CT Super-Resolution GAN Constrained by the Identical, Residual, and Cycle Learning Ensemble (GAN-CIRCLE)

Chenyu You; Wenxiang Cong; Michael W Vannier; Punam K Saha; Eric A Hoffman; Ge Wang; Guang Li; Yi Zhang; Xiaoliu Zhang; Hongming Shan; Mengzhou Li; Shenghong Ju; Zhen Zhao; Zhuiyang Zhang

doi:10.1109/TMI.2019.2922960

Back

CT Super-Resolution GAN Constrained by the Identical, Residual, and Cycle Learning Ensemble (GAN-CIRCLE)

Journal article

Open access

Peer reviewed

CT Super-Resolution GAN Constrained by the Identical, Residual, and Cycle Learning Ensemble (GAN-CIRCLE)

Chenyu You, Wenxiang Cong, Michael W Vannier, Punam K Saha, Eric A Hoffman, Ge Wang, Guang Li, Yi Zhang, Xiaoliu Zhang, Hongming Shan, …

IEEE transactions on medical imaging, Vol.39(1), pp.188-203

01/2020

DOI: 10.1109/TMI.2019.2922960

PMID: 31217097

Files and links (1)

url

https://arxiv.org/pdf/1808.04256View

Open Access

Abstract

In this paper, we present a semi-supervised deep learning approach to accurately recover high-resolution (HR) CT images from low-resolution (LR) counterparts. Specifically, with the generative adversarial network (GAN) as the building block, we enforce the cycle-consistency in terms of the Wasserstein distance to establish a nonlinear end-to-end mapping from noisy LR input images to denoised and deblurred HR outputs. We also include the joint constraints in the loss function to facilitate structural preservation. In this process, we incorporate deep convolutional neural network (CNN), residual learning, and network in network techniques for feature extraction and restoration. In contrast to the current trend of increasing network depth and complexity to boost the imaging performance, we apply a parallel 1 × 1 CNN to compress the output of the hidden layer and optimize the number of layers and the number of filters for each convolutional layer. The quantitative and qualitative evaluative results demonstrate that our proposed model is accurate, efficient and robust for super-resolution (SR) image restoration from noisy LR input images. In particular, we validate our composite SR networks on three large-scale CT datasets, and obtain promising results as compared to the other state-of-the-art methods.

Computed Tomography

Noise Reduction

adversarial learning

Computed tomography (CT)

deep learning

Gallium nitride

Generative adversarial networks

Image reconstruction

Image resolution

residual learning

super-resolution

Training

Details

Title: Subtitle: CT Super-Resolution GAN Constrained by the Identical, Residual, and Cycle Learning Ensemble (GAN-CIRCLE)
Creators: Chenyu You - Stanford University
Wenxiang Cong - Rensselaer Polytechnic Institute
Michael W Vannier - University of Chicago
Punam K Saha - University of Iowa
Eric A Hoffman - University of Iowa
Ge Wang - Rensselaer Polytechnic Institute
Guang Li - Rensselaer Polytechnic Institute
Yi Zhang - Sichuan University
Xiaoliu Zhang - University of Iowa
Hongming Shan - Rensselaer Polytechnic Institute
Mengzhou Li - Rensselaer Polytechnic Institute
Shenghong Ju - Southeast University
Zhen Zhao - Southeast University
Zhuiyang Zhang - Wuxi No.2 People's Hospital
Resource Type: Journal article
Publication Details: IEEE transactions on medical imaging, Vol.39(1), pp.188-203
DOI: 10.1109/TMI.2019.2922960
PMID: 31217097
NLM abbreviation: IEEE Trans Med Imaging
ISSN: 0278-0062
eISSN: 1558-254X
Publisher: Institute of Electrical and Electronics Engineers
Language: English
Date published: 01/2020
Academic Unit: Roy J. Carver Department of Biomedical Engineering; Radiology; Electrical and Computer Engineering; Internal Medicine
Record Identifier: 9984197259202771

Metrics

19 Record Views

384 Times Cited - Web of Science