Recurrent Inference Machines as Inverse Problem Solvers for MR Relaxometry

Emanoel Ribeiro Sabidussi, Stefan Klein, Matthan W. A. Caan, Shabab Bazrafkan, Arjan J. den Dekker, Jan Sijbers, Wiro Niessen, Dirk Poot

Keywords: Quantitative MRI, Relaxometry, Deep Learning, Mapping, Recurrent Inference Machines

TL;DR: In this paper we present a novel, deep learning based methodology for MR Relaxometry using Recurrent Inference Machines

Abstract: In this work, we propose the use of Recurrent Inference Machines (RIMs) to perform $T_1$ mapping. The RIM is a neural network framework that learns an iterative inference process using a model of the signal, similar to conventional statistical methods for quantitative MRI (QMRI), such as the Maximum Likelihood Estimator (MLE). Previously, RIMs were used to solve linear inverse reconstruction problems. Here, we show that they can also be used to optimize non-linear problems. The developed RIM framework is evaluated in terms of accuracy and precision and compared to an MLE method and an implementation of the ResNet. The results show that, compared to the other techniques in Monte Carlo experiments with simulated data, the RIM improves the precision of estimates without compromising in accuracy.

Paper Type: both

Primary Subject Area: Image Acquisition and Reconstruction

Secondary Subject Area: Application: Radiology

Paper Status: based on accepted/submitted journal paper

Source Code Url: https://gitlab.com/e.ribeirosabidussi/emcqmri_relaxometry

Data Set Url: https://gitlab.com/e.ribeirosabidussi/emcqmri_relaxometry

Registration: I acknowledge that publication of this at MIDL and in the proceedings requires at least one of the authors to register and present the work during the conference.

Authorship: I confirm that I am the author of this work and that it has not been submitted to another publication before.