Metamaterial-based transmit and receive system for whole-body magnetic resonance imaging at ultra-high magnetic fields

Magnetic resonance imaging (MRI) at ultra-high fields (UHF), such as 7 T, provides an enhanced signal-to-noise ratio and has led to unprecedented high-resolution anatomic images and brain activation maps. Although a variety of radio frequency (RF) coil architectures have been developed for imaging at UHF conditions, they usually are specialized for small volumes of interests (VoI). So far, whole-body coil resonators are not available for commercial UHF human whole-body MRI systems. The goal of the present study was the development and validation of a transmit and receive system for large VoIs that operates at a 7 T human whole-body MRI system. A Metamaterial Ring Antenna System (MRAS) consisting of several ring antennas was developed, since it allows for the imaging of extended VoIs. Furthermore, the MRAS not only requires lower intensities of the irradiated RF energy, but also provides a more confined and focused injection of excitation energy on selected body parts. The MRAS consisted of several antennas with 50 cm inner diameter, 10 cm width and 0.5 cm depth. The position of the rings was freely adjustable. Conformal resonant right-/left-handed metamaterial was used for each ring antenna with two quadrature feeding ports for RF power. The system was successfully implemented and demonstrated with both a silicone oil and a water-NaCl-isopropanol phantom as well as in vivo by acquiring whole-body images of a crab-eating macaque. The potential for future neuroimaging applications was demonstrated by the acquired high-resolution anatomic images of the macaque’s head. Phantom and in vivo measurements of crab-eating macaques provided high-resolution images with large VoIs up to 40 cm in xy-direction and 45 cm in z-direction. The results of this work demonstrate the feasibility of the MRAS system for UHF MRI as proof of principle. The MRAS shows a substantial potential for MR imaging of larger volumes at 7 T UHF. This new technique may provide new diagnostic potential in spatially extended pathologies such as searching for spread-out tumor metastases or monitoring systemic inflammatory processes.


Introduction
Human whole-body magnetic resonance imaging (MRI) systems with a static magnetic flux density field (B 0 field) of either 1.5 T or 3 T are the workhorses of routine clinical applications [1][2][3][4][5]. During the last decade, a trend towards performing MRI at higher magnetic flux densities (B 0 ) is observed, since higher B 0 fields lead to both an improved signal-to-noise ratio (SNR) and an enhanced spatial resolution, thus providing for a more sensitive detection. Currently, these features of ultra-high field (UHF) MRI are especially exploited in functional MRI studies, where the brain activity is investigated. Recently, the certification process of human whole-body 7 T MRI systems for clinical diagnostics has been started. It is to be expected that clinical 7 T MRI examinations will become more frequent in near future. Although substantial progress has been made since the introduction of 7 T MR scanners in research some problems still remain and call for new solutions. In MRI performed at magnetic flux densities of up to 3 T, usually a body coil irradiates the required excitation energy (B 1 + field) into the object, whereas dedicated, smaller radio frequency (RF) coils are used for detection of the MR signals, which are ultimately transformed into the well-known MR images. The electromagnetic characteristic of body coils is known to depend on the resonant frequency, which is linearly dependent on the magnetic field strength. This was shown by Vaughn et al. where a transverse electromagnetic (TEM) body coil was developed and demonstrated at a 7 T MRI system [6] with the outcome of low efficiency and interference effects in the transverse plane. At lower field strengths body coils act as resistively loaded resonators that conduct the RF energy into the body [7]. By contrast, at higher magnetic field strengths (7 T and beyond), RF coils tend to behave like antennas. In the latter, the feeding power is spread over the surface of the body of the subject, from where it propagates into the inner parts of the body [8], thus mimicking radiation behavior. Usually, the body coil of standard clinical MRI systems is a large RF coil in birdcage architecture [9] that transmits the excitation energy in a highly homogeneous and efficient manner (Fig 1A). Systems equipped with such coils may provide large fields of view (FoV), even though smaller receive coils are typically used. Wholebody imaging is then achieved by moving the patient slowly through this optimal volume, i.e., using a moving table [10,11].
During the last decade MRI systems with static B 0 fields of 7 T or more [12] were introduced in research in order to increase the SNR and spatial resolution in functional MRI (fMRI) [13,14] and anatomic studies as well as in other promising areas [15]. However, application were mainly restricted to small or medium-sized volumes of interests (VoI) as at ultrahigh magnetic fields, the higher resolution and the enhanced SNR come with a decreased wavelength of the excitation field (B 1 + ) which is often as small or even smaller as the dimensions of the examined object. In this case, severe B 1 + field inhomogeneities may occur [16] due to standing wave effects. Furthermore, at UHF MRI the RF energy absorbed by the investigated body part is considerably higher when compared to MRI systems operating at lower B 0 fields. So far, the vendor-equipped UHF MRI systems support imaging of VoIs with dimensions determined by the characteristics of the transmit (Tx) and receive (Rx) coils. At UHF MRI volume or multi-element transmit coils are being employed for investigating the brain, [17] heart, [18] or knee [19]. The limitation to smaller VoIs is also reflected by the fact that body coils or arrays are not yet available for commercial 7 T (or higher) whole-body MRI. During the last years, however, efforts have been undertaken to develop concepts and RF coils that make larger VoIs accessible for UHF MRI. A step in this direction is the so-called travelling-wave (TW) excitation approach that makes use of the RF shield as part of the gradient system in combination with a simple patch antenna [20]. This approach yields a sufficiently uniform B 1 + field distribution for the imaging of a human extremity (leg) [20] or for neuroimaging of smaller non-human primates [21], thus enabling the imaging at VoIs of intermediate size. Unfortunately, the B 1 + transmit efficiency, i.e., the required RF energy to generate the B 1 + field is lower than that of both the body coils conventionally used in clinical 3 T MRI systems [22] and of the UHF body-part volume resonators of birdcage architecture [23,24]. Usually, birdcage volume coils confine the RF energy in the structure enclosed by the RF coil. The acquisition of whole-body images at UHF MRI requires the control of the RF energy introduced into the examined body region. This calls for novel conceptual solutions, like the one provided by the theory of conformal resonant right-/left-handed (CRLH) metamaterial ring antennas [25], which relies on the excitation of the fundamental transverse electric circular waveguide mode (TE 11 ) in the MRI scanner bore by an adaptive RF antenna system. This concept is briefly summarized in the next section. Whereas the theoretical foundations have been recently put forward [25] and examined in field simulations [26] no experimental realization or even an in vivo examination of a whole-body transmit system based on CRLH metamaterial ring antennas has been reported so far (Fig 1B).
In the present paper we describe the first experimental study of a potential whole-body imaging system based on CRLH metamaterial ring antennas. The rationale of our study is the following: We first show the design of CRLH metamaterial ring antennas and unravel the detailed functionalities of the developed metamaterial ring antenna system (MRAS). Than we show that CRLH metamaterial ring antennas are indeed appropriate for the construction and operation of a whole-body imaging system run in an UHF MRI system. Characteristic functionalities of the MRAS will be investigated in experiments, namely its high SNR, its ability to image large VoIs, as well as its efficient use of RF energy. To these purposes three experimental ring antenna setups were built, which consist of (1) a MRAS for measuring phantoms, (2) a MRAS for whole-body image acquisition of a macaque, and (3) a MRAS to allow high-resolution imaging of the head of a macaque. These setups differ in the geometric and functional arrangements of their ring antennas. We demonstrate that MRASs allow in vivo 7 T imaging of VoIs that are 4-6 times larger than that of standard 7 T imaging capability. The MRAS concept is also different from the standard volume body coil concept with its fixed architecture as the rings can be positioned and combined in a flexible way allowing different experimental setups. This will be shown in the present work and the outcome of these three experimental ring antenna setups will give an overview of how many ring antennas the Metamaterial Ring Antenna System should consists of to provide a high SNR and B 1 + transmit efficiency ( Fig 1B).
This new concept may also be of interest for diagnostic imaging as recently, the certification process of introducing human whole-body 7 T MRI systems into clinical routine has been started. Therefore, clinical 7 T MRI examinations will become more frequent in near future and the presented concept may support additional capabilities to use 7 T MRI for new diagnostic techniques such as whole-body imaging at UHF.

Metamaterial ring antennas
The CRLH metamaterial ring antennas where build in-house. They were designed to support a full wavelength resonance pattern along the circumference of the ring structure to match the surface-current distribution of the excited TE 11 mode.
The whole diameter of all three constructed ring antenna was 51 cm for a mobile versatile installation inside the 7 T MRI system bore (diameter = 60 cm) because an installation behind the MR bore housing wasn't possible. Each ring structure is formed by three (outer / intermediate / inner) metal layers, which encompass a spacing of 1.28 mm followed by 10 mm, where the outer and intermediate layer are part of a Rogers 3010 substrate (ε r = 10.2). The inner metal layer corresponds to an acrylic glass ring (inner diameter 50 cm) acting as the overall carrier. The resulting CRLH ring antenna has 24 unit cells with a length p = 62.83 mm and a width w = 11 mm, with a gap width of 13.83 mm (Fig 2A). The resulting length of the metal patch on the inner layer is therefore 49.00 mm. The CRLH metamaterial ring antenna is based on a periodic arrangement of such elongated metal patches on both sides of the Rogers substrate, namely on the inner and intermediate metal layer, where the intermediate periodic structure is shifted by p/2 against the inner layer structure. A 50 mm wide metal ring acts as ground plane on the outer layer and completes the antenna topology. This ground plane ring is actually divided into 8 sections that are equidistantly distributed along the circumference. These sections are capacitively coupled and thus best suited to inhibit eddy currents stemming from field gradient switching.

Metamaterial Ring Antenna System
Periodic material arrangements whose structural properties are designed to modulate electromagnetic (EM) waves in such a way that desired surface-current density distributions can be tailored are known as electromagnetic metamaterials. Their structural features have both inductive and capacitive characteristics. Components made of metamaterials are small, their dimensions typically being shorter than the wavelengths of the RF field of the UHF MRI system. Therefore, a suitable choice of both the metamaterial and of the geometry of the components made of it yields optimal and adaptable surface-current density distributions, which can be tailored to mimic the surface-current distribution of the targeted travelling-wave mode. Hence, any desired excitation of B 1 + mode fields may be generated by a suitable experimental design [27].
The test setup of our MRAS is based on a set of conformal resonant right-/left-handed ring antennas [25]. The MRAS is composed of at least one pair of ring antennas, but more ring antennas can be added easily. The rings may act as transmitters and receivers, but also as reflectors which is a function usually not required in standard body coils at lower fields. When working as reflector, the ring antenna must be short-circuited. The MRAS utilizes the RF shield as a circular waveguide, where waveguide modes are excited at any desired position depending on the number, function, and position of the ring antennas installed in the MRI system bore. Depending on the ring antenna configuration either travelling or standing EM waves can be formed. The optimal surface-current density distribution closely matches the surface-current of the fundamental transverse electric TE 11 mode of a circular waveguide represented by the RF shield at the given diameter ( Fig 2B). Hence, when using quadrature feeding of the RF energy into the antenna, any of the ring antennas contributes to the excitation of a propagating, circularly polarized (CP) B 1 + field.

Experimental setups with Metamaterial Ring Antenna System
In our experiments up to three ring antennas (diameter = 51 cm) were placed into the MRI system bore (inner diameter = 60 cm). The insertion of the ring antennas prevents the operation of the patient table (Fig 2C). As a consequence whole-body imaging of an adult human with all experimental setups was not allowed at this time. To cope with this restriction, we have performed 3 sets of proof of principle experiments, which were designed to study different key features of MRAS. In conjunction, these 3 experimental ring antenna setups demonstrate that MRAS systems are well suitable for whole-body imaging. The first experimental ring antenna setup consists of three ring antennas (Fig 3A). Two ring antennas were placed at the end positions of the MRAS, and one ring was placed in the middle position. The end rings served for both transmit and receive whereas the central ring was used for receive only. A cylindrical silicone oil phantom of 40 cm diameter and 12 cm length was used as load. These experiments were performed to test the SNR characteristics of the MRAS in a large homogeneous VoI (silicone oil has low permittivity of ε r = 2.3 which leads to an almost undistorted B 1 + field distribution and thus images free of interference artifacts). In addition, the MR images of the phantom also allowed assessing the quality of the excited fundamental transverse electric circular waveguide mode (TE 11 ) under low dielectric load. For comparison of the MRAS sensitivity, the SNR was measured in a selected image dataset. The signal values of each two-port ring antenna were sum-of-squares combined into a whole image dataset. The mean value of the signal intensity in a selected ROI signal in the center of the phantom and a second ROI noise in a signal free area outside of the phantom was measured to form the quotient of both. For calculation of the SNR the associated backgroundnoise coil-element correction factors for this mean value approach have to be applied by multiplication with the ROI signal /ROI noise ratio [28,29]. The second experimental ring antenna setup should investigate the capability of MRAS to image larger (in z-direction) VoIs and objects with higher permittivity as typical for in vivo studies. We (1) performed experiments on a cylindrical water phantom that contained water (53%), isopropanol (45%) and sodium chloride (2%) (diameter = 20 cm, length = 40 cm) (Fig 4A), and (2) acquired whole-body images of an adult female crab-eating macaque (Fig 5A) [30]. The higher permittivity (ε r = 58.2) and conductivity (σ = 0.92 S/m) of the water-NaCl-isopropanol solution may lead to stronger distorted B 1 + field distributions, which usually yield to more realistic images with typical interference artifacts. The material parameters represent averaged parameters to mimic the average properties of a human torso based on human tissue material parameters [31]. The choice to select a macaque, which is about 45 cm tall, as an in vivo subject for imaging is motivated by the goal to acquire data from complex biological structures under the above-mentioned spatial restrictions of the MRAS-equipped bore of the MRI system. A second rationale was to provide the neuroscience research group of our site working with macaques monkeys [21,24,32] with a new and potentially better data acquisition concept. For the experiments with the water-NaCl-isopropanol phantom and crab-eating macaque, the MRAS was configured as a two ring antenna systems both for transmit and receive which were placed at the end positions of the VoI (Fig 5A). This configuration was necessary, as the fixation system for the macaque monkey did not allow positioning any additional ring antenna closer to the animal. A third experimental ring antenna setup examined the ability of MRAS to increase the B 1 + transmit efficiency and the SNR for high-resolution imaging. For this purpose, the bore of the MRI system was equipped with one ring antenna for transmit whereas the remaining ring antenna served as an idle-running reflector (Fig 6A). Note that the two antennas were placed at the same side of the subject. Whereas the ring antenna close to the macaque acted as the transmitter, the more distant ring reflected the EM waves back towards the subject. This arrangement of antennas allowed an increase in the B 1 + field directivity toward the object. The smaller receive coil was placed over the head of the subject. For in vivo high-resolution brain imaging, again, adult female crab-eating macaque was used. This setup should maximize the B 1 + transmit efficiency of the MRAS and provided good SNR with the phased array coil. All experimental ring antenna setup graphics were created with the field simulation software CST Microwave Studio 2016 (CST GmbH, Darmstadt, Germany).

In vivo experiments
The in vivo experiments of adult female crab-eating macaques (Macaca fascicularis; age 6 years, weight 7-8 kg) were approved by the Authority for Animal Care and Ethics (Landesverwaltungsamt) of the federal State of Sachsen-Anhalt (No. 28-42502-2-1129 IfN) and conformed to the rules for animal experimentation of the European Communities Council Directive (2010/63/EU). During MRI measurements, the monkey was kept under general anesthesia with a mixture of ketamine (2mg/kg) and xylazine (5 mg/kg). The monkey was held in the sphinx position with a fixation device for the head developed in-house.

MRI systems
The investigated mobile versatile metamaterial-ring antennas were inserted into the bore of a human whole-body 7 T UHF MRI research system (Magnetom 7 T, Siemens Healthcare GmbH, Erlangen, Germany), with inner bore diameter of 60 cm. The 7 T MRI system was equipped with a SC72d gradient system with a maximum amplitude of 70 mT/m and maximum slew rate of 200 mT/m/ms. The gradient system was equipped with a 147 cm long RF  shield with an inner diameter of 64 cm. The system control software version was VB17 (UHF). The linear pulse power amplifier (LPPA 13080W, COMET Stolberg, Stolberg, Germany) providing the RF power was equipped with eight 1 kW RF power amplifiers combined with a power combiner and generates 8 kW (excluding the transmissions losses).
Furthermore, a human whole-body 3 T MRI system (Magnetom Trio 3 T, Siemens Healthcare GmbH, Erlangen, Germany) was used for body coil comparison SNR measurements. The 3 T MRI system was equipped with a gradient system with a maximum amplitude of 40 mT/m and maximum slew rate of 200 mT/m/ms. The linear pulse power amplifier provides an amount of 35 kW (excluding the transmissions losses) RF power.

MRI sequences
To obtain the B 1 + transmit efficiency and the B 1 + field distribution a turbo Fast Low-Angle Shot (FLASH) B 1 + flip angle mapping sequence [33] based on a turbo FLASH sequence preceded by a magnetization preparation (Siemens work-in-progress package, provided by Hans-Peter Fautz) was used. The turbo FLASH B 1 + flip angle mapping sequence was used with the following parameter settings for the silicone oil phantom: time of echo (TE) = 1.83 ms, time of repetition (TR) = 6000 ms, resolution = 3.125×3.125×8 mm 3 , flip angle = 85˚, rectangular presaturation pulse duration = 1.5 ms. For the water-NaCl-isopropanol phantom: TE = 1.83 ms, TR = 5000 ms, resolution = 1.95×1.95×8 mm 3 , flip angle = 85˚, rectangular pre-saturation pulse duration = 1.5 ms (Fig 4C and 4D). For the crab-eating macaques: TE = 1.83 ms, TR = 5000 ms, resolution = 3.35×3.35×8 mm 3 , flip angle = 85˚, rectangular pre-saturation pulse duration = 1.5 ms (Fig 5B). Images of the cylindrical silicone oil phantom were acquired with a 2D gradient-echo (GRE) sequence with the following parameter settings: TE = 4.23 ms, TR = 113 ms, flip  Fig 6A. The high-resolution images show a detailed view of the main parts of the crab-eating macaque brain (gray and white matter, cerebellum, and hippocampus). In the upper left corner a small back-folding artifact is visible which does not affect the imaging of the brain parts. The ticks indicate the distance to the y and z position of the iso-center of the MRI system. https://doi.org/10.1371/journal.pone.0191719.g006 angle = 45˚, resolution = 1.75×1.75×2.5 mm 3 . Images of the cylindrical water-NaCl-isopropanol phantom were acquired with a 2D gradient-echo (GRE) sequence with the following parameter settings: TE = 3.69 ms, TR = 200 ms, flip angle = 60˚, resolution = 1.17×1.17×8 mm 3 . For whole-body MRI of the macaque monkey (Fig 5C) a 2D FLASH sequence was used with the following parameter settings: TE = 4 ms, TR = 435 ms, flip angle = 45˚, resolution = 0.87×0.87×2 mm 3 . In the third experimental setup used for image the monkey brain, images were acquired using a 2D Turbo Spin Echo (TSE) sequence with the following parameter settings: TE = 68 ms, TR = 12220 ms, resolution = 0.31×0.31×1.5 mm 3 , flip angle = 90˚. All DICOM images of this study were managed by the eXtensible Neuroimaging Archive Toolkit (XNAT) [34].

Results
In the first experimental setup we examined the B 1 + transmit efficiency using a large cylindrical silicone oil phantom in a three-ring MRAS configuration. The magnitude of the B 1 + field of the undistorted circularly polarized TE 11 mode in the RF shield is displayed in Fig 3B. The image of the phantom acquired perpendicularly to the cylinder axis is slightly distorted. This is due to small deviations of in the linearity of the UHF gradient system. The B 1 + transmit efficiency along the transversal plane of the cylindrical silicone oil phantom was determined by applying a turbo FLASH B 1 + flip angle mapping sequence [33]. It was found to be B 1 + Tx_eff_7T = 2.2 μTÁ(kW) -0.5 for the MRAS, which is approximately twice that of the B 1 + transmit efficiency of the conventional TW approach [24]. In the central region of interest (ROI) of about 10 cm diameter, the SNR = 101, which is nearly twice as large as that obtained in a standard 3 T Siemens Trio MRI body coil using the same phantom (SNR = 61) but just a quarter in comparison with the B 1 + transmit efficiency (B 1 + Tx_eff_3T = 8.4 μTÁ(kW) -0.5 ). When matter of low permittivity like air (ε r = 1.0) or silicone oil (ε r = 2.3) is present in the bore of the MRI system a undistorted fundamental TE 11 mode is formed and the CP B 1 + field typically possesses a high transversal field homogeneity within the central part of the ROI compromising about 10 cm diameter. Outside of this central part, the field homogeneity decreases due to the inherent transversal profile of the TE 11 mode. During in vivo applications, however, there is a significant dielectric loading (ε r = 50-80) in the MRI system bore that either distorts the TE 11 mode or even leads to the excitation of higher order transverse magnetic (TM) modes [35]. The second experimental ring antenna setup with the water-NaCl-isopropanol phantom as load showed good imaging results (Fig 4B) in z-direction. The distortion is most pronounced at the borders of the water-NaCl-isopropanol phantom, which is caused by deviations from the linearity of the UHF gradient system. The B 1 + transmit efficiency was determined along the coronal and transversal plane (Fig 4C and 4D) to yield B 1 + Cor_Tx_eff = 1.48 μTÁ(kW) -0.5 and B 1 + Tra_Tx_eff = 1.95 μTÁ(kW) -0.5 . In the next step, the second experimental ring antenna setup was tested with in vivo measurements of an adult crab-eating macaque using the maximum VoI provided by the gradient system of the whole-body 7 T MRI system. Fig 5B shows  The 2D TSE sequence applied in the third experimental ring antenna setup required the maximum possible RF power and the highest B 1 + transmit efficiency provided by the system.
In order to maximize the SNR for neuroimaging applications, an optimally adapted 3-element phased array [36] primate head coil was used. The ring configuration in the third experimental setup led to an increase of the B 1 + transmit efficiency to B 1 + Tx_eff = 4 μTÁ(kW) -0.5 inside the monkey's brain tissue. The anatomic high-resolution image in Fig 6B displays both excellent contrast and homogeneity across the entire brain. Gray and white matter can be differentiated with great accuracy and the high-resolution allows a very detailed picture of the brain anatomy, which is most visible in the cerebellum, where even small structures are delineated with high precision. These results are comparable in quality to volume coil excitation approach results [37]. The Q-factor of metamaterial ring antennas for loaded and unloaded conditions in all experimental setups was only minimal varying with Q = [1.2,. . .,1.3].

Discussion
In a theoretical study, the concept of MRAS was proposed [25] for UHF MRI, which combines a volume-selective B 1 + field excitation in z-direction [26]. Simulations of the spatial distribution of the EM waves [38] showed that the MRAS is very versatile, as it provides highly confined field profiles appropriate, for instance, to larynx diagnostics, [26] while reducing the exposure of the irradiated RF energy. Thus, theoretical and numerical studies point to MRAS having a high potential for offering increased B 1 + transmit efficiency at reduced exposure of the whole human body of the irradiated RF energy.
Here, we report on the first experimental realization of such a system, and studied its efficiency for imaging both small as well as whole-body volumes. Furthermore, the B 1 + transmit efficiencies, the resolution, and the SNR of a metamaterial-based ring antenna system were investigated. The MRAS relies on the focused excitation of the fundamental transverse electric circular waveguide mode (TE 11 ). The results show the possibility for selective excitation in z-plane for a research whole-body 7 T MRI system. Measurements of a silicone oil phantom (Fig 3B) indicate the presence of this mode, even if the image of the mode field within the phantom is slightly distorted due to small inhomogeneities in the static magnetic field of the MRI system. The distortions become more pronounced when the permittivity of the medium or subject increases, as seen when comparing the measurements of the silicone oil phantom (ε r = 2.3) with the imaging data of the water-NaCl-isopropanol phantom and a crab-eating macaque. The B 1 + transmit efficiency of our MRAS was determined to B 1 + Tx_eff = 2.2 μTÁ(kW) -0.5 using the silicone oil phantom, a value comparable to that reported for travelling-wave approach [20]. In the cylindrical water-NaCl-isopropanol phantom, the B 1 + transmit efficiency of the second experimental setup with only two ring antennas was minimal lower in transversal plane. This was caused by several effects: first, the inhomogeneous B 1 + field distribution due to the higher permittivity (about 25 times higher than silicone oil) and electrical conductivity of water in a large volume and second by the dual propagation direction of the B 1 + field when using this ring antenna configuration. Even the monkey as load does not lead to an increase of the B 1 + transmit efficiency in the torso or in the head for this ring antenna setup.
The third experimental ring antenna setup (Fig 6A) can increase the B 1 + transmit efficiency to highest value of all measurements with B 1 + Tx_eff = 4 μTÁ(kW) -0.5 inside the monkey brain tissue. The main reason was the more focused B 1 + field resulting from using one ring antenna as reflector. This according transmit efficiency measured in the 7 T MRI system is only around 2 times smaller than those obtained for standard clinical 3 T MRI systems (B 1 + Tx_eff = 8.4 μTÁ (kW) -0.5 measured at a 3 T Siemens Trio MRI body coil. Compared to other recent approaches destined to whole-body imaging, like coaxial waveguide MRI [23] or travelling-wave approach [20], the in vivo B 1 + transmit efficiency of our MRAS is approximately twice as high.
The big advantage of the MRAS for whole-body imaging is its geometric and functional flexibility. On the one hand, MRAS offers the possibility to place the antennas flexibly and with ease at suitable positions, on the other hand, the functional behavior of each antenna can be chosen to act either act as transmitter, receiver, or reflector. Thus, the RF magnetic field distribution in the bore can be tailored using a particular arrangement of ring antennas according to the needs of each experiment. This freedom in the design of the MRAS was demonstrated here with three different experimental ring antenna setups (Figs 3A, 4A and 6A). This flexibility in the design makes the MRAS unique compared to any other experimental UHF MRI whole-body transmit concept [23,[39][40][41][42], and allows a convenient imaging of the subject in the MR bore. For instance, in comparison to the conventional TW approach [20] the MRAS does not require any patch antenna for excitation. Such patch antennas may possibly block one side of the MRI bore thus potentially cause stress to patients due to the restricted space left in the bore.
The imaging studies using crab-eating macaques also demonstrated that high-resolution images could be acquired using an MRAS (Fig 5A) as well as its suitability for imaging of whole monkey (Fig 5C). Keeping in mind that the macaque fixation device occupied the central part of the MR bore ( Fig 5A) and prevented the installation of any additional metamaterial ring antenna in this study. The high SNR enabled a resolution of 0.31×0.31×1.5 mm 3 , providing a very detailed view of the main parts of the crab-eating macaque brain. In fact, the gray and white matters, cerebellum and hippocampus are well delineated with a good contrast ( Fig  6B). Even in low permittivity media like the silicone oil phantom the MRAS exhibited a high SNR, which is nearly twice as large as the SNR of a body coil inside a clinical 3 T MRI system.
The MRAS can also be applied for large volumes up to the limits of gradient system usually with 40 cm in z-direction and 40 cm in x-y direction. This was demonstrated by imaging a large silicone oil phantom and a water-NaCl-isopropanol phantom (with realistic material parameters of permittivity and conductivity) and the whole body of a macaque. In fact, a MRAS-equipped 7 T UHF MRI system opens up the possibility of imaging a maximum FoV. The reconstructed MR image (Figs 4B, 4D and 5C), however, shows non-homogenous illumination along the z-direction, i.e., the contrasts are weaker in the center of the image and brighter at its borders. Two factors contribute to this non-homogeneity, namely the distance of the ring antennas in terms of sensitivity to the phantom and subject, and the distribution of the B 1 + field inside the water phantom and subject. The B 1 + field homogeneity and B 1 + transmit efficiency are given by the resulting mix of transverse electric and transverse magnetic modes, as well as the distribution of permittivities and conductivities. Nevertheless, the inhomogeneous B 1 + field distribution is still an ongoing challenge at UHF MRI. To obtain a more homogeneous B 1 + field distribution across the whole body, RF shimming [43,44] and application of dielectric pads [45] are suggested as promising complements.
In the present study the MRAS ability of acquiring data with high SNR from rather large VoIs was demonstrated using a silicone oil cylinder phantom (diameter = 40 cm) and a crabeating macaque (length about 45 cm) as a load. The ring antennas installed inside the bore of the MRI system blocked any use of the patient table and prevents human subject tests. This shortcoming can be easily overcome by using metamaterial ring antennas whose inner diameters slightly exceed the 60 cm of the MRI bore. They could be integrated directly into the MRI system and placed behind its housing, such that the MRAS system no longer occupies a part of the volume of the bore, nor obstructs the operation of the moving table. Orzada et al. [46] demonstrated the possibility of installation of a 32-element transmit/receive body array for 7 T integrated between the bore liner and the gradient system. This body array uses a moving table and has 4 times higher RF power (32 kW). Thus, the combination of a MRAS and a moving table opens up the possibility of imaging the entire body of a human using a 7 T (or higher) MRI system [47]. The metamaterial ring antennas in all experimental setups didn't show strong load dependency of the Q-factor. This behavior is an advantage for applications as transmit system case but a disadvantage for application as receive system.
In conclusion, we have provided an experimental proof of concept [26,38] that UHF MRI systems (7 T or higher) equipped with a MRAS may be employed to measure VoIs of different sizes, including extended VoIs. The flexibility of the ring configuration and the ring functions opens up the opportunity for a tailor-made design of experimental setups, which then allow for imaging under a wide variety of conditions. Furthermore, the high SNR of MRASequipped MRI systems also suggest that such a system is well suited for high-resolution studies of small VoIs. Last but not least, the MRAS provides new options not only for proton but also for other heteronuclei ( 19 F, 23 Na) [48] due to its inherent, dual-band capability and good SNR for receive.
The outcome of the three experimental ring antenna setups leads us the conclusion that the Metamaterial Ring Antenna System should consists of a minimum number of five ring antennas to provide both high SNR and B 1 + transmit efficiency. This system has to be fully integrated into the MRI system and should use two outer ring antennas as reflector, two ring antennas as transmitter and one central ring antenna for receive only. As soon as 7 T UHF whole-body MRI systems obtain clinical approval, we expect that fullbody MRAS-equipped MRI systems will improve the diagnostic potential of the UHF MRI when monitoring extended inflammation, searching for disseminated tumors and metastases, or analyzing systemic diseases. Ultimately, we believe that MRAS can boost the use of UHF whole-body MRI systems in clinical applications.