Learning Bloch Simulations for MR Fingerprinting by Invertible Neural Networks

TL;DR

This paper introduces an invertible neural network approach for MR fingerprinting that jointly learns the forward and backward processes, improving parameter estimation especially under challenging physical constraints.

Contribution

It proposes using invertible neural networks to learn both the Bloch simulation and parameter estimation processes in MR fingerprinting, enhancing accuracy and scalability.

Findings

Learning the forward process improves MR parameter estimation.

The approach is especially effective under physical restrictions.

INNs offer a feasible alternative to traditional backward-only NNs.

Abstract

Magnetic resonance fingerprinting (MRF) enables fast and multiparametric MR imaging. Despite fast acquisition, the state-of-the-art reconstruction of MRF based on dictionary matching is slow and lacks scalability. To overcome these limitations, neural network (NN) approaches estimating MR parameters from fingerprints have been proposed recently. Here, we revisit NN-based MRF reconstruction to jointly learn the forward process from MR parameters to fingerprints and the backward process from fingerprints to MR parameters by leveraging invertible neural networks (INNs). As a proof-of-concept, we perform various experiments showing the benefit of learning the forward process, i.e., the Bloch simulations, for improved MR parameter estimation. The benefit especially accentuates when MR parameter estimation is difficult due to MR physical restrictions. Therefore, INNs might be a feasible alternative to the current solely backward-based NNs for MRF reconstruction.