Fourier neural operator for coupled Brown-Néel rotation model

Abstract

Modelling magnetization dynamics of magnetic nanoparticles (MNPs) is crucial to understand and predict their signal response in magnetic particle imaging (MPI). Coupled Brown-Néel rotation model expresses MNP magnetization as a system of ordinary differential equations (ODEs). However, numerical solution of these ODEs can be computationally intensive and time consuming using classical solvers. In this work, we propose a neural solver that utilizes a Fourier Neural Operator (FNO) to speed up the computation time for the coupled Brown-Néel rotation model. We show that the FNO model provides high signal fidelity with 5 orders of magnitude acceleration in computation time.