Variational adversarial networks for accelerated MR image reconstruction

Kerstin Hammernik; Erich Kobler; Thomas Pock; Michael P Recht; Daniel K  Sodickson; Florian Knoll

Variational adversarial networks for accelerated MR image reconstruction

Kerstin Hammernik, Erich Kobler, Thomas Pock, Michael P Recht, Daniel K Sodickson, Florian Knoll

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

Abstract

Inspired by variational networks and adversarial training, we introduce variational adversarial networks for accelerated MR image reconstruction to overcome typical limitations of using simple image quality measures as loss functions for training. While simple loss functions, such as mean-squared-error and structural similarity index, result in low resolution and blurry images, we show that the proposed variational adversarial network leads to sharper images and preserves fine details for clinical low and high SNR patient data.

Original language	English
Title of host publication	ISMRM-ESMRMB 2018 Abstracts
Subtitle of host publication	International Society for Magnetic Resonance in Medicine (ISMRM)
Place of Publication	Berkeley, Kalifornien (USA
Pages	1091-1091
Number of pages	1
Publication status	Published - Jun 2018
Externally published	Yes
Event	Joint Annual Meeting ISMRM-ESMRMB 2018 - Paris, France Duration: 16 Jun 2018 → 21 Jun 2018

Conference

Conference	Joint Annual Meeting ISMRM-ESMRMB 2018
Country/Territory	France
City	Paris
Period	16.06.2018 → 21.06.2018

Fields of science

102037 Visualisation

Cite this

@inproceedings{202bf2722e6e469b83ca68a3e5b0b7dd,

title = "Variational adversarial networks for accelerated MR image reconstruction",

abstract = "Inspired by variational networks and adversarial training, we introduce variational adversarial networks for accelerated MR image reconstruction to overcome typical limitations of using simple image quality measures as loss functions for training. While simple loss functions, such as mean-squared-error and structural similarity index, result in low resolution and blurry images, we show that the proposed variational adversarial network leads to sharper images and preserves fine details for clinical low and high SNR patient data. ",

author = "Kerstin Hammernik and Erich Kobler and Thomas Pock and Recht, \{Michael P\} and Sodickson, \{Daniel K\} and Florian Knoll",

year = "2018",

month = jun,

language = "English",

pages = "1091--1091",

booktitle = "ISMRM-ESMRMB 2018 Abstracts",

note = "Joint Annual Meeting ISMRM-ESMRMB 2018 ; Conference date: 16-06-2018 Through 21-06-2018",

}

TY - GEN

T1 - Variational adversarial networks for accelerated MR image reconstruction

AU - Hammernik, Kerstin

AU - Kobler, Erich

AU - Pock, Thomas

AU - Recht, Michael P

AU - Sodickson, Daniel K

AU - Knoll, Florian

PY - 2018/6

Y1 - 2018/6

N2 - Inspired by variational networks and adversarial training, we introduce variational adversarial networks for accelerated MR image reconstruction to overcome typical limitations of using simple image quality measures as loss functions for training. While simple loss functions, such as mean-squared-error and structural similarity index, result in low resolution and blurry images, we show that the proposed variational adversarial network leads to sharper images and preserves fine details for clinical low and high SNR patient data.

AB - Inspired by variational networks and adversarial training, we introduce variational adversarial networks for accelerated MR image reconstruction to overcome typical limitations of using simple image quality measures as loss functions for training. While simple loss functions, such as mean-squared-error and structural similarity index, result in low resolution and blurry images, we show that the proposed variational adversarial network leads to sharper images and preserves fine details for clinical low and high SNR patient data.

UR - https://archive.ismrm.org/2018/1091.html

M3 - Conference proceedings

SP - 1091

EP - 1091

BT - ISMRM-ESMRMB 2018 Abstracts

CY - Berkeley, Kalifornien (USA

T2 - Joint Annual Meeting ISMRM-ESMRMB 2018

Y2 - 16 June 2018 through 21 June 2018

ER -

Variational adversarial networks for accelerated MR image reconstruction

Abstract

Conference

Fields of science

Other files and links

Cite this