Estimation of Field Inhomogeneity Map Following Magnitude-Based Ambiguity-Resolved Water-Fat Separation

TL;DR

This paper presents a novel method to accurately estimate magnetic field inhomogeneity maps from magnitude-based water-fat separation, improving robustness and agreement with complex-based methods, especially at high field strengths.

Contribution

The study introduces a least-squares matrix formulation for field map estimation from phase images, extending magnitude-based water-fat separation to include field inhomogeneity mapping.

Findings

Accurate field maps beyond double the complex-based limits.

High agreement with reference methods in UK Biobank data.

Robust performance at 3 T with inhomogeneities over 300 Hz.

Abstract

PURPOSE: To extend magnitude-based PDFF (Proton Density Fat Fraction) and $R_2^*$ mapping with resolved water-fat ambiguity to calculate field inhomogeneity (field map) using the phase images. THEORY: The estimation is formulated in matrix form, resolving the field map in a least-squares sense. PDFF and $R_2^*$ from magnitude fitting may be updated using the estimated field maps. METHODS: The limits of quantification of our voxel-independent implementation were assessed. Bland-Altman was used to compare PDFF and field maps from our method against a reference complex-based method on 152 UK Biobank subjects (1.5 T Siemens). A separate acquisition (3 T Siemens) presenting field inhomogeneities was also used. RESULTS: The proposed field mapping was accurate beyond double the complex-based limit range. High agreement was obtained between the proposed method and the reference in UK Biobank (PDFF bias = -0.03 %, LoA (limits of agreement) [-0.1,0.1] %; Field map bias = 0.06 Hz, LoA = [-0.2,0.3] Hz). Robust field mapping was observed at 3 T, for inhomogeneities over 300 Hz including rapid variation across edges. CONCLUSION: Field mapping following magnitude-based water-fat separation with resolved water-fat ambiguity was demonstrated in-vivo and showed potential at high field.