Waveguide effects and implications for cardiac magnetic resonance elastography

A finite element study

Armando Manduca, T. L. Rossman, D. S. Lake, K. J. Glaser, Arvin Forghanian-Arani, S. P. Arunachalam, P. J. Rossman, Joshua D Trazasko, Richard Lorne Ehman, M. (Dan) Dragomir Daescu, Philip A Araoz

Research output: Contribution to journalArticle

Abstract

Magnetic resonance elastography (MRE) is increasingly being applied to thin or small structures in which wave propagation is dominated by waveguide effects, which can substantially bias stiffness results with common processing approaches. The purpose of this work was to investigate the importance of such biases and artifacts on MRE inversion results in: (i) various idealized 2D and 3D geometries with one or more dimensions that are small relative to the shear wavelength; and (ii) a realistic cardiac geometry. Finite element models were created using simple 2D geometries as well as a simplified and a realistic 3D cardiac geometry, and simulated displacements acquired by MRE from harmonic excitations from 60 to 220 Hz across a range of frequencies. The displacement wave fields were inverted with direct inversion of the Helmholtz equation with and without the application of bandpass filtering and/or the curl operator to the displacement field. In all geometries considered, and at all frequencies considered, strong biases and artifacts were present in inversion results when the curl operator was not applied. Bandpass filtering without the curl was not sufficient to yield accurate recovery. In the 3D geometries, strong biases and artifacts were present in 2D inversions even when the curl was applied, while only 3D inversions with application of the curl yielded accurate recovery of the complex shear modulus. These results establish that taking the curl of the wave field and performing a full 3D inversion are both necessary steps for accurate estimation of the shear modulus both in simple thin-walled or small structures and in a realistic cardiac geometry when using simple inversions that neglect the hydrostatic pressure term. In practice, sufficient wave amplitude, signal-to-noise ratio, and resolution will be required to achieve accurate results.

Original languageEnglish (US)
JournalNMR in Biomedicine
DOIs
StateAccepted/In press - Jan 1 2018

Fingerprint

Elasticity Imaging Techniques
Magnetic resonance
Artifacts
Waveguides
Geometry
Hydrostatic Pressure
Signal-To-Noise Ratio
Elastic moduli
Recovery
Helmholtz equation
Hydrostatic pressure
Wave propagation
Mathematical operators
Signal to noise ratio
Stiffness
Wavelength
Processing

Keywords

  • cardiac elastography
  • cardiac MRE
  • elastography
  • finite element modeling
  • MR elastography
  • myocardial stiffness
  • shear waves
  • waveguide

ASJC Scopus subject areas

  • Molecular Medicine
  • Radiology Nuclear Medicine and imaging
  • Spectroscopy

Cite this

@article{1130db815d6f481c84398a1ee3f99cb1,
title = "Waveguide effects and implications for cardiac magnetic resonance elastography: A finite element study",
abstract = "Magnetic resonance elastography (MRE) is increasingly being applied to thin or small structures in which wave propagation is dominated by waveguide effects, which can substantially bias stiffness results with common processing approaches. The purpose of this work was to investigate the importance of such biases and artifacts on MRE inversion results in: (i) various idealized 2D and 3D geometries with one or more dimensions that are small relative to the shear wavelength; and (ii) a realistic cardiac geometry. Finite element models were created using simple 2D geometries as well as a simplified and a realistic 3D cardiac geometry, and simulated displacements acquired by MRE from harmonic excitations from 60 to 220 Hz across a range of frequencies. The displacement wave fields were inverted with direct inversion of the Helmholtz equation with and without the application of bandpass filtering and/or the curl operator to the displacement field. In all geometries considered, and at all frequencies considered, strong biases and artifacts were present in inversion results when the curl operator was not applied. Bandpass filtering without the curl was not sufficient to yield accurate recovery. In the 3D geometries, strong biases and artifacts were present in 2D inversions even when the curl was applied, while only 3D inversions with application of the curl yielded accurate recovery of the complex shear modulus. These results establish that taking the curl of the wave field and performing a full 3D inversion are both necessary steps for accurate estimation of the shear modulus both in simple thin-walled or small structures and in a realistic cardiac geometry when using simple inversions that neglect the hydrostatic pressure term. In practice, sufficient wave amplitude, signal-to-noise ratio, and resolution will be required to achieve accurate results.",
keywords = "cardiac elastography, cardiac MRE, elastography, finite element modeling, MR elastography, myocardial stiffness, shear waves, waveguide",
author = "Armando Manduca and Rossman, {T. L.} and Lake, {D. S.} and Glaser, {K. J.} and Arvin Forghanian-Arani and Arunachalam, {S. P.} and Rossman, {P. J.} and Trazasko, {Joshua D} and Ehman, {Richard Lorne} and {Dragomir Daescu}, {M. (Dan)} and Araoz, {Philip A}",
year = "2018",
month = "1",
day = "1",
doi = "10.1002/nbm.3996",
language = "English (US)",
journal = "NMR in Biomedicine",
issn = "0952-3480",
publisher = "John Wiley and Sons Ltd",

}

TY - JOUR

T1 - Waveguide effects and implications for cardiac magnetic resonance elastography

T2 - A finite element study

AU - Manduca, Armando

AU - Rossman, T. L.

AU - Lake, D. S.

AU - Glaser, K. J.

AU - Forghanian-Arani, Arvin

AU - Arunachalam, S. P.

AU - Rossman, P. J.

AU - Trazasko, Joshua D

AU - Ehman, Richard Lorne

AU - Dragomir Daescu, M. (Dan)

AU - Araoz, Philip A

PY - 2018/1/1

Y1 - 2018/1/1

N2 - Magnetic resonance elastography (MRE) is increasingly being applied to thin or small structures in which wave propagation is dominated by waveguide effects, which can substantially bias stiffness results with common processing approaches. The purpose of this work was to investigate the importance of such biases and artifacts on MRE inversion results in: (i) various idealized 2D and 3D geometries with one or more dimensions that are small relative to the shear wavelength; and (ii) a realistic cardiac geometry. Finite element models were created using simple 2D geometries as well as a simplified and a realistic 3D cardiac geometry, and simulated displacements acquired by MRE from harmonic excitations from 60 to 220 Hz across a range of frequencies. The displacement wave fields were inverted with direct inversion of the Helmholtz equation with and without the application of bandpass filtering and/or the curl operator to the displacement field. In all geometries considered, and at all frequencies considered, strong biases and artifacts were present in inversion results when the curl operator was not applied. Bandpass filtering without the curl was not sufficient to yield accurate recovery. In the 3D geometries, strong biases and artifacts were present in 2D inversions even when the curl was applied, while only 3D inversions with application of the curl yielded accurate recovery of the complex shear modulus. These results establish that taking the curl of the wave field and performing a full 3D inversion are both necessary steps for accurate estimation of the shear modulus both in simple thin-walled or small structures and in a realistic cardiac geometry when using simple inversions that neglect the hydrostatic pressure term. In practice, sufficient wave amplitude, signal-to-noise ratio, and resolution will be required to achieve accurate results.

AB - Magnetic resonance elastography (MRE) is increasingly being applied to thin or small structures in which wave propagation is dominated by waveguide effects, which can substantially bias stiffness results with common processing approaches. The purpose of this work was to investigate the importance of such biases and artifacts on MRE inversion results in: (i) various idealized 2D and 3D geometries with one or more dimensions that are small relative to the shear wavelength; and (ii) a realistic cardiac geometry. Finite element models were created using simple 2D geometries as well as a simplified and a realistic 3D cardiac geometry, and simulated displacements acquired by MRE from harmonic excitations from 60 to 220 Hz across a range of frequencies. The displacement wave fields were inverted with direct inversion of the Helmholtz equation with and without the application of bandpass filtering and/or the curl operator to the displacement field. In all geometries considered, and at all frequencies considered, strong biases and artifacts were present in inversion results when the curl operator was not applied. Bandpass filtering without the curl was not sufficient to yield accurate recovery. In the 3D geometries, strong biases and artifacts were present in 2D inversions even when the curl was applied, while only 3D inversions with application of the curl yielded accurate recovery of the complex shear modulus. These results establish that taking the curl of the wave field and performing a full 3D inversion are both necessary steps for accurate estimation of the shear modulus both in simple thin-walled or small structures and in a realistic cardiac geometry when using simple inversions that neglect the hydrostatic pressure term. In practice, sufficient wave amplitude, signal-to-noise ratio, and resolution will be required to achieve accurate results.

KW - cardiac elastography

KW - cardiac MRE

KW - elastography

KW - finite element modeling

KW - MR elastography

KW - myocardial stiffness

KW - shear waves

KW - waveguide

UR - http://www.scopus.com/inward/record.url?scp=85052378165&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85052378165&partnerID=8YFLogxK

U2 - 10.1002/nbm.3996

DO - 10.1002/nbm.3996

M3 - Article

JO - NMR in Biomedicine

JF - NMR in Biomedicine

SN - 0952-3480

ER -