Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2019 Sep;82(3):1177-1186.
doi: 10.1002/mrm.27786. Epub 2019 Apr 29.

Separation of water and fat signal in whole-body gradient echo scans using convolutional neural networks

Jonathan Andersson  1 , Håkan Ahlström  1   2 , Joel Kullberg  1   2
Affiliations

Separation of water and fat signal in whole-body gradient echo scans using convolutional neural networks

Jonathan Andersson et al. Magn Reson Med. 2019 Sep.

Abstract

Purpose: To perform and evaluate water-fat signal separation of whole-body gradient echo scans using convolutional neural networks.

Methods: Whole-body gradient echo scans of 240 subjects, each consisting of 5 bipolar echoes, were used. Reference fat fraction maps were created using a conventional method. Convolutional neural networks, more specifically 2D U-nets, were trained using 5-fold cross-validation with 1 or several echoes as input, using the squared difference between the output and the reference fat fraction maps as the loss function. The outputs of the networks were assessed by the loss function, measured liver fat fractions, and visually. Training was performed using a graphics processing unit (GPU). Inference was performed using the GPU as well as a central processing unit (CPU).

Results: The loss curves indicated convergence, and the final loss of the validation data decreased when using more echoes as input. The liver fat fractions could be estimated using only 1 echo, but results were improved by use of more echoes. Visual assessment found the quality of the outputs of the networks to be similar to the reference even when using only 1 echo, with slight improvements when using more echoes. Training a network took at most 28.6 h. Inference time of a whole-body scan took at most 3.7 s using the GPU and 5.8 min using the CPU.

Conclusion: It is possible to perform water-fat signal separation of whole-body gradient echo scans using convolutional neural networks. Separation was possible using only 1 echo, although using more echoes improved the results.

Keywords: Dixon; convolutional neural network; deep learning; magnetic resonance imaging; neural network; water-fat separation.

PubMed Disclaimer

Figures

Figure 1
Figure 1
A visual representation of 1 of the networks used in this manuscript, with an axial slice containing the real and the imaginary parts of the first echo as input and the corresponding fat fraction map as output. The cyan boxes represent feature maps. The white boxes represent feature maps that have been transferred by the skip connections. The horizontal numbers represent the number of features in a layer and the vertical numbers represent the number of elements per feature of the layer
Figure 2
Figure 2
Curves showing loss per foreground voxel for the networks using echoes of both polarities. Dashed lines are used for training data, solid lines for validation data
Figure 3
Figure 3
Liver fat fractions estimated by the neural networks using echoes of both polarities plotted against the reference fat fraction. A, All data points shown. In case a normal liver was misclassified as fatty or vice versa, the corresponding data point is found in the red shaded area. B, Zoom‐in on the area with the cases were the liver fat fractions were correctly classified as normal, where most cases can be found
Figure 4
Figure 4
Axial fat fraction maps of the abdomen of a subject with a fatty liver (reference fat fraction 12.28%). Background has been removed from all images for clarity. The images are in grayscale with range 0% FF to 100% FF. (A–E) Results using neural networks. A, Using the 1st echo, B, using the 1st and the 2nd echoes, C, using the 1st through the 3rd echoes, D, using the 1st through the 4th echoes, and E, using all 5 echoes. F, Reference
Figure 5
Figure 5
Coronal water signal images of a representative subject. A, Image created using a neural network with the first echo as input. B, Reference
See this image and copyright information in PMC

References

    1. Dixon WT. Simple proton spectroscopic imaging. Radiology. 1984;153:189–194. - PubMed
    1. Ma J. Dixon techniques for water and fat imaging. J Magn Reson Imaging. 2008;28:543–558. - PubMed
    1. Yu H, Shimakawa A, McKenzie CA, Brodsky E, Brittain JH, Reeder SB. Multiecho water‐fat separation and simultaneous R2* estimation with multifrequency fat spectrum modeling. Magn Reson Med. 2008;60:1122–1134. - PMC - PubMed
    1. Szczepaniak LS, Nurenberg P, Leonard D, et al. Magnetic resonance spectroscopy to measure hepatic triglyceride content: prevalence of hepatic steatosis in the general population. Am J Physiol Endocrinol Metab. 2005;288:E462–E468. - PubMed
    1. Ma J. A single‐point Dixon technique for fat‐suppressed fast 3D gradient‐echo imaging with a flexible echo time. J Magn Reson Imaging. 2008;27:881–890. - PubMed

Publication types

MeSH terms