, Cheng-Chung ;
Hsu, Wen-Yen ;
Tsai, Meng-Tsan .
BIOMEDICAL SIGNAL PROCESSING AND CONTROL, 2025, 100
[43]   Reference-Driven Compressed Sensing MR Image Reconstruction Using Deep Convolutional Neural Networks without Pre-Training [J].
Zhao, Di ;
Zhao, Feng ;
Gan, Yongjin .
SENSORS, 2020, 20 (01)
[44]   Generative adversarial network-based post-processed image super-resolution technology for accelerating brain MRI: comparison with compressed sensing [J].
Ueki, Wataru ;
Nishii, Tatsuya ;
Umehara, Kensuke ;
Ota, Junko ;
Higuchi, Satoshi ;
Ohta, Yasutoshi ;
Nagai, Yasuhiro ;
Murakawa, Keizo ;
Ishida, Takayuki ;
Fukuda, Tetsuya .
ACTA RADIOLOGICA, 2023, 64 (01) :336-345
[45]   Reconstruction of 3D knee MRI using deep learning and compressed sensing: a validation study on healthy volunteers [J].
Dratsch, Thomas ;
Zaeske, Charlotte ;
Siedek, Florian ;
Rauen, Philip ;
Hokamp, Nils Grosse ;
Sonnabend, Kristina ;
Maintz, David ;
Bratke, Grischa ;
Iuga, Andra .
EUROPEAN RADIOLOGY EXPERIMENTAL, 2024, 8 (01)
[46]   Real-Time Image Reconstruction for Low-Dose CT using Deep Convolutional Generative Adversarial Networks (GANs) [J].
Choi, Kihwan ;
Kim, Sung Won ;
Lim, Joon Seok .
MEDICAL IMAGING 2018: PHYSICS OF MEDICAL IMAGING, 2018, 10573
[47]   Compressed sensing and deep learning reconstruction for women's pelvic MRI denoising: Utility for improving image quality and examination time in routine clinical practice [J].
Ueda, Takahiro ;
Ohno, Yoshiharu ;
Yamamoto, Kaori ;
Iwase, Akiyoshi ;
Fukuba, Takashi ;
Hanamatsu, Satomu ;
Obama, Yuki ;
Ikeda, Hirotaka ;
Ikedo, Masato ;
Yui, Masao ;
Murayama, Kazuhiro ;
Toyama, Hiroshi .
EUROPEAN JOURNAL OF RADIOLOGY, 2021, 134
[48]   Efficacy of compressed sensing and deep learning reconstruction for adult female pelvic MRI at 1.5 T (vol 8, 103, 2024) [J].
Ueda, Takahiro ;
Yamamoto, Kaori ;
Yazawa, Natsuka ;
Tozawa, Ikki ;
Ikedo, Masato ;
Yui, Masao ;
Nagata, Hiroyuki ;
Nomura, Masahiko ;
Ozawa, Yoshiyuki ;
Ohno, Yoshiharu .
EUROPEAN RADIOLOGY EXPERIMENTAL, 2024, 8 (01)
[49]   Towards High-Quality MRI Reconstruction With Anisotropic Diffusion-Assisted Generative Adversarial Networks and Its Multi-Modal Images Extension [J].
Luo, Yuyang ;
Wu, Gengshen ;
Liu, Yi ;
Liu, Wenjian ;
Han, Jungong .
IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS, 2025, 29 (05) :3098-3111
[50]   Evaluation of high-resolution pituitary dynamic contrast-enhanced MRI using deep learning-based compressed sensing and super-resolution reconstruction [J].
Zhang, Meng ;
Xia, Chunchao ;
Tang, Jing ;
Yao, Li ;
Hu, Na ;
Li, Jiaqi ;
Peng, Wanlin ;
Hu, Sixian ;
Ye, Zheng ;
Zhang, Xiaoyong ;
Huang, Jin ;
Li, Zhenlin .
EUROPEAN RADIOLOGY, 2025,