High-Permittivity Pad Design for Dielectric Shimming in Magnetic Resonance Imaging Using Projection-Based Model Reduction and a Nonlinear Optimization Scheme
Abstract:Abstract-Inhomogeneities in the transmit RF magnetic field (B + 1 ) reduce the quality of MR images. This quality can be improved by using high-permittivity pads that tailor the B + 1 fields. The design of an optimal pad is application-specific and not straightforward, and would therefore benefit from a systematic optimization approach. In this paper, we propose such a method to efficiently design dielectric pads. To this end, a projection based model order reduction technique is used that significantly decrea… Show more
“…Magnetic resonance imaging (MRI) [1] is an essential medical diagnostic tool which is used for observing the tissue changes of the patients in the absence of ionizing radiation. It provides high-resolution images that contain important anatomical information.…”
Existing model-based or data-driven methods have achieved a high-quality reconstruction in compressive sensing magnetic resonance imaging (CS-MRI). However, most methods are designed for a specific type of sampling mask or sampling rate while ignoring the existence of external noise, resulting in poor robustness. In this work, we propose a probabilistic model-based method based on Laplacian scale mixture (LSM) modeling and denoising based approximate message passing (D-AMP) algorithm to address this issue. Sparse coefficients of similar packed patches are modeled with LSM distribution to exploit the nonlocal self-similarity prior of MR image, and a maximum a posterior estimation problem for sparse coding is formulated. It is shown that both hidden scale parameters i.e. variances of sparse coefficients and location parameters can be jointly estimated along with the unknown sparse coefficients via the method of alternating optimization. Moreover, the variance of noise is also iteratively updated based on maximum likelihood estimation. With plug-and-play prior method, the above structured sparse coding procedure can be regarded as a nonlocal filtering operation and be incorporated into DAMP for MR image reconstruction. Owing to the power of our nonlocal filtering which takes both signal and noise estimation into account, the proposed method not only outperforms many state-of-the-art methods for most situations of observation, but also delivers the best qualitative reconstruction results with finer details and less artifacts in experiments. INDEX TERMS Compressive sensing magnetic resonance imaging (CS-MRI), Laplacian scale mixture (LSM), denoising based approximate message passing (D-AMP), nonlocal filtering.
“…Magnetic resonance imaging (MRI) [1] is an essential medical diagnostic tool which is used for observing the tissue changes of the patients in the absence of ionizing radiation. It provides high-resolution images that contain important anatomical information.…”
Existing model-based or data-driven methods have achieved a high-quality reconstruction in compressive sensing magnetic resonance imaging (CS-MRI). However, most methods are designed for a specific type of sampling mask or sampling rate while ignoring the existence of external noise, resulting in poor robustness. In this work, we propose a probabilistic model-based method based on Laplacian scale mixture (LSM) modeling and denoising based approximate message passing (D-AMP) algorithm to address this issue. Sparse coefficients of similar packed patches are modeled with LSM distribution to exploit the nonlocal self-similarity prior of MR image, and a maximum a posterior estimation problem for sparse coding is formulated. It is shown that both hidden scale parameters i.e. variances of sparse coefficients and location parameters can be jointly estimated along with the unknown sparse coefficients via the method of alternating optimization. Moreover, the variance of noise is also iteratively updated based on maximum likelihood estimation. With plug-and-play prior method, the above structured sparse coding procedure can be regarded as a nonlocal filtering operation and be incorporated into DAMP for MR image reconstruction. Owing to the power of our nonlocal filtering which takes both signal and noise estimation into account, the proposed method not only outperforms many state-of-the-art methods for most situations of observation, but also delivers the best qualitative reconstruction results with finer details and less artifacts in experiments. INDEX TERMS Compressive sensing magnetic resonance imaging (CS-MRI), Laplacian scale mixture (LSM), denoising based approximate message passing (D-AMP), nonlocal filtering.
“…Furthermore, we note that the stability of our solution against a range of w b values implies the robustness of our evanescent mode-shaping solution to the shape or displacement of the used phantom. In the optimization process, only three factors (effectively two) need to be considered for homogenization, which is less complex than the pad optimization scheme in other studies 30 , 31 . Figure 4 Effects of the air-gap potential barrier width w b on the mode.…”
The notion of mode shaping based on evanescent coupling has been successfully applied in various fields of optics, such as in the dispersion engineering of optical waveguides. Here, we show that the same concept provides an opportunity for the seemingly different field of ultra-high-field MRI, addressing transmit RF magnetic field (B1+) inhomogeneity. In this work, treating the human phantom as a resonator, we employ an evanescently coupled high-index cladding layer to study the effects of the auxiliary potential on shaping the B1+ field distribution inside the phantom. Controlling the strength and coupling of the auxiliary potential ultimately determining the hybridized mode, we successfully demonstrate the global 2D homogenization of axial B1+ for a simplified cylindrical phantom and for a more realistic phantom of spheroidal geometry. The mode-shaping potentials with a magnetic permeability or material loss are also tested to offer additional degrees of freedom in the selection of materials as well as in the manipulation of the B1+ distribution, opening up the possibility of B1+ homogenization for 3D MRI scanning.
“…Furthermore, the pad design domain was divided into small subdomains, on average 2.5 cm in width and 2.3 cm in height, by assigning the same material properties to grid edges that belong to a given non‐overlapping subdomain. To exploit the reduced solution space because of the subdomains and the parameterization, a reduced order model was created following the procedure described in van Gemert et al In this same work, it was shown that the reduced order model produces a negligible error. This allows for field computations that are up to 4 orders of magnitude faster than conventional EM solvers such as XFdtd.…”
Section: Methodsmentioning
confidence: 99%
“…In previous works, we have developed an advanced reduced order modeling technique to accelerate pad evaluations by characterizing stationary components such as the RF coil and body model in an offline‐stage and compressing the resulting model. This yielded up to 4 orders of magnitude of acceleration compared to commercial software and enabled the automated design of dielectric pads in under 1 min . The resulting framework is, however, body‐model‐specific, and optimization results obtained in a generic model are not likely to be optimal when the subject anatomy is substantially different, particularly in the case of fetal imaging.…”
Purpose
One of the main concerns in fetal MRI is the radiofrequency power that is absorbed both by the mother and the fetus. Passive shimming using high permittivity materials in the form of “dielectric pads” has previously been shown to increase the B1+ efficiency and homogeneity in different applications, while reducing the specific absorption rate (SAR). In this work, we study the effect of optimized dielectric pads for 3 pregnant models.
Methods
Pregnant models in the 3rd, 7th, and 9th months of gestation were used for simulations in a birdcage coil at 3T. Dielectric pads were optimized regions of interest (ROI) using previously developed methods for B1+ efficiency and homogeneity and were designed for 2 ROIs: the entire fetus and the brain of the fetus. The SAR was evaluated in terms of the whole‐body SAR, average SAR in the fetus and amniotic fluid, and maximum 10 g‐averaged SAR in the mother, fetus, and amniotic fluid.
Results
The optimized dielectric pads increased the transmit efficiency up to 55% and increased the B1+ homogeneity in almost every tested configuration. The B1+‐normalized whole‐body SAR was reduced by more than 31% for all body models. The B1+‐normalized local SAR was reduced in most scenarios by up to 62%.
Conclusion
Simulations have shown that optimized high permittivity pads can reduce SAR in pregnant subjects at the 3rd, 7th, and 9th month of gestation, while improving the transmit field homogeneity in the fetus. However, significantly more work is required to demonstrate that fetal imaging is safe under standard operating conditions.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
