A general framework for hepatic iron overload quantification using MRI

Magnetic resonance imaging (MRI) has been considered for the quantification of iron overload in the liver. Iron overload was found to correlate with T2* measurement using T2* weighted images. In this work, we address the problem of iron overload estimation in the liver using MRI. We propose a general framework for all liver models proposed in the literature. The iron overload estimation task is then formulated as a minimization problem, and suitable regularization functions are assigned to the unknown model parameters. Subsequently, an alternating direction method of multipliers (ADMM) is used to estimate these unknown parameters. Three different models are derived, tested and compared; namely the single exponential (SEXP), the bi-exponential (BiEXP), and the exponential plus constant (CEXP). Simulations conducted using synthetic datasets indicate good correlation between estimated and ground truth T2* for all models. Moreover, the algorithms are evaluated using MRI scans of nine patients of different iron concentrations, using a 3-Tesla MRI scanner. The estimated T2* values of the proposed approaches are found to correlate with those obtained by the MRI scanner console. Moreover, the proposed approaches outperform several existing methods in the literature for iron overload estimation.(c) 2023 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY license (http://creativecommons .org /licenses /by /4 .0/).

Automated summary

This article proposes a general framework for estimating iron overload in the liver using magnetic resonance imaging (MRI), considering various liver models. The iron overload estimation is formulated as a minimization problem with regularization functions assigned to unknown model parameters. An alternating direction method of multipliers (ADMM) is applied for parameter estimation. Three models were tested and compared using synthetic datasets, showing good correlation with ground truth T2* measurements. The proposed approaches were also evaluated using MRI scans of patients, demonstrating better performance than existing methods.