Volume 58, Issue 4 pp. 1200-1210

Research Article

DeepSWI: Using Deep Learning to Enhance Susceptibility Contrast on T2*-Weighted MRI

Ozan Genc MS,

Ozan Genc MS

Department of Radiology and Biomedical Imaging, University of California, San Francisco, California, USA

Institute of Biomedical Engineering, Boğaziçi University, Istanbul, Turkey

Search for more papers by this author

Melanie A. Morrison PhD,

Melanie A. Morrison PhD

orcid.org/0000-0002-5993-0515

Department of Radiology and Biomedical Imaging, University of California, San Francisco, California, USA

Search for more papers by this author

Javier E. Villanueva-Meyer MD,

Javier E. Villanueva-Meyer MD

Department of Radiology and Biomedical Imaging, University of California, San Francisco, California, USA

Department of Neurological Surgery, University of California, San Francisco, California, USA

Search for more papers by this author

Brian Burns PhD,

Brian Burns PhD

GE Healthcare, Menlo Park, California, USA

Search for more papers by this author

Christopher P. Hess MD, PhD,

Christopher P. Hess MD, PhD

Department of Radiology and Biomedical Imaging, University of California, San Francisco, California, USA

Department of Neurology, University of California, San Francisco, California, USA

Search for more papers by this author

Suchandrima Banerjee PhD,

Suchandrima Banerjee PhD

GE Healthcare, Menlo Park, California, USA

Search for more papers by this author

Janine M. Lupo PhD,

Corresponding Author

Janine M. Lupo PhD

[email protected]

orcid.org/0000-0002-0051-6387

Department of Radiology and Biomedical Imaging, University of California, San Francisco, California, USA

UCSF/UC Berkeley Graduate Group of Bioengineering, University of California, Berkeley, San Francisco, California, USA

Address reprint requests to: J.M.L., Byers Hall UCSF, Box 2532, 1700 4th Street, Suite 303D, San Francisco, California 94158-2330, USA. E-mail: [email protected]

Search for more papers by this author

Ozan Genc MS,

Ozan Genc MS

Department of Radiology and Biomedical Imaging, University of California, San Francisco, California, USA

Institute of Biomedical Engineering, Boğaziçi University, Istanbul, Turkey

Search for more papers by this author

Melanie A. Morrison PhD,

Melanie A. Morrison PhD

orcid.org/0000-0002-5993-0515

Department of Radiology and Biomedical Imaging, University of California, San Francisco, California, USA

Search for more papers by this author

Javier E. Villanueva-Meyer MD,

Javier E. Villanueva-Meyer MD

Department of Radiology and Biomedical Imaging, University of California, San Francisco, California, USA

Department of Neurological Surgery, University of California, San Francisco, California, USA

Search for more papers by this author

Brian Burns PhD,

Brian Burns PhD

GE Healthcare, Menlo Park, California, USA

Search for more papers by this author

Christopher P. Hess MD, PhD,

Christopher P. Hess MD, PhD

Department of Radiology and Biomedical Imaging, University of California, San Francisco, California, USA

Department of Neurology, University of California, San Francisco, California, USA

Search for more papers by this author

Suchandrima Banerjee PhD,

Suchandrima Banerjee PhD

GE Healthcare, Menlo Park, California, USA

Search for more papers by this author

Janine M. Lupo PhD,

Corresponding Author

Janine M. Lupo PhD

[email protected]

orcid.org/0000-0002-0051-6387

Department of Radiology and Biomedical Imaging, University of California, San Francisco, California, USA

UCSF/UC Berkeley Graduate Group of Bioengineering, University of California, Berkeley, San Francisco, California, USA

Address reprint requests to: J.M.L., Byers Hall UCSF, Box 2532, 1700 4th Street, Suite 303D, San Francisco, California 94158-2330, USA. E-mail: [email protected]

Search for more papers by this author

First published: 02 February 2023

https://doi.org/10.1002/jmri.28622

Share a link

Email
Wechat
Bluesky

Abstract

Background

Although susceptibility-weighted imaging (SWI) is the gold standard for visualizing cerebral microbleeds (CMBs) in the brain, the required phase data are not always available clinically. Having a postprocessing tool for generating SWI contrast from T2*-weighted magnitude images is therefore advantageous.

Purpose

To create synthetic SWI images from clinical T2*-weighted magnitude images using deep learning and evaluate the resulting images in terms of similarity to conventional SWI images and ability to detect radiation-associated CMBs.

Study Type

Retrospective.

Population

A total of 145 adults (87 males/58 females; 43.9 years old) with radiation-associated CMBs were used to train (16,093 patches/121 patients), validate (484 patches/4 patients), and test (2420 patches/20 patients) our networks.

Field Strength/Sequence

3D T2*-weighted, gradient-echo acquired at 3 T.

Assessment

Structural similarity index (SSIM), peak signal-to-noise-ratio (PSNR), normalized mean-squared-error (nMSE), CMB counts, and line profiles were compared among magnitude, original SWI, and synthetic SWI images. Three blinded raters (J.E.V.M., M.A.M., B.B. with 8-, 6-, and 4-years of experience, respectively) independently rated and classified test-set images.

Statistical Tests

Kruskall–Wallis and Wilcoxon signed-rank tests were used to compare SSIM, PSNR, nMSE, and CMB counts among magnitude, original SWI, and predicted synthetic SWI images. Intraclass correlation assessed interrater variability. P values <0.005 were considered statistically significant.

Results

SSIM values of the predicted vs. original SWI (0.972, 0.995, 0.9864) were statistically significantly higher than that of the magnitude vs. original SWI (0.970, 0.994, 0.9861) for whole brain, vascular structures, and brain tissue regions, respectively; 67% (19/28) CMBs detected on original SWI images were also detected on the predicted SWI, whereas only 10 (36%) were detected on magnitude images. Overall image quality was similar between the synthetic and original SWI images, with less artifacts on the former.

Conclusions

This study demonstrated that deep learning can increase the susceptibility contrast present in neurovasculature and CMBs on T2*-weighted magnitude images, without residual susceptibility-induced artifacts. This may be useful for more accurately estimating CMB burden from magnitude images alone.

Evidence Level

Technical Efficacy

Stage 2.

REFERENCES

1Bowers DC, Liu Y, Leisenring W, et al. Late-occurring stroke among long-term survivors of childhood leukemia and brain tumors: A report from the Childhood Cancer Survivor Study. J Clin Oncol off J Am Soc Clin Oncol 2006; 24(33): 5277-5282. https://doi.org/10.1200/JCO.2006.07.2884.
10.1200/JCO.2006.07.2884
PubMed Web of Science® Google Scholar
2Halefoglu A, Yousem D. Susceptibility weighted imaging: Clinical applications and future directions. World J Radiol 2018; 10(4): 30-45.
10.4329/wjr.v10.i4.30
PubMed Web of Science® Google Scholar
3Haacke E, Xu Y, Cheng Y, Reichenbach J. Susceptibility weighted imaging (SWI). Magn Reson Med 2004; 52: 612-618.
10.1002/mrm.20198
PubMed Web of Science® Google Scholar
4Bian W, Hess CP, Chang SM, Nelson SJ, Lupo JM. Susceptibility-weighted MR imaging of radiation therapy-induced cerebral microbleeds in patients with glioma: A comparison between 3T and 7T. Neuroradiology 2014; 56(2): 91-96. https://doi.org/10.1007/s00234-013-1297-8.
10.1007/s00234-013-1297-8
PubMed Web of Science® Google Scholar
5Li N, Wang W, Sati P, Pham D, Butman J. Quantitative assessment of susceptibility-weighted imaging processing methods. J Magn Reson Imaging 2014; 40(6): 1463-1473.
10.1002/jmri.24501
PubMed Web of Science® Google Scholar
6Xiao S, Wu Y, Lee AY, Rokem A. MRI2MRI: A deep convolutional network that accurately transforms between brain MRI contrasts. bioRxiv 2018;289926. https://doi.org/10.1101/289926.
10.1101/289926
Google Scholar
7Wolterink JM, Dinkla AM, Savenije MHF, Seevinck PR, van den Berg CAT, Išgum I. Deep MR to CT synthesis using unpaired data. Simulation and synthesis in medical imaging. Cham, Switzerland: Springer; 2017. p. 14-23. https://doi.org/10.1007/978-3-319-68127-6_2.
10.1007/978-3-319-68127-6_2
Google Scholar
8Leynes A, Yang J, Wiesinger F, et al. Zero-echo-time and dixon deep pseudo-CT (ZeDD CT): Direct generation of pseudo-CT images for pelvic PET/MRI attenuation correction using deep convolutional neural networks with multiparametric MRI. J Nucl Med 2018; 59: 852-858.
10.2967/jnumed.117.198051
PubMed Web of Science® Google Scholar
9“ Synthesis of Positron Emission Tomography (PET) Images via Multi-channel Generative Adversarial Networks (GANs) springerprofessional.de. Accessed on April 27, 2021. Available from: https://www.springerprofessional.de/en/synthesis-of-positron-emission-tomography-pet-images-via-multi-c/15033206
Google Scholar
10Chartsias A, Joyce T, Giuffrida M, Tsaftaris S. Multimodal MR synthesis via modality-invariant latent representation. IEEE Trans Med Imaging 2018; 37: 803-814.
10.1109/TMI.2017.2764326
PubMed Web of Science® Google Scholar
11Dar S, Yurt M, Karacan L, Erdem A, Erdem E, Cukur T. Image synthesis in multi-contrast MRI with conditional generative adversarial networks. IEEE Trans Med Imaging 2019; 38: 2375-2388.
10.1109/TMI.2019.2901750
PubMed Web of Science® Google Scholar
12Wei W, Poirion E, Bodini B, et al. Fluid-attenuated inversion recovery MRI synthesis from multisequence MRI using three-dimensional fully convolutional networks for multiple sclerosis. J Med Imaging (Bellingham) 2019; 6: 1.
10.1117/1.JMI.6.1.014005
Web of Science® Google Scholar
13Mani A, Santini T, Puppala R, et al. Applying deep learning to accelerated clinical brain magnetic resonance imaging for multiple sclerosis. Front Neurol 2021; 12:685276. https://doi.org/10.3389/fneur.2021.685276.
10.3389/fneur.2021.685276
PubMed Web of Science® Google Scholar
14Nie D, Trullo R, Lian J, et al. Medical image synthesis with deep convolutional adversarial networks. IEEE Trans Biomed Eng 2018; 65(12): 2720-2730. https://doi.org/10.1109/TBME.2018.2814538.
10.1109/TBME.2018.2814538
PubMed Web of Science® Google Scholar
15Sharma A, Kaur P, A. Nigam, et al. Learning to decode 7T-like MR image reconstruction from 3T MR images. 2018, Accessed on April 27, 2021. [Online]. Available from: http://arxiv.org/abs/1806.06886
Google Scholar
16 Convolutional neural network for reconstruction of 7T-like images from 3T MRI using appearance and anatomical features, springerprofessional.de Accessed on April 27, 2021. https://www.springerprofessional.de/en/convolutional-neural-network-for-reconstruction-of-7t-like-image/10832734.
Google Scholar
17Roddy E, Sear K, Felton E, et al. Presence of cerebral microbleeds is associated with worse executive function in pediatric brain tumor survivors. Neuro Oncol 2016; 18: 1548-1558.
10.1093/neuonc/now082.02
PubMed Web of Science® Google Scholar
18Morrison M, Mueller S, Felton E, et al. Rate of radiation-induced microbleed formation on 7T MRI relates to cognitive impairment in young patients treated with radiation therapy for a brain tumor. Radiother Oncol 2021; 154: 145-153.
10.1016/j.radonc.2020.09.028
CAS PubMed Web of Science® Google Scholar
19Lupo J, Banerjee S, Hammond KE, et al. GRAPPA-based susceptibility-weighted imaging of normal volunteers and patients with brain tumor at 7T. Magn Reson Imaging 2009; 27: 480-488.
10.1016/j.mri.2008.08.003
PubMed Web of Science® Google Scholar
20Chen Y, Jakary A, Avadiappan S, Hess CP, Lupo JM. QSMGAN: Improved quantitative susceptibility mapping using 3D generative adversarial networks with increased receptive field. Neuroimage 2020; 207:116389. https://doi.org/10.1016/j.neuroimage.2019.116389.
10.1016/j.neuroimage.2019.116389
PubMed Web of Science® Google Scholar
21Bian W, Hess CP, Chang SM, Nelson SJ, Lupo JM. Computer-aided detection of radiation-induced cerebral microbleeds on susceptibility-weighted MR images. NeuroImage Clin. 2013; 2: 282-290. https://doi.org/10.1016/j.nicl.2013.01.012.
10.1016/j.nicl.2013.01.012
PubMed Web of Science® Google Scholar
22Ronneberger O, Fischer P, Brox T. U-Net: Convolutional networks for biomedical image segmentation. 2015, Accessed on April 27, 2021. [Online]. Available: http://arxiv.org/abs/1505.04597
Google Scholar
23Goodfellow IJ, Pouget-Abadie J, Mirza M, et al. Generative adversarial networks,” ArXiv14062661 Cs Stat. 2014, Accessed on April 27, 2021. [Online]. Available: http://arxiv.org/abs/1406.2661
Google Scholar
24Gulrajani I, Ahmed F, Arjovsky M, et al. Improved training of wasserstein GANs ArXiv170400028 Cs Stat. 2017 Accessed on April 27, 2021. [Online]. Available: http://arxiv.org/abs/1704.00028
Google Scholar
25Mao X, Li Q, Xie H, et al. Least squares generative adversarial networks ArXiv161104076 Cs. 2017, Accessed on April 27, 2021. [Online]. Available: http://arxiv.org/abs/1611.04076
Google Scholar
26Wang Z, She Q, Ward TE, Generative adversarial networks in computer vision: A survey and taxonomy. ArXiv190601529 Cs. 2020, Accessed on April 27, 2021. [Online]. Available from: http://arxiv.org/abs/1906.01529
Google Scholar
27Akiba T, Sano S, Yanase T, et al. Optuna: A next-generation hyperparameter optimization framework, ArXiv190710902 Cs Stat. 2019, Accessed on April 27, 2021. [Online]. Available: http://arxiv.org/abs/1907.10902.
Google Scholar
28Morrison MA, Payabvash S, Chen Y, et al. A user-guided tool for semi-automated cerebral microbleed detection and volume segmentation: Evaluating vascular injury and data labelling for machine learning. NeuroImage Clin 2018; 20: 498-505. https://doi.org/10.1016/j.nicl.2018.08.002.
10.1016/j.nicl.2018.08.002
PubMed Web of Science® Google Scholar
29Shrout P, Fleiss J. Intraclass correlations: Uses in assessing rater reliability. Psychol Bull 1979; 86: 420-428.
10.1037/0033-2909.86.2.420
CAS PubMed Web of Science® Google Scholar

Volume58, Issue4

October 2023

Pages 1200-1210

DeepSWI: Using Deep Learning to Enhance Susceptibility Contrast on T2*-Weighted MRI

Abstract

Background

Purpose

Study Type

Population

Field Strength/Sequence

Assessment

Statistical Tests

Results

Conclusions

Evidence Level

Technical Efficacy

REFERENCES

References

Information

About Wiley Online Library

Help & Support

Opportunities

Connect with Wiley

DeepSWI: Using Deep Learning to Enhance Susceptibility Contrast on T2*-Weighted MRI

Abstract

Background

Purpose

Study Type

Population

Field Strength/Sequence

Assessment

Statistical Tests

Results

Conclusions

Evidence Level

Technical Efficacy

REFERENCES

References

Related

Information