Comparative review of algorithms and methods for chemical‐shift‐encoded quantitative fat‐water imaging

Abstract

PurposeTo propose a standardized comparison between state‐of‐the‐art open‐source fat‐water separation algorithms for proton density fat fraction (PDFF) and quantification using an open‐source multi‐language toolbox.MethodsEight recent open‐source fat‐water separation algorithms were compared in silico, in vitro, and in vivo. Multi‐echo data were synthesized with varying fat‐fractions, B0 off‐resonance, SNR and TEs. Experimental evaluation was conducted using calibrated fat‐water phantoms acquired at 3T and multi‐site open‐source phantoms data. Algorithms' performances were observed on challenging in vivo datasets at 3T. Finally, reconstruction algorithms were investigated with different fat spectra to evaluate the importance of the fat model.ResultsIn silico and in vitro results proved most algorithms to be not sensitive to fat‐water swaps and offsets with five or more echoes. However, two methods remained inaccurate even with seven echoes and SNR = 50, and two other algorithms' precision depended on the echo spacing scheme (p < 0.05). The remaining four algorithms provided reliable performances with limits of agreement under 2% for PDFF and 6 s−1 for . The choice of fat spectrum model influenced quantification of PDFF mildly (<2% bias) and of more severely, with errors up to 20 s−1.ConclusionIn promoting standardized comparisons of MRI‐based fat and iron quantification using chemical‐shift encoded multi‐echo methods, this benchmark work has revealed some discrepancies between recent approaches for PDFF and mapping. Explicit choices and parameterization of the fat‐water algorithm appear necessary for reproducibility. This open‐source toolbox further enables the user to optimize acquisition parameters by predicting algorithms' margins of errors.

Comparison of Vendor-Independent Software Tools for Liver Proton Density Fat Fraction Estimation at 1.5 T

“…The overall bias was smallest, with MAG-R (0.01%), followed closely by QPBO (0.08%), and it was well below 1% with all methods. Our results align with previous studies, which detected a similarly strong correlation and low bias between estimated and true PDFF in fat–water phantoms using CSE-MRI [ 32 , 33 , 38 ]. Moreover, open-source algorithms demonstrated accuracy comparable to commercially available CSE-MRI methods developed by major vendors [ 7 ].…”

Trajectory correction enables free-running chemical shift encoded imaging for accurate cardiac proton-density fat fraction quantification at 3T

“…Synthetic CSE-CMR data y were simulated with parameters matching practical acquisitions, for monopolar readouts: echo time number/first value/spacing: NTE/TE1/ΔTE = 8/1.16 ms/1.96 ms, and for bipolar readouts: NTE/TE1/ΔTE = 13/1.12/1.07 ms, and a noisy fat-water signal model defined as: with W and F corresponding to normalized water and fat magnitude signals, the off-resonance frequency, the common initial phase which holds only for low flip angles [26] , R 2 * = 50 s −1 the transversal decay, η(t) the complex Gaussian noise (with SNR = 10 or 50) and , the relative normalized amplitudes ( ), and frequency offsets of a 10-peaks resolved subcutaneous fat spectrum [27] , [28] , respectively. The virtual CSE-CMR data were synthesized as volumes using an open-source toolbox [28] with along the x-axis, PDFF varied from 0% to 100% with 1% step, along the y-axis, f 0 was uniformly distributed from −200 Hz to 200 Hz with 4 Hz step, and the z-axis consisted in 100 repetitions. Synthetic volumes were normalized based on 99% of the maximum of the first echo.…”