Tags

Type your tag names separated by a space and hit enter

Compressed sensing effects on quantitative analysis of undersampled human brain sodium MRI.
Magn Reson Med. 2020 03; 83(3):1025-1033.MR

Abstract

PURPOSE

The clinical application of sodium MRI is hampered due to relatively low image quality and associated long acquisition times. Compressed sensing (CS) aims at a reduction of measurement time, but has been found to encompass quantitative estimation bias when used in low SNR x-Nuclei imaging. This work analyses CS in quantitative human brain sodium MRI from undersampled acquisitions and provides recommendations for tissue sodium concentration (TSC) estimation.

METHODS

CS reconstructions from 3D radial acquisitions of 5 healthy volunteers were investigated over varying undersampling factors (USFs) and CS penalty weights on different sparsity domains, Wavelet, Discrete Cosine Transform (DCT), and Identity. Resulting images were compared with highly sampled and undersampled NUFFT-based images and evaluated for image quality (i.e. structural similarity), image intensity bias, and its effect on TSC estimates in gray and white matter.

RESULTS

Wavelet-based CS reconstructions show highest image quality with stable TSC estimates for most USFs. Up to an USF of 4, images showed good structural detail. DCT and Identity-based CS enable good image quality, however show a bias in TSC with a reduction in estimates across USFs.

CONCLUSIONS

The image intensity bias is lowest in Wavelet-based reconstructions and enables an up to fourfold acquisition speed up while maintaining good structural detail. The associated acquisition time reduction can facilitate a translation of sodium MRI into clinical routine.

Authors+Show Affiliations

Department of Biomedical Engineering, The University of Melbourne, Parkville, Australia. Melbourne Brain Centre Imaging Unit, Department of Medicine and Radiology, The University of Melbourne, Parkville, Australia.Melbourne Brain Centre Imaging Unit, Department of Medicine and Radiology, The University of Melbourne, Parkville, Australia.Melbourne Brain Centre Imaging Unit, Department of Medicine and Radiology, The University of Melbourne, Parkville, Australia.Melbourne Brain Centre Imaging Unit, Department of Medicine and Radiology, The University of Melbourne, Parkville, Australia.Department of Radiology, Guy's and St. Thomas' NHS Foundation Trust, London, UK.Department of Biomedical Engineering, The University of Melbourne, Parkville, Australia. Melbourne Brain Centre Imaging Unit, Department of Medicine and Radiology, The University of Melbourne, Parkville, Australia.

Pub Type(s)

Journal Article
Research Support, Non-U.S. Gov't

Language

eng

PubMed ID

31502711

Citation

Blunck, Yasmin, et al. "Compressed Sensing Effects On Quantitative Analysis of Undersampled Human Brain Sodium MRI." Magnetic Resonance in Medicine, vol. 83, no. 3, 2020, pp. 1025-1033.
Blunck Y, Kolbe SC, Moffat BA, et al. Compressed sensing effects on quantitative analysis of undersampled human brain sodium MRI. Magn Reson Med. 2020;83(3):1025-1033.
Blunck, Y., Kolbe, S. C., Moffat, B. A., Ordidge, R. J., Cleary, J. O., & Johnston, L. A. (2020). Compressed sensing effects on quantitative analysis of undersampled human brain sodium MRI. Magnetic Resonance in Medicine, 83(3), 1025-1033. https://doi.org/10.1002/mrm.27993
Blunck Y, et al. Compressed Sensing Effects On Quantitative Analysis of Undersampled Human Brain Sodium MRI. Magn Reson Med. 2020;83(3):1025-1033. PubMed PMID: 31502711.
* Article titles in AMA citation format should be in sentence-case
TY - JOUR T1 - Compressed sensing effects on quantitative analysis of undersampled human brain sodium MRI. AU - Blunck,Yasmin, AU - Kolbe,Scott C, AU - Moffat,Bradford A, AU - Ordidge,Roger J, AU - Cleary,Jon O, AU - Johnston,Leigh A, Y1 - 2019/09/10/ PY - 2019/05/06/received PY - 2019/07/30/revised PY - 2019/08/19/accepted PY - 2019/9/11/pubmed PY - 2019/9/11/medline PY - 2019/9/11/entrez KW - compressed sensing KW - iterative reconstruction KW - quantitative imaging KW - sodium (23 Na) MRI KW - x-Nuclei SP - 1025 EP - 1033 JF - Magnetic resonance in medicine JO - Magn Reson Med VL - 83 IS - 3 N2 - PURPOSE: The clinical application of sodium MRI is hampered due to relatively low image quality and associated long acquisition times. Compressed sensing (CS) aims at a reduction of measurement time, but has been found to encompass quantitative estimation bias when used in low SNR x-Nuclei imaging. This work analyses CS in quantitative human brain sodium MRI from undersampled acquisitions and provides recommendations for tissue sodium concentration (TSC) estimation. METHODS: CS reconstructions from 3D radial acquisitions of 5 healthy volunteers were investigated over varying undersampling factors (USFs) and CS penalty weights on different sparsity domains, Wavelet, Discrete Cosine Transform (DCT), and Identity. Resulting images were compared with highly sampled and undersampled NUFFT-based images and evaluated for image quality (i.e. structural similarity), image intensity bias, and its effect on TSC estimates in gray and white matter. RESULTS: Wavelet-based CS reconstructions show highest image quality with stable TSC estimates for most USFs. Up to an USF of 4, images showed good structural detail. DCT and Identity-based CS enable good image quality, however show a bias in TSC with a reduction in estimates across USFs. CONCLUSIONS: The image intensity bias is lowest in Wavelet-based reconstructions and enables an up to fourfold acquisition speed up while maintaining good structural detail. The associated acquisition time reduction can facilitate a translation of sodium MRI into clinical routine. SN - 1522-2594 UR - https://www.unboundmedicine.com/medline/citation/31502711/Compressed_sensing_effects_on_quantitative_analysis_of_undersampled_human_brain_sodium_MRI_ L2 - https://doi.org/10.1002/mrm.27993 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.