Zhang, L, Pereañez, M, Bowles, C et al. (4 more authors) (2019) Unsupervised Standard Plane Synthesis in Population Cine MRI via Cycle-Consistent Adversarial Networks. In: Lecture Notes in Computer Science. Medical Image Computing and Computer Assisted Intervention – MICCAI 2019, 13-17 Oct 2019, Shenzhen, China. Springer Verlag , pp. 660-668. ISBN 978-3-030-32244-1
Abstract
In clinical studies or population imaging settings, cardiac magnetic resonance (CMR) images may suffer from artifacts due to variability in the breath-hold position adopted by the patient during the scan. Consistent orientation of image planes with respect to the cardiac ventricles in CMR sequences forms a crucial step in the assessment of cardiac function via parameters such as the Ejection Fraction (EF) and Cardiac Output (CO) of both ventricles, which are the most immediate indicators of normal/abnormal cardiac function. In this paper, we present a novel unsupervised approach for the realistic transformation of acquired CMR images to a standard orientation using Cycle-Consistent Adversarial Networks (Cycle-GANs). We tackle this challenge by splitting the problem into two principal subtasks. First, we consider a bidirectional generator mapping between the re-oriented image and the original, hence allowing direct comparison to the input image without the need to resort to paired training data. Second, we devise a novel loss function incorporating intensity and orientation terms, and aims to produce images of high perceptual quality. Extensive experiments conducted on the CMR images in the UK Biobank dataset demonstrate that the images rendered by our model can improve the accuracy of the image derived cardiac parameters.
Metadata
Item Type: | Proceedings Paper |
---|---|
Authors/Creators: |
|
Copyright, Publisher and Additional Information: | © Springer Nature Switzerland AG 2019. This is an author produced version of a conference paper published in Lecture Notes in Computer Science. Uploaded in accordance with the publisher's self-archiving policy. |
Keywords: | Deep learning; Cycle-Consistent Adversarial Networks; Cardiac orientation; Ventricular volume; MRI; Population imaging |
Dates: |
|
Institution: | The University of Leeds |
Academic Units: | The University of Leeds > Faculty of Engineering & Physical Sciences (Leeds) > School of Computing (Leeds) |
Depositing User: | Symplectic Publications |
Date Deposited: | 12 May 2020 15:43 |
Last Modified: | 13 Jun 2020 10:56 |
Status: | Published |
Publisher: | Springer Verlag |
Identification Number: | 10.1007/978-3-030-32245-8_73 |
Open Archives Initiative ID (OAI ID): | oai:eprints.whiterose.ac.uk:160599 |