Magnetic resonance imaging (MRI) is a widely used noninvasive imaging modality with excellent diagnostic accuracy. A major determinant for advanced MRI is the strength of the main static magnetic field B0, which determines the ultimate signal-to-noise...
Magnetic resonance imaging (MRI) is a widely used noninvasive imaging modality with excellent diagnostic accuracy. A major determinant for advanced MRI is the strength of the main static magnetic field B0, which determines the ultimate signal-to-noise ratio (SNR) and spatial resolution. While most clinical MRI systems currently operate at a B0 strength of 1.5 or 3 Tesla (T), the growing demand for diagnostics with higher precision has led to the advent of ultrahigh field (UHF) MRI operating at 7T or higher.
However, UHF MRI introduces new challenges because the transmit radiofrequency (RF) magnetic field B1+ must operate at the Larmor frequency, which increases in proportion to the static magnetic field B0 – e.g., approximately 300 MHz at 7T. In case of the RF excitation frequency ≥ 300 MHz, the wavelength of the B1+ field inside human tissues becomes comparable to or even shorter than the dimensions of the human head, leading to the B1+ inhomogeneity. For example, if not mitigated, this effect results in field concentration near the center of the human head and signal attenuation in peripheral regions. Therefore, advanced UHF MRI necessitates addressing the following challenge: achieving homogeneous B1+ field while minimizing electromagnetic absorption within the biological resonator.
To manipulate the B1+ field distribution in the region of interest (ROI), passive structures, such as high-permittivity materials (HPMs) and metasurfaces, have been employed as clinically safe options. However, their impact largely remains restricted to local field enhancement in the vicinity of the structures, which can even lead to degradation of three-dimensional (3D) B1+ field homogeneity over the entire brain volume. As an alternative, the parallel RF transmission (pTx) actively modulates the B1+ field distribution in the ROI by independently controlling the magnitudes and phases of multiple RF sources. In this context, the pTx-based redistribution of the field can be interpreted as the complex-valued superposition of individual eigenmodes excited by multi-channel Tx coils. Despite the advantages in its flexibility, the pTx approach requires dedicated coil designs and complex hardware, limiting the compatibility with conventional MRI systems.
In this dissertation, we present a new metasurface design method for volumetric B1+ field homogenization, providing seamless integration into existing MRI systems. In essence, this approach can be considered as a pTx-like passive implementation with phase-controlled metasurfaces under the existence of near-field coupling to the transmitter. For this problem, analogous to the Green’s function method, we semi-analytically find an optimal complex-valued superposition of scattering responses by using gradient descent method combined with sequential pruning for stable implementation. The optimized field distribution is then realized through the weighted assembly of high-permittivity metasurface scatterers. Our proposal not only reveals a fundamental solution for volumetric B1+ field homogenization for UHF MRI, but also provides a general recipe for the noninvasive field homogenization in electromagnetic environments.