Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Learning MRI artefact removal with unpaired data

A Publisher Correction to this article was published on 27 January 2021

This article has been updated


Retrospective artefact correction (RAC) improves image quality post acquisition and enhances image usability. Recent machine-learning-driven techniques for RAC are predominantly based on supervised learning, so practical utility can be limited as data with paired artefact-free and artefact-corrupted images are typically insufficient or even non-existent. Here we show that unwanted image artefacts can be disentangled and removed from an image via an RAC neural network learned with unpaired data. This implies that our method does not require matching artefact-corrupted data to be either collected via acquisition or generated via simulation. Experimental results demonstrate that our method is remarkably effective in removing artefacts and retaining anatomical details in images with different contrasts.

This is a preview of subscription content, access via your institution

Relevant articles

Open Access articles citing this article.

Access options

Rent or buy this article

Get just this article for as long as you need it


Prices may be subject to local taxes which are calculated during checkout

Fig. 1: Overview of DUNCAN.
Fig. 2: Visual comparison of corrected in vivo images.
Fig. 3: Visual comparison of corrected in silico images.
Fig. 4: Quantitative comparison of corrected in silico T1-weighted images.
Fig. 5: Quantitative comparison of corrected in silico T2-weighted images.
Fig. 6: Segmentation accuracy of in silico images.

Data availability

The data used in this paper were provided by the investigative team of the UNC/UMN Baby Connectome Project. The data can be obtained from the National Institute of Mental Health Data Archive (NDA) ( or by contacting the investigative team23.

Code availability

The source code and trained models for this study are publicly available on Zenodo (

Change history


  1. Budde, J., Shajan, G., Scheffler, K. & Pohmann, R. Ultra-high resolution imaging of the human brain using acquisition-weighted imaging at 9.4 T. Neuroimage 86, 592–598 (2014).

    Article  Google Scholar 

  2. Zhuo, J. & Gullapalli, R. P. MR artifacts, safety and quality control. Radiographics 26, 275–297 (2006).

    Article  Google Scholar 

  3. Andre, J. et al. Toward quantifying the prevalence, severity and cost associated with patient motion during clinical MR examinations. J. Am. Coll. Radiol. 12, 689–695 (2015).

    Article  Google Scholar 

  4. Zaitsev, M., Maclaren, J. & Herbst, M. Motion artifacts in MRI: a complex problem with many partial solutions. J. Magn. Reson. Imaging 42, 887–901 (2015).

    Article  Google Scholar 

  5. Zaitsev, M., Dold, C., Sakas, G., Hennig, J. & Speck, O. Magnetic resonance imaging of freely moving objects: prospective real-time motion correction using an external optical motion tracking system. Neuroimage 31, 1038–1050 (2006).

    Article  Google Scholar 

  6. Qin, L. et al. Prospective head-movement correction for high-resolution MRI using an in-bore optical tracking system. Magn. Reson. Med. 62, 924–934 (2009).

    Article  Google Scholar 

  7. Ooi, M. B., Krueger, S., Thomas, W. J., Swaminathan, S. V. & Brown, T. R. Prospective real-time correction for arbitrary head motion using active markers. Magn. Reson. Med. 62, 943–954 (2009).

    Article  Google Scholar 

  8. Schulz, J. et al. An embedded optical tracking system for motion-corrected magnetic resonance imaging at 7 T. Magn. Reson. Mater. Phys. Biol. Med. 25, 443–453 (2012).

    Article  Google Scholar 

  9. Maclaren, J. et al. Measurement and correction of microscopic head motion during magnetic resonance imaging of the brain. PLoS ONE 7, e48088 (2012).

    Article  Google Scholar 

  10. Maclaren, J., Herbst, M., Speck, O. & Zaitsev, M. Prospective motion correction in brain imaging: a review. Magn. Reson. Med. 69, 621–636 (2012).

    Article  Google Scholar 

  11. Pipe, J. G. Motion correction with propeller MRI: application to head motion and freebreathing cardiac imaging. Magn. Reson. Med. 42, 963–969 (1999).

    Article  Google Scholar 

  12. Vertinsky, A. T. et al. Performance of PROPELLER relative to standard FSE T2-weighted imaging in pediatric brain MRI. Pediatric Radiol. 39, 1038–1047 (2009).

    Article  Google Scholar 

  13. Jin, K. H., McCann, M. T., Froustey, E. & Unser, M. Deep convolutional neural network for inverse problems in imaging. IEEE Trans. Image Process. 26, 4509–4522 (2017).

    Article  MathSciNet  Google Scholar 

  14. Haskell, M. W. et al. Network accelerated motion estimation and reduction (NAMER): convolutional neural network guided retrospective motion correction using a separable motion model. Magn. Reson. Med. 82, 1452–1461 (2019).

    Article  Google Scholar 

  15. Johnson, P. M. & Drangova, M. Motion correction in MRI using deep learning. In Proc. 26th Annual Meeting ISMRM 4098 (ISMRM, 2018).

  16. Tamada, D., Kromrey, M.-L., Ichikawa, S., Onishi, H. & Motosugi, U. Motion artifact reduction using a convolutional neural network for dynamic contrast enhanced MR imaging of the liver. Magn. Reson. Med. Sci. 19, 64–76 (2020).

    Article  Google Scholar 

  17. Küstner, T. et al. Retrospective correction of motion-affected MR images using deep learning frameworks. Magn. Reson. Med. 82, 1527–1540 (2019).

    Article  Google Scholar 

  18. Johnson, P. M. & Drangova, M. Conditional generative adversarial network for 3D rigid-body motion correction in MRI. Magn. Reson. Med. 82, 901–910 (2019).

    Google Scholar 

  19. Liu, M.-Y., Breuel, T. & Kautz, J. Unsupervised image-to-image translation networks. In Proc. Advances in Neural Information Processing Systems 700–708 (NIPS, 2017).

  20. Zhu, J.-Y., Park, T., Isola, P. & Efros, A. A. Unpaired image-to-image translation using cycle-consistent adversarial networks. In Proc. 2017 IEEE International Conference on Computer Vision (ICCV) 2223–2232 (IEEE, 2017).

  21. Zhu, J.-Y. et al. Toward multimodal image-to-image translation. In Proc. Advances in Neural Information Processing Systems 465–476 (NIPS, 2017).

  22. Isola, P., Zhu, J.-Y., Zhou, T. & Efros, A. A. Image-to-image translation with conditional adversarial networks. In Proc. 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 1125–1134 (IEEE, 2017).

  23. Howell, B. R. et al. The UNC/UMN baby connectome project (BCP): an overview of the study design and protocol development. Neuroimage 185, 891–905 (2019).

    Article  Google Scholar 

  24. Perlin, K. An image synthesizer. ACM SIGGRAPH Comput. Graph. 19, 287–296 (1985).

    Article  Google Scholar 

  25. Wang, Z., Bovik, A., Sheikh, H. & Simoncelli, E. Image quality assessment: from error visibility to structural similarity. IEEE Trans. Image Process. 13, 600–612 (2004).

    Article  Google Scholar 

  26. Wang, Z., Simoncelli, E. & Bovik, A. Multiscale structural similarity for image quality assessment. In Proc. 37th IEEE Asilomar Conference on Signals, Systems and Computers 1398–1402 (IEEE, 2003).

  27. Sheikh, H. & Bovik, A. Image information and visual quality. IEEE Trans. Image Process. 15, 430–444 (2006).

    Article  Google Scholar 

  28. Wang, Z. & Bovik, A. A universal image quality index. IEEE Signal Process. Lett. 9, 81–84 (2002).

    Article  Google Scholar 

  29. Mao, X. et al. Least squares generative adversarial networks. In Proc. 2017 IEEE International Conference on Computer Vision (ICCV) 2794–2802 (IEEE, 2017).

  30. Smith, S. M. Fast robust automated brain extraction. Human Brain Mapp. 17, 143–155 (2002).

    Article  Google Scholar 

  31. Zhang, Y., Brady, M. & Smith, S. Segmentation of brain MR images through a hidden Markov random field model and the expectation-maximization algorithm. IEEE Trans. Med. Imaging 20, 45–57 (2001).

    Article  Google Scholar 

  32. Ulyanov, D., Vedaldi, A. & Lempitsky, V. Improved texture networks: maximizing quality and diversity in feed-forward stylization and texture synthesis. In Proc. IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 6924–6932 (IEEE, 2017).

  33. Simonyan, K. & Zisserman, A. Very deep convolutional networks for large-scale image recognition. In Proc. International Conference on Learning Representations 1–14 (CLR, 2015).

  34. Liu, S. et al. Code used in “Learning MRI artefact removal with unpaired data”. Zenodo (2020).

Download references


This work was supported in part by National Institutes of Health grants (EB006733, AG053867, MH117943, MH104324, MH110274) and the efforts of the UNC/UMN Baby Connectome Project Consortium. The authors thank X. Zong of the University of North Carolina at Chapel Hill for an initial discussion on motion artefact simulation and Y. Hong of the University of North Carolina at Chapel Hill and Y. Chen of Case Western Reserve University for proofreading the manuscript.

Author information

Authors and Affiliations



S.L. designed the framework and network architecture, carried out the implementation, performed the experiments and analysed the data. S.L. and P.-T.Y. wrote the manuscript. S.L., K.-H.T. and P.-T.Y. revised the manuscript. L.Q. contributed to the initial formulation of the method before moving to Stanford University. W.L. provided the infant data for training and testing. P.-T.Y. conceived the study and were in charge of overall direction and planning. D.S. was involved in the initial discussion of the problem when he was with the University of North Carolina at Chapel Hill. All work was done at the University of North Carolina at Chapel Hill.

Corresponding author

Correspondence to Pew-Thian Yap.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Peer review information

Peer review information Nature Machine Intelligence thanks Chuyang Ye and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Supplementary Discussion and Supplementary Figs. 1–12.

Rights and permissions

Reprints and Permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Liu, S., Thung, KH., Qu, L. et al. Learning MRI artefact removal with unpaired data. Nat Mach Intell 3, 60–67 (2021).

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI:

This article is cited by


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing