Perfusion quantification by model-free arterial spin labeling using nonlinear stochastic regularization deconvolution.
(2013) In Magnetic Resonance in Medicine 70(5). p.1470-1480- Abstract
- Purpose: Quantification of cerebral blood flow can be accomplished by model-free arterial spin labeling using the quantitative STAR labeling of arterial regions (QUASAR) sequence. The required deconvolution is normally based on block-circulant singular value decomposition (cSVD)/oscillation SVD (oSVD), an algorithm associated with nonphysiological residue functions and potential effects of arterial dispersion. The aim of this work was to amend this by implementing nonlinear stochastic regularization (NSR) deconvolution, previously used to retrieve realistic residue functions in dynamic susceptibility contrast MRI. METHODS: To characterize the residue function in model-free arterial spin labeling, and possibly to improve absolute cerebral... (More)
- Purpose: Quantification of cerebral blood flow can be accomplished by model-free arterial spin labeling using the quantitative STAR labeling of arterial regions (QUASAR) sequence. The required deconvolution is normally based on block-circulant singular value decomposition (cSVD)/oscillation SVD (oSVD), an algorithm associated with nonphysiological residue functions and potential effects of arterial dispersion. The aim of this work was to amend this by implementing nonlinear stochastic regularization (NSR) deconvolution, previously used to retrieve realistic residue functions in dynamic susceptibility contrast MRI. METHODS: To characterize the residue function in model-free arterial spin labeling, and possibly to improve absolute cerebral blood flow quantification, NSR was applied to deconvolution of QUASAR data. For comparison, SVD-based deconvolution was also employed. Residue function characteristics and cerebral blood flow values from 10 volunteers were obtained. Simulations were performed to support the in vivo results. RESULTS: NSR was able to resolve realistic residue functions in contrast to the SVD-based methods. Mean cerebral blood flow estimates in gray matter were 36.6 ± 2.6, 28.6 ± 3.3, 40.9 ± 3.6, and 42.9 ± 3.9 mL/100 g/min for cSVD, oSVD, NSR, and NSR with correction for arterial dispersion, respectively. In simulations, the NSR-based perfusion estimates showed better accuracy than the SVD-based approaches. CONCLUSION: Perfusion quantification by model-free arterial spin labeling is evidently dependent on the selected deconvolution method, and NSR is a feasible alternative to SVD-based methods. Magn Reson Med, 2012. © 2012 Wiley Periodicals, Inc. (Less)
Please use this url to cite or link to this publication:
https://lup.lub.lu.se/record/3439060
- author
- Ahlgren, André LU ; Wirestam, Ronnie LU ; Petersen, Esben Thade ; Ståhlberg, Freddy LU and Knutsson, Linda LU
- organization
- publishing date
- 2013
- type
- Contribution to journal
- publication status
- published
- subject
- in
- Magnetic Resonance in Medicine
- volume
- 70
- issue
- 5
- pages
- 1470 - 1480
- publisher
- John Wiley & Sons Inc.
- external identifiers
-
- wos:000326115000031
- pmid:23281031
- scopus:84886782698
- pmid:23281031
- ISSN
- 1522-2594
- DOI
- 10.1002/mrm.24587
- project
- MRI brain perfusion quantification at 3 tesla using arterial spin labeling
- language
- English
- LU publication?
- yes
- id
- e08bd971-c4e2-4666-aab3-ef612e18912e (old id 3439060)
- alternative location
- http://www.ncbi.nlm.nih.gov/pubmed/23281031?dopt=Abstract
- date added to LUP
- 2016-04-01 10:11:14
- date last changed
- 2022-04-04 03:14:21
@article{e08bd971-c4e2-4666-aab3-ef612e18912e, abstract = {{Purpose: Quantification of cerebral blood flow can be accomplished by model-free arterial spin labeling using the quantitative STAR labeling of arterial regions (QUASAR) sequence. The required deconvolution is normally based on block-circulant singular value decomposition (cSVD)/oscillation SVD (oSVD), an algorithm associated with nonphysiological residue functions and potential effects of arterial dispersion. The aim of this work was to amend this by implementing nonlinear stochastic regularization (NSR) deconvolution, previously used to retrieve realistic residue functions in dynamic susceptibility contrast MRI. METHODS: To characterize the residue function in model-free arterial spin labeling, and possibly to improve absolute cerebral blood flow quantification, NSR was applied to deconvolution of QUASAR data. For comparison, SVD-based deconvolution was also employed. Residue function characteristics and cerebral blood flow values from 10 volunteers were obtained. Simulations were performed to support the in vivo results. RESULTS: NSR was able to resolve realistic residue functions in contrast to the SVD-based methods. Mean cerebral blood flow estimates in gray matter were 36.6 ± 2.6, 28.6 ± 3.3, 40.9 ± 3.6, and 42.9 ± 3.9 mL/100 g/min for cSVD, oSVD, NSR, and NSR with correction for arterial dispersion, respectively. In simulations, the NSR-based perfusion estimates showed better accuracy than the SVD-based approaches. CONCLUSION: Perfusion quantification by model-free arterial spin labeling is evidently dependent on the selected deconvolution method, and NSR is a feasible alternative to SVD-based methods. Magn Reson Med, 2012. © 2012 Wiley Periodicals, Inc.}}, author = {{Ahlgren, André and Wirestam, Ronnie and Petersen, Esben Thade and Ståhlberg, Freddy and Knutsson, Linda}}, issn = {{1522-2594}}, language = {{eng}}, number = {{5}}, pages = {{1470--1480}}, publisher = {{John Wiley & Sons Inc.}}, series = {{Magnetic Resonance in Medicine}}, title = {{Perfusion quantification by model-free arterial spin labeling using nonlinear stochastic regularization deconvolution.}}, url = {{https://lup.lub.lu.se/search/files/1635639/8682585.pdf}}, doi = {{10.1002/mrm.24587}}, volume = {{70}}, year = {{2013}}, }