Tags

Type your tag names separated by a space and hit enter

Reconstruction of Compressed-sensing MR Imaging Using Deep Residual Learning in the Image Domain.
Magn Reson Med Sci. 2020 Jul 02 [Online ahead of print]MR

Abstract

PURPOSE

A deep residual learning convolutional neural network (DRL-CNN) was applied to improve image quality and speed up the reconstruction of compressed sensing magnetic resonance imaging. The reconstruction performances of the proposed method was compared with iterative reconstruction methods.

METHODS

The proposed method adopted a DRL-CNN to learn the residual component between the input and output images (i.e., aliasing artifacts) for image reconstruction. The CNN-based reconstruction was compared with iterative reconstruction methods. To clarify the reconstruction performance of the proposed method, reconstruction experiments using 1D-, 2D-random under-sampling and sampling patterns that mix random and non-random under-sampling were executed. The peak-signal-to-noise ratio (PSNR) and the structural similarity index (SSIM) were examined for various numbers of training images, sampling rates, and numbers of training epochs.

RESULTS

The experimental results demonstrated that reconstruction time is drastically reduced to 0.022 s per image compared with that for conventional iterative reconstruction. The PSNR and SSIM were improved as the coherence of the sampling pattern increases. These results indicate that a deep CNN can learn coherent artifacts and is effective especially for cases where the randomness of k-space sampling is rather low. Simulation studies showed that variable density non-random under-sampling was a promising sampling pattern in 1D-random under-sampling of 2D image acquisition.

CONCLUSION

A DRL-CNN can recognize and predict aliasing artifacts with low incoherence. It was demonstrated that reconstruction time is significantly reduced and the improvement in the PSNR and SSIM is higher in 1D-random under-sampling than in 2D. The requirement of incoherence for aliasing artifacts is different from that for iterative reconstruction.

Authors+Show Affiliations

Department of Innovation Systems Engineering, Graduate School of Engineering, Utsunomiya University.Department of Innovation Systems Engineering, Graduate School of Engineering, Utsunomiya University.

Pub Type(s)

Journal Article

Language

eng

PubMed ID

32611937

Citation

Ouchi, Shohei, and Satoshi Ito. "Reconstruction of Compressed-sensing MR Imaging Using Deep Residual Learning in the Image Domain." Magnetic Resonance in Medical Sciences : MRMS : an Official Journal of Japan Society of Magnetic Resonance in Medicine, 2020.
Ouchi S, Ito S. Reconstruction of Compressed-sensing MR Imaging Using Deep Residual Learning in the Image Domain. Magn Reson Med Sci. 2020.
Ouchi, S., & Ito, S. (2020). Reconstruction of Compressed-sensing MR Imaging Using Deep Residual Learning in the Image Domain. Magnetic Resonance in Medical Sciences : MRMS : an Official Journal of Japan Society of Magnetic Resonance in Medicine. https://doi.org/10.2463/mrms.mp.2019-0139
Ouchi S, Ito S. Reconstruction of Compressed-sensing MR Imaging Using Deep Residual Learning in the Image Domain. Magn Reson Med Sci. 2020 Jul 2; PubMed PMID: 32611937.
* Article titles in AMA citation format should be in sentence-case
TY - JOUR T1 - Reconstruction of Compressed-sensing MR Imaging Using Deep Residual Learning in the Image Domain. AU - Ouchi,Shohei, AU - Ito,Satoshi, Y1 - 2020/07/02/ PY - 2020/7/3/entrez PY - 2020/7/3/pubmed PY - 2020/7/3/medline KW - compressed sensing KW - deep learning KW - reconstruction JF - Magnetic resonance in medical sciences : MRMS : an official journal of Japan Society of Magnetic Resonance in Medicine JO - Magn Reson Med Sci N2 - PURPOSE: A deep residual learning convolutional neural network (DRL-CNN) was applied to improve image quality and speed up the reconstruction of compressed sensing magnetic resonance imaging. The reconstruction performances of the proposed method was compared with iterative reconstruction methods. METHODS: The proposed method adopted a DRL-CNN to learn the residual component between the input and output images (i.e., aliasing artifacts) for image reconstruction. The CNN-based reconstruction was compared with iterative reconstruction methods. To clarify the reconstruction performance of the proposed method, reconstruction experiments using 1D-, 2D-random under-sampling and sampling patterns that mix random and non-random under-sampling were executed. The peak-signal-to-noise ratio (PSNR) and the structural similarity index (SSIM) were examined for various numbers of training images, sampling rates, and numbers of training epochs. RESULTS: The experimental results demonstrated that reconstruction time is drastically reduced to 0.022 s per image compared with that for conventional iterative reconstruction. The PSNR and SSIM were improved as the coherence of the sampling pattern increases. These results indicate that a deep CNN can learn coherent artifacts and is effective especially for cases where the randomness of k-space sampling is rather low. Simulation studies showed that variable density non-random under-sampling was a promising sampling pattern in 1D-random under-sampling of 2D image acquisition. CONCLUSION: A DRL-CNN can recognize and predict aliasing artifacts with low incoherence. It was demonstrated that reconstruction time is significantly reduced and the improvement in the PSNR and SSIM is higher in 1D-random under-sampling than in 2D. The requirement of incoherence for aliasing artifacts is different from that for iterative reconstruction. SN - 1880-2206 UR - https://www.unboundmedicine.com/medline/citation/32611937/Reconstruction_of_Compressed_sensing_MR_Imaging_Using_Deep_Residual_Learning_in_the_Image_Domain_ L2 - https://dx.doi.org/10.2463/mrms.mp.2019-0139 DB - PRIME DP - Unbound Medicine ER -
Try the Free App:
Prime PubMed app for iOS iPhone iPad
Prime PubMed app for Android
Prime PubMed is provided
free to individuals by:
Unbound Medicine.