EMnet: An unrolled deep neural network for PET image reconstruction

Kuang Gong; Dufan Wu; Kyungsang Kim; Jaewon Yang; Georges El Fakhri; Youngho Seo; Quanzheng Li

doi:10.1117/12.2513096

EMnet: An unrolled deep neural network for PET image reconstruction

Kuang Gong, Dufan Wu, Kyungsang Kim, Jaewon Yang, Georges El Fakhri, Youngho Seo, Quanzheng Li

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

19 Scopus citations

Abstract

PET image reconstruction is challenging due to the ill-poseness of the inverse problem and limited number of detected photons. Recently deep neural networks have been widely applied to medical imaging denoising applications. In this work, based on the expectation maximization (EM) algorithm, we propose an unrolled neural network framework for PET image reconstruction, named EMnet. An innovative feature of the proposed framework is that the deep neural network is combined with the EM update steps in a whole graph. Thus data consistency can act as a constraint during network training. Both simulation data and real data are used to evaluate the proposed method. Quantification results show that our proposed EMnet method can outperform the neural network denoising and Gaussian denoising methods.

Original language	English (US)
Title of host publication	Medical Imaging 2019
Subtitle of host publication	Physics of Medical Imaging
Editors	Taly Gilat Schmidt, Guang-Hong Chen, Hilde Bosmans
Publisher	SPIE
ISBN (Electronic)	9781510625433
DOIs	https://doi.org/10.1117/12.2513096
State	Published - 2019
Externally published	Yes
Event	Medical Imaging 2019: Physics of Medical Imaging - San Diego, United States Duration: Feb 17 2019 → Feb 20 2019

Publication series

Name	Progress in Biomedical Optics and Imaging - Proceedings of SPIE
Volume	10948
ISSN (Print)	1605-7422

Conference

Conference	Medical Imaging 2019: Physics of Medical Imaging
Country/Territory	United States
City	San Diego
Period	2/17/19 → 2/20/19

Keywords

Image reconstruction
PET
Unrolled neural network

ASJC Scopus subject areas

Electronic, Optical and Magnetic Materials
Atomic and Molecular Physics, and Optics
Radiology Nuclear Medicine and imaging
Biomaterials

Access to Document

10.1117/12.2513096

Cite this

Gong, K., Wu, D., Kim, K., Yang, J., El Fakhri, G., Seo, Y., & Li, Q. (2019). EMnet: An unrolled deep neural network for PET image reconstruction. In T. G. Schmidt, G.-H. Chen, & H. Bosmans (Eds.), Medical Imaging 2019: Physics of Medical Imaging Article 1094853 (Progress in Biomedical Optics and Imaging - Proceedings of SPIE; Vol. 10948). SPIE. https://doi.org/10.1117/12.2513096

EMnet: An unrolled deep neural network for PET image reconstruction. / Gong, Kuang; Wu, Dufan; Kim, Kyungsang et al.
Medical Imaging 2019: Physics of Medical Imaging. ed. / Taly Gilat Schmidt; Guang-Hong Chen; Hilde Bosmans. SPIE, 2019. 1094853 (Progress in Biomedical Optics and Imaging - Proceedings of SPIE; Vol. 10948).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Gong, K, Wu, D, Kim, K, Yang, J, El Fakhri, G, Seo, Y & Li, Q 2019, EMnet: An unrolled deep neural network for PET image reconstruction. in TG Schmidt, G-H Chen & H Bosmans (eds), Medical Imaging 2019: Physics of Medical Imaging., 1094853, Progress in Biomedical Optics and Imaging - Proceedings of SPIE, vol. 10948, SPIE, Medical Imaging 2019: Physics of Medical Imaging, San Diego, United States, 2/17/19. https://doi.org/10.1117/12.2513096

@inproceedings{63ddd278ddce45378929385637541d4f,

title = "EMnet: An unrolled deep neural network for PET image reconstruction",

abstract = "PET image reconstruction is challenging due to the ill-poseness of the inverse problem and limited number of detected photons. Recently deep neural networks have been widely applied to medical imaging denoising applications. In this work, based on the expectation maximization (EM) algorithm, we propose an unrolled neural network framework for PET image reconstruction, named EMnet. An innovative feature of the proposed framework is that the deep neural network is combined with the EM update steps in a whole graph. Thus data consistency can act as a constraint during network training. Both simulation data and real data are used to evaluate the proposed method. Quantification results show that our proposed EMnet method can outperform the neural network denoising and Gaussian denoising methods.",

keywords = "Image reconstruction, PET, Unrolled neural network",

author = "Kuang Gong and Dufan Wu and Kyungsang Kim and Jaewon Yang and {El Fakhri}, Georges and Youngho Seo and Quanzheng Li",

note = "Publisher Copyright: {\textcopyright} SPIE. Downloading of the abstract is permitted for personal use only.; Medical Imaging 2019: Physics of Medical Imaging ; Conference date: 17-02-2019 Through 20-02-2019",

year = "2019",

doi = "10.1117/12.2513096",

language = "English (US)",

series = "Progress in Biomedical Optics and Imaging - Proceedings of SPIE",

publisher = "SPIE",

editor = "Schmidt, {Taly Gilat} and Guang-Hong Chen and Hilde Bosmans",

booktitle = "Medical Imaging 2019",

}

TY - GEN

T1 - EMnet

T2 - Medical Imaging 2019: Physics of Medical Imaging

AU - Gong, Kuang

AU - Wu, Dufan

AU - Kim, Kyungsang

AU - Yang, Jaewon

AU - El Fakhri, Georges

AU - Seo, Youngho

AU - Li, Quanzheng

PY - 2019

Y1 - 2019

N2 - PET image reconstruction is challenging due to the ill-poseness of the inverse problem and limited number of detected photons. Recently deep neural networks have been widely applied to medical imaging denoising applications. In this work, based on the expectation maximization (EM) algorithm, we propose an unrolled neural network framework for PET image reconstruction, named EMnet. An innovative feature of the proposed framework is that the deep neural network is combined with the EM update steps in a whole graph. Thus data consistency can act as a constraint during network training. Both simulation data and real data are used to evaluate the proposed method. Quantification results show that our proposed EMnet method can outperform the neural network denoising and Gaussian denoising methods.

AB - PET image reconstruction is challenging due to the ill-poseness of the inverse problem and limited number of detected photons. Recently deep neural networks have been widely applied to medical imaging denoising applications. In this work, based on the expectation maximization (EM) algorithm, we propose an unrolled neural network framework for PET image reconstruction, named EMnet. An innovative feature of the proposed framework is that the deep neural network is combined with the EM update steps in a whole graph. Thus data consistency can act as a constraint during network training. Both simulation data and real data are used to evaluate the proposed method. Quantification results show that our proposed EMnet method can outperform the neural network denoising and Gaussian denoising methods.

KW - Image reconstruction

KW - PET

KW - Unrolled neural network

UR - http://www.scopus.com/inward/record.url?scp=85068357873&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85068357873&partnerID=8YFLogxK

U2 - 10.1117/12.2513096

DO - 10.1117/12.2513096

M3 - Conference contribution

AN - SCOPUS:85068357873

T3 - Progress in Biomedical Optics and Imaging - Proceedings of SPIE

BT - Medical Imaging 2019

A2 - Schmidt, Taly Gilat

A2 - Chen, Guang-Hong

A2 - Bosmans, Hilde

PB - SPIE

Y2 - 17 February 2019 through 20 February 2019

ER -

EMnet: An unrolled deep neural network for PET image reconstruction

Abstract

Publication series

Conference

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this