Premium
Phase processing for quantitative susceptibility mapping of regions with large susceptibility and lack of signal
Author(s) -
Fortier Véronique,
Levesque Ives R.
Publication year - 2018
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.26989
Subject(s) - quantitative susceptibility mapping , computer science , context (archaeology) , artifact (error) , algorithm , signal processing , signal (programming language) , phase (matter) , image quality , reduction (mathematics) , artificial intelligence , pattern recognition (psychology) , mathematics , physics , image (mathematics) , digital signal processing , medicine , paleontology , geometry , radiology , quantum mechanics , magnetic resonance imaging , computer hardware , biology , programming language
Purpose Phase processing impacts the accuracy of quantitative susceptibility mapping (QSM). Techniques for phase unwrapping and background removal have been proposed and demonstrated mostly in brain. In this work, phase processing was evaluated in the context of large susceptibility variations (Δχ) and negligible signal, in particular for susceptibility estimation using the iterative phase replacement (IPR) algorithm. Methods Continuous Laplacian, region‐growing, and quality‐guided unwrapping were evaluated. For background removal, Laplacian boundary value (LBV), projection onto dipole fields (PDF), sophisticated harmonic artifact reduction for phase data (SHARP), variable‐kernel sophisticated harmonic artifact reduction for phase data (V‐SHARP), regularization enabled sophisticated harmonic artifact reduction for phase data (RESHARP), and 3D quadratic polynomial field removal were studied. Each algorithm was quantitatively evaluated in simulation and qualitatively in vivo. Additionally, IPR‐QSM maps were produced to evaluate the impact of phase processing on the susceptibility in the context of large Δχ with negligible signal. Results Quality‐guided unwrapping was the most accurate technique, whereas continuous Laplacian performed poorly in this context. All background removal algorithms tested resulted in important phase inaccuracies, suggesting that techniques used for brain do not translate well to situations where large Δχ and no or low signal are expected. LBV produced the smallest errors, followed closely by PDF. Conclusion Results suggest that quality‐guided unwrapping should be preferred, with PDF or LBV for background removal, for QSM in regions with large Δχ and negligible signal. This reduces the susceptibility inaccuracy introduced by phase processing. Accurate background removal remains an open question. Magn Reson Med 79:3103–3113, 2017. © 2017 International Society for Magnetic Resonance in Medicine.