Fig. 1. Main directions and typical advances achieved by metamaterials to improve MRI. Three key aspects discussed in this review are metamaterials with exotic effective electromagnetic parameters [including negative-permeability metamaterials (NPMs), high-permeability/permittivity metamaterials (HPMs), and zero-index metamaterials (ZIMs)], metamaterials with peculiar dispersion relations [including magneto-inductive metamaterials (MIMs), wire-medium metamaterials (WMMs), and composed right-/left-handed transmission line metamaterials (TLMs)], and metamaterial resonators with tailored electromagnetic field distributions. There are two fundamental types of metamaterial resonators: those using optimized structures possessing different characteristics (e.g., flexible geometry, dual-band response, and volumetric arrangement) and those with integrated tunability and/or nonlinearity, featuring intelligence, adaptivity, or auxeticity. The figures are reproduced with permission from Ref. 35 © 2017 Springer Nature, licensed under a CC-BY 4.0 International License; Ref. 36 © 2017 American Chemical Society, licensed under a CC-BY-ND 4.0 License; Ref. 37 © 2018 Wiley-VCH; Ref. 38 © 2021 Wiley-VCH; Ref. 39 © 2021 Wiley-VCH; and Ref. 40 © 2019 Wiley-VCH.

Fig. 2. Principle of MRI. (a) Several nuclei are randomly orientated, thus exhibiting zero NMV. (b) After the introduction of the static external magnetic field $B_0$, more nuclei tend to align in the direction of $B_0$, contributing to a nonzero NMV. Inexactly aligned magnetic moments will precess around $B_0$, as will the NMV. (c) The RF magnetic field $B_1$ can cause the NMV to flip into the transverse direction and become detectable. The flip angle (FA) is dependent on the strength and duration of $B_1$. (d) Schematic of an open MRI machine. The shaded arcs indicate the radial positions of different components integrated within the scanner.

Fig. 3. Metamaterials with extreme electromagnetic parameters for MRI. (a) Schematic of a single Swiss roll. (b) Layout of the MRI imaging experiment. (c) 0.5-T MRI images of a thumb acquired for three cases: with the body coil in transceiver mode as a reference (left panel), and with a local coil as the receiver in the absence (middle panel) or presence of Swiss rolls (right panel). (d) Photograph of the M-shaped antenna. (e) Effective permeability as the function of frequency. (f) The field pattern observed above the parameter-optimized Swiss rolls (outlined in red). (g) Sketch of the experimental setup for the NPM lens and (h) obtained 1.5-T MRI images with and without the $\mu =-1$ metamaterial lens between two ankles. The SNR in the ROI indicated by a circle was increased by 2.6 times in the presence of the lens realized by NPMs. (i) Sketch of the magnetic-induction lines for a single coil with a $\mu =0$ slab (left panel) and with a $\mu \to \infty$ slab (right panel). SNR maps measured by a 1.5-T receiver coil with and without (j) the $\mu =0$ slab perpendicular to the coil or (k) the $\mu \to \infty$ slab parallel to the coil. The figures are reproduced with permission from (a)–(c) Ref. 66 © 2001 AAAS, (d)–(f) Ref. 67 © 2003 Optica, (g) and (h) Ref. 68 © 2010 Elsevier, and (j) and (k) Ref. 69 © 2011 American Institute of Physics (AIP).

Fig. 4. MIMs and WMMs for MRI. (a) Sketch of a 1D MI waveguide consisting of coaxial capacitively loaded loops. (b) Equivalent circuit for the structure. (c) Photograph of a distorted flexible octagonal MI ring detector. (d) 1.5-T MRI experiment images of a pomelo fruit (upper images) and water-filled bottle (lower images) obtained from an undistorted and distorted MI ring detector. (e) Schematic diagram of WMM containing a dense array of metal wires. (f) Dispersions of a WMM formed from silver nanorods in three cases: with the PEC approximation (gray dashed lines), with spatial dispersion (thick green lines), and without spatial dispersion (thin magenta lines). 3-T MRI images of different phantoms obtained with a loop receiver coil and transferred (g) without the WMM or (h) with a 63 deg curved WMM. The position of the loop coil is indicated by an orange rectangle. The figures are reproduced with permission from (b) Ref. 86 © 2006 AIP, (c) and (d) Ref. 87 © 2010 Elsevier, (e) and (f) Ref. 88 © 2012 Wiley-VCH, and (g) and (h) Ref. 89 © 2009 Elsevier.

Fig. 5. CRLH-TL metamaterials for MRI. (a) Schematic of a TL, in which the electric and magnetic field distributions are marked by the black solid and red dashed lines, respectively. (b) TL model of a circuit-based metamaterial. (c) The corresponding equivalent-circuit model of the 1D TL structure. (d) Simulated model and magnetic field distribution of a CRLH ring antenna with the current flowing along the surface at resonance. (e) Schematic diagram of the adaptivity of a CRLH ring antenna system realized by tailored current distribution: unidirectional transmission (antenna 1), coherent amplification (antennas 2 and 3), and active attenuation (antenna 4). The frequency is exaggerated for visibility. (f) CRLH ring antenna system with on-demand magnetic field distribution (top panel) for a larynx imaging case. The amplitudes and phases of the in-phase and quadrature current excitations of each unit cell are shown in the bottom panels. The figures are reproduced with permission from (d) and (e) Ref. 103 © 2011 IEEE, and (f) Ref. 104 © 2013 IEEE.

Fig. 6. SNR enhancement ratios of typical MMRs. Different types are distinguished by symbols and colors. For planar MMRs, the ratio is measured 2 cm above the surface; for volumetric MMRs, the ratio is measured in the ROI.

Fig. 7. General MMRs for MRI. (a) Schematic view of a planar MMR composed of metallic wires. (b) The electromagnetic field distribution above this MMR for the fundamental mode. (c) 1.5-T ex vivo MRI images of fish with scanning times of 1020 s [(i) and (ii)] or 120 s [(iii) and (iv)] in the absence of the MMR [(i) and (iii)] or in the presence of the MMR [(ii) and (iv)]. (d) Sketch of the design of a volumetric MMR. (e) Illustration of possible clinical scenarios for deploying volumetric MMRs: supine position (left panel) and superman position (right panel). (f) and (g) Experimental setups and SNR maps of 1.5-T wrist MRI of two volunteers in the superman position with different settings: whole-body birdcage coil in transceiver mode in the presence of volumetric MMR (the first and third columns); extremity coil in transceiver mode (the second and fourth columns). The figures are reproduced with permission from (a) and (c) Ref. 59 © 2016 Wiley-VCH; (b) Ref. 116 © 2018 Elsevier, licensed under a CC-BY 4.0 International License; (d) and (e) Ref. 37 © 2018 Wiley-VCH; and (f) and (g) Ref. 117 © 2022 Elsevier.

Fig. 8. Tunable MMRs for MRI. (a) Sketch of an MMR with tunable water filling factor. (b) Configuration of an MMR with movable metallic patches and an on-demand adjustable wedge shape. (c) Illustrations of structural deformation between adjacent unit cells and of a complete 2D auxetic MMR. (d) Similar to (c), but for the 3D case. The figures are reproduced with permission from (a) Ref. 126 © 2018 American Physical Society, (b) Ref. 127 © 2019 AIP, and (c) and (d) Ref. 38 © 2021 Wiley-VCH.

Fig. 9. Nonlinear MMRs for MRI. (a) Conceptual illustration of the working principle for linear and nonlinear MMRs. (b) Schematic diagram of MRI pulse sequence and response of nonlinear MMRs. (c) Sketch of the configuration of a nonlinear MMR consisting of a helix array and a varactor-SRR. (d) 3-T MRI images of a mineral-oil phantom captured for three cases: body coil only (top left), and body coil with a nonlinear (top right) and linear (bottom left) MMR. The relevant SNR enhancement ratio curves as functions of distance are also shown (bottom right). (e) Experimental setups and images of a 1.5-T wrist MRI using different sequences: with flexible coil (top panels); and with adaptative cylindrical nonlinear MMR (bottom panels). The figures are reproduced with permission from (a), (c), and (d) Ref. 40 © 2019 Wiley-VCH, and (e) Ref. 39 © 2021 Wiley-VCH.