Physics-informed neural networks for myocardial perfusion MRI quantification

Rudolf L. M. van Herten, Amedeo Chiribiri, Marcel Breeuwer, Mitko Veta, Cian M. Scannell (Corresponding author)

Research output: Contribution to journalArticleAcademic

4 Downloads (Pure)

Abstract

Tracer-kinetic models allow for the quantification of kinetic parameters such as blood flow from dynamic contrast-enhanced magnetic resonance (MR) images. Fitting the observed data with multi-compartment exchange models is desirable, as they are physiologically plausible and resolve directly for blood flow and microvascular function. However, the reliability of model fitting is limited by the low signal-to-noise ratio, temporal resolution, and acquisition length. This may result in inaccurate parameter estimates. This study introduces physics-informed neural networks (PINNs) as a means to perform myocardial perfusion MR quantification, which provides a versatile scheme for the inference of kinetic parameters. These neural networks can be trained to fit the observed perfusion MR data while respecting the underlying physical conservation laws described by a multi-compartment exchange model. Here, we provide a framework for the implementation of PINNs in myocardial perfusion MR. The approach is validated both in silico and in vivo. In the in silico study, an overall reduction in mean-squared error with the ground-truth parameters was observed compared to a standard non-linear least squares fitting approach. The in vivo study demonstrates that the method produces parameter values comparable to those previously found in literature, as well as providing parameter maps which match the clinical diagnosis of patients.
Original languageEnglish
Article number2011:12844v2
Number of pages18
JournalarXiv
Volume2020
Publication statusPublished - 25 Nov 2020

Bibliographical note

Submitted to Medical Image Analysis

Fingerprint Dive into the research topics of 'Physics-informed neural networks for myocardial perfusion MRI quantification'. Together they form a unique fingerprint.

Cite this