Premium
Robust water fat separated dual‐echo MRI by phase‐sensitive reconstruction
Author(s) -
Romu Thobias,
Dahlström Nils,
Leinhard Olof Dahlqvist,
Borga Magnus
Publication year - 2017
Publication title -
magnetic resonance in medicine
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.696
H-Index - 225
eISSN - 1522-2594
pISSN - 0740-3194
DOI - 10.1002/mrm.26488
Subject(s) - voxel , artifact (error) , computer science , artificial intelligence , phase (matter) , magnetic resonance imaging , robustness (evolution) , pattern recognition (psychology) , poisson distribution , swap (finance) , nuclear magnetic resonance , mathematics , physics , chemistry , radiology , statistics , medicine , finance , economics , biochemistry , quantum mechanics , gene
Purpose The purpose of this work was to develop and evaluate a robust water‐fat separation method for T1‐weighted symmetric two‐point Dixon data. Theory and Methods A method for water‐fat separation by phase unwrapping of the opposite‐phase images by phase‐sensitive reconstruction (PSR) is introduced. PSR consists of three steps; (1), identification of clusters of tissue voxels; (2), unwrapping of the phase in each cluster by solving Poisson's equation; and (3), finding the correct sign of each unwrapped opposite‐phase cluster, so that the water‐fat images are assigned the correct identities. Robustness was evaluated by counting the number of water‐fat swap artifacts in a total of 733 image volumes. The method was also compared to commercial software. Results In the water‐fat separated image volumes, the PSR method failed to unwrap the phase of one cluster and misclassified 10. One swap was observed in areas affected by motion and was constricted to the affected area. Twenty swaps were observed surrounding susceptibility artifacts, none of which spread outside the artifact affected regions. The PSR method had fewer swaps when compared to commercial software. Conclusion The PSR method can robustly produce water‐fat separated whole‐body images based on symmetric two‐echo spoiled gradient echo images, under both ideal conditions and in the presence of common artifacts. Magn Reson Med 78:1208–1216, 2017. © 2016 International Society for Magnetic Resonance in Medicine.