Artifact Correction and Signal Quantification in High Field Breast MRI

Michael J van Rijssel

Research output: ThesisDoctoral thesis 1 (Research UU / Graduation UU)


This thesis investigates artifact correction and signal quantification in high field breast MRI. We focus on dynamic contrast-enhanced MRI (DCE-MRI) and diffusion-weighted imaging (DWI). DCE-MRI is sensitive to inhomogeneities in the radiofrequency transmit (B1+) field. Chapter 2 investigates B1+ field characterization and Chapter 3 explores the possibilities of correcting DCE-MRI for the related contrast artifact. DWI is prone to spatial distortion artifacts due to inhomogeneities in the static magnetic (B0) field. In Chapter 4 we propose a new way to better correct for these distortions, specifically in the breast. Chapter 5 focuses on better quantification of the diffusion signal measured in DWI.

Chapter 2 proposes a fast and noise-free way to estimate the B1+ distribution when using local transmit coils at 7 T. In simulations, intersubject differences in local transmit B1+ fields of the breast were found to be comparable to the accuracy of B1+ mapping methods. Therefore, a generic template was proposed and tested in 15 healthy volunteers. In a subset of three volunteers, repeated measurements had an error of up to 15% of the nominal angle; this error range increased slightly by approximately 6% when using a B1+ template. Consequently, a single generic B1+ template suits subjects over a wide range of breast anatomies, eliminating the need for a time-consuming and noise-prone B1+ mapping protocol.

In Chapter 3 we use the template approach developed in Chapter 2 as a basis for DCE contrast correction. Using the template as a source of B1+ information, we investigated the correctable B1+ range post acquisition. A direct mapping from measured to true signal intensities was devised to limit noise amplification during correction. Simulations showed that the correctable B1+ range extends down to 43% of the nominal angle. The distribution of curve types in a 7 T patient dataset with a wide range of B1+ levels corresponded better to those reported in literature after correction.

Chapter 4 shows that the B0 field in breast has high discontinuities at gland-fat tissue interfaces. Therefore, we developed a distortion correction method that incorporates high-resolution off-resonance maps to better solve severe distortions at tissue interfaces. Quantitative comparisons showed an increase in conformity between corrected EPI images and a non-EPI high-bandwidth reference scan, both ex-vivo and in-vivo. All metrics showed a significant improvement when a high-resolution off-resonance map was used, in particular at tissue boundaries. It is this improvement at tissue interfaces, which is due to the use of high-resolution off-resonance maps that gives our method the advantage over existing distortion correction techniques.

Chapter 5 explores whether the phasor transform can aid in producing more stable mixed-signal parameter maps in DWI; first in the context of fixed-diffusivity fraction estimation and second in the context of fitting the intravoxel incoherent motion (IVIM) model. While the phasor-based approach for fixed-diffusivity fraction estimation didn’t improve upon simply solving a linear system, phasor-based IVIM fitting did produce more stable parameter maps for two parameters (the pseudodiffusion fraction f, and the diffusion constant D) compared to nonlinear fitting and segmented fitting.
Original languageEnglish
QualificationDoctor of Philosophy
Awarding Institution
  • University Medical Center (UMC) Utrecht
  • Pluim, JPW, Primary supervisor
  • Klomp, Dennis, Supervisor
  • Gilhuijs, Kenneth, Co-supervisor
Award date26 Nov 2019
Place of PublicationUtrecht
Print ISBNs978-90-393-7196-1
Publication statusPublished - 26 Nov 2019


  • Breast
  • MRI
  • 7T
  • high field
  • DCE
  • DWI
  • artifact correction
  • distortion correction
  • B1+
  • B0


Dive into the research topics of 'Artifact Correction and Signal Quantification in High Field Breast MRI'. Together they form a unique fingerprint.

Cite this