Recurrent inference machines as inverse problem solvers for MR relaxometry

• Published in: Medical image analysis 2021-12, Vol.74, p.102220-102220, Article 102220
• Main Authors: Sabidussi, E.R., Klein, S., Caan, M.W.A., Bazrafkan, S., den Dekker, A.J., Sijbers, J., Niessen, W.J., Poot, D.H.J.
• Format: Article
• Language: English
• Subjects: Artificial neural networks; Deep learning; Humans; Image Processing, Computer-Assisted; Inverse problems; Inverse reconstruction; Iterative methods; Magnetic Resonance Imaging; Mapping; Mathematical models; Maximum likelihood estimators; Monte Carlo Method; Neural networks; Neural Networks, Computer; Phantoms, Imaging; Quantitative MRI; Recurrent inference machines; Relaxometry; Signal processing; Source code; Statistical inference; Statistical methods
• ISSN: 1361-8415; 1361-8423
• DOI: 10.1016/j.media.2021.102220

•Recurrent Inference Machines iteratively reconstruct T1 and T2 relaxometry maps.•Higher precision, comparable accuracy as a state-of-the-art mapping method.•Generalizes across anatomies and slight variations in acquisition settings.•Inference is 150 times faster than conventional state-of-the-art mapping method.•Simulated data for training reduces requirements on costly image acquisition. [Display omitted] In this paper, we propose the use of Recurrent Inference Machines (RIMs) to perform T1 and T2 mapping. The RIM is a neural network framework that learns an iterative inference process based on the signal model, similar to conventional statistical methods for quantitative MRI (QMRI), such as the Maximum Likelihood Estimator (MLE). This framework combines the advantages of both data-driven and model-based methods, and, we hypothesize, is a promising tool for QMRI. Previously, RIMs were used to solve linear inverse reconstruction problems. Here, we show that they can also be used to optimize non-linear problems and estimate relaxometry maps with high precision and accuracy. The developed RIM framework is evaluated in terms of accuracy and precision and compared to an MLE method and an implementation of the Residual Neural Network (ResNet). The results show that the RIM improves the quality of estimates compared to the other techniques in Monte Carlo experiments with simulated data, test-retest analysis of a system phantom, and in-vivo scans. Additionally, inference with the RIM is 150 times faster than the MLE, and robustness to (slight) variations of scanning parameters is demonstrated. Hence, the RIM is a promising and flexible method for QMRI. Coupled with an open-source training data generation tool, it presents a compelling alternative to previous methods.