Enabling Radiofrequency Antenna Technology

A plethora of reports eloquently refer to the development of enabling RF technology tailored for CMR at 7 T. Research directions include local transceiver (TX/RX) arrays and multichannel transmission arrays in conjunction with multichannel local receiver arrays.

Multichannel RF coil designs tailored for UHF-CMR involve rigid, flexible, and modular configurations. The gestation process revealed a trend toward a larger number of transmit and receive elements to improve anatomic coverage and to advance the capabilities for transmission field shaping. Pioneering developments made good use of building blocks that include stripline elements, electrical dipoles, dielectric resonant antennas, and loop elements, with up to 32 independent elements.

Loop element-based CMR optimized 7 T transceiver configurations were reported for a 4-channel TX/RX ( Fig. 13.1A ), an 8-channel TX/RX ( Fig. 13.1B ), and a two-dimensional (2D) 16-channel TX/RX design ( Fig. 13.1C ). The 2D approach was extended to a more sophisticated modular 32-channel TX/RX array shown in Fig. 13.1D .

Examples of multichannel transceiver arrays tailored for cardiovascular magnetic resonance at 7 T. Top row, Photographs of cardiac optimized 7 T transceiver coil arrays including (left to right) a 4-channel, an 8-channel, a 16-channel, and a 32-channel loop array configuration together with an 8-channel and 16-channel bow-tie antenna array. For all configurations, the coil elements are used for transmission and reception. Four-chamber (middle row) and short-axis (bottom row) views of the heart derived from two-dimensional CINE fast low angle shot acquisitions using the radiofrequency coil arrays shown on top and a spatial resolution of (1.4 × 1.4. × 4) mm ³ .

(From Niendorf T, Paul K, Ozerdem C, et al. W(h)ither human cardiac and body magnetic resonance at ultrahigh fields? Technical advances, practical considerations, applications, and clinical opportunities. NMR Biomed. 2016;29(9):1173–1197.)

Various stripline element-based 7 T transceiver configurations were proposed for cardiac UHF-MR. In a pioneering 8-element transverse electromagnetic field (TEM) transceiver array design, each element was independently connected to a dedicated RF power amplifier. Other stripline array configurations run flexible designs consisting of a pair of 4 or 8 stripline elements. Another practical solution comprising an 8-channel stripline transceive array that uses sophisticated automated tuning with piezoelectric actuators was accomplished.

Electric dipole configurations have been shown to benefit MR of the upper torso and the heart at 7 T. Electric dipoles run the trait of a linearly polarized current pattern, where RF energy is directed perpendicular to the dipole along the Poynting vector to the subject, resulting in a symmetrical, rather uniform excitation field with increased depth penetration. Early implementations include straight dipole elements. The need for densely packed multichannel transceiver coil arrays tailored for UHF-CMR has inspired explorations into electric dipole configurations that comprise 8 or 16 bow-tie antenna building blocks, as shown in Fig. 13.1E and F . Each of the building blocks contains a bow-tie–shaped λ/2-dipole antenna immersed in D ₂ O. This approach helps to shorten the effective antenna length.

Improvements in the compactness of dipole antenna arrays were also accomplished by incorporating lumped element inductors, by using fractionated dipoles, or by employing meander structures. The requirements of UHF-CMR along with ultimate intrinsic SNR and current pattern considerations has inspired alternative dipole antenna designs. One important development is bent electric dipoles. Other configurations include circular dipoles, which consist of a circular conductor with a feed point on one side and a gap on the other. Recently, proposed dipole antenna variants include folded dipoles or combinations of loop elements and dipole configuration. The snake antenna building block presents an alternative for the design of transmit arrays tailored for UHF-CMR.

Ultimate intrinsic signal-to-noise ratio (UISNR) considerations outlined that loop and dipole current patterns contribute equally to UISNR at 7 T. For field strengths of B ₀ ≥ 9.4 T, current patterns are dominated by linear (dipole-type) current patterns, which provide motivation for shifting the weight to dipoles versus loop and stripline elements at 7 T and beyond.

Multichannel Radiofrequency Transmission

To translate the capabilities of multichannel RF coil configurations into practical value, advances in MR systems hardware abounded that help to mitigate signal variations across the heart and to address blood/myocardium contrast heterogeneity. State-of-the-art UHF-MR systems architecture supports multichannel transmit arrays to be driven in two modes, using (1) an array of independent-feeding RF power amplifiers (RFPA) with exquisite control of phase, amplitude, or even complete RF waveforms for all individual channels or (2) a single-feeding RFPA. For the single-feeding RFPA regime, the amplifier output is commonly split into fixed-intensity signals using RF power splitters. Phase adjustments for individual coil elements or groups of coil elements are accomplished with phase-shifting networks or coaxial cables. This setup provides a practical solution that can be made readily available for a large number of MR sites. In today’s research implementations, RFPAs with up to 16 kW peak power are used for the single-feeding RFPA mode. State-of-the-art multifeeding RFPA implementations support 8 to 16 RF amplifiers, each providing 1 to 2 kW adjustable RF output power. A multifeeding RFPA setup provides software-driven control over phase and amplitude by modulating the applied input signal. Moreover, it enables sophisticated parallel transmission of independent RF waveforms to every channel. On the journey to a broader range of UHF-CMR applications, it is beneficial if RF coil configurations provide flexibility which supports single-feeding as well as multiple-feeding RFPA regimes without major changes in cabling and other components. To this end, Fig. 13.2 shows an example of a 16-channel loop array tailored for UHF-CMR which is connected to a universal interface that supports the single-feeding and multifeeding RFPA modes. Transition between both modes can be accomplished in 5 seconds without the need for moving the patient or the patient table.

Example of a 16-channel radiofrequency (RF) coil array tailored for cardiovascular magnetic resonance at 7 T which is connected to a universal and magnetic resonance vendor-independent interface that supports two RF transmission modes using (left) an array of up to 16 independent-feeding RF power amplifiers (RFPAs) with exquisite control of phase, amplitude, or even complete RF waveforms for all individual channels or (right) a single-feeding RFPA connected to a network of RF splitters and phase shifters used for transmission field shaping. For the single-feeding RFPA mode, the amplifier output is split into 16 fixed-intensity signals using RF power splitters. Phase adjustments for individual coil elements (or groups of coil elements) are accomplished with phase-shifting networks incorporated in the coil interface. Transition between both transmission modes can be accomplished in 5 seconds without the need for moving the patient or the patient table.

(Courtesy Dr. Helmar Waiczies, MRI.TOOLS GmbH, Berlin, Germany.)

Progress in Pulse Sequence Development

Progress in UHF-CMR pulse sequence methodology includes two main categories: (1) novel approaches for transmission field mapping and (2) new technologies for transmission field shaping.

The knowledge of the spatial B ₁ ⁺ distribution generated by RF coil arrays is pivotal for UHF-CMR to manage transmission fields across the heart, including magnetization preparation, signal excitation, and signal refocusing. B ₁ ⁺ mapping approaches commonly used are mainly magnitude-based and are generally confined to the ratios or the fit of signal intensity images. For this purpose, sets of images are acquired using either two flip angles, identical flip angles but different repetition times (TR), variable flip angles, or signals from spin echoes and stimulated echoes, as well as signals from gradient echoes and stimulated echoes. It should be noted that B ₁ ⁺ mapping is not a need limited to transceiver arrays tailored for ¹ H UHF-MR but includes heteronuclear MR applications. For instance, B ₁ ⁺ mapping has been applied to ²³ Na MRI to support quantification of sodium tissue content.

Phase-based techniques present an alternative for B ₁ ⁺ mapping, and were reported to be more accurate versus magnitude-based methods exploiting the low flip angle regime. Phase-based techniques commonly take advantage of a composite or off-resonance RF pulse to encode the spatial B ₁ ⁺ information into the phase of the magnetization vector. Most phase-based methods, however, make use of nonselective pulses for volume excitation in conjunction with three-dimensional (3D) imaging schemes. This approach results in relatively long acquisition times that often exceed clinically acceptable breath-hold periods, along with a pronounced susceptibility for respiratory motion artifacts. Realizing these limitations, the feasibility of cardiac-gated, single-breath-hold B ₁ ⁺ mapping has been demonstrated using a 2D phase-based Bloch-Siegert approach.

Notwithstanding the advances in B ₁ ⁺ -mapping methodology, transmission field mapping of individual coil elements at 7 T has been primarily established for head imaging but remains a challenge for UHF-CMR because of depth penetration constraints and susceptibility artifacts. To cope with this challenge, B ₁ ⁺ profiles are often derived from numerical electromagnetic field (EMF) simulations. For this purpose, the transmit phases for each coil element are adjusted using iterative algorithms to improve transmission field homogeneity in a region of interest encompassing the heart using a human voxel model. To ensure that the relative B ₁ ⁺ distributions of the individual coil elements derived from the EMF simulations accord with reality, it is essential that EMF simulations are validated against MR measurements using phantoms filled with a dielectric liquid that resembles the dielectric properties of the target organ/tissue, as outlined in Fig. 13.3 .

Example for transmission field assessment with electromagnetic field simulations and experimental B ₁ ⁺ mapping using a building block equipped with four loop radiofrequency (RF) elements. For this purpose a cylindrical phantom setup was used in the simulations and in the experiments. (A) Virtual model of the setup used for validation (RF shield and casing are not shown but included in the simulations). The experimental setup was arranged analogously to the virtual setup. (B) Simulated B ₁ ⁺ distribution (absolute values) of four channels which form one curved building block. For B ₁ ⁺ evaluation, transversal slices through the cylindrical phantom were positioned in alignment with the center of the loop elements. (C) Absolute transmission fields derived from B ₁ ⁺ mapping of the experimental setup. (D) Difference maps in percentage of the simulated results. The results demonstrate the qualitative and quantitative agreement between the numerical simulations and the measurements.

(From Niendorf T, Paul K, Oezerdem C, et al. W(h)ither human cardiac and body magnetic resonance at ultrahigh fields? Technical advances, practical considerations, applications, and clinical opportunities. NMR Biomed . 2016;29(9):1173–1197.)

Based upon a priori acquired B ₁ ⁺ maps derived for all individual transmit channels of a coil array, B ₁ ⁺ shimming can be used to reduce RF nonuniformities across the heart using the degrees of freedom of multichannel transmission to shape the transmission field for UHF-CMR.

Dynamic shimming is the forte of the multifeeding RFPA’s mode in conjunction with sophisticated pulse sequence developments, which support cascades of dynamic B ₁ ⁺ shimming stages that balance the needs of preparation modules common in cardiac imaging such as inversion, fat suppression or blood suppression, pencil beam excitation across the diaphragm for navigator-gated respiratory motion compensation, and rapid imaging modules. Frontier CMR implementations of dynamic B ₁ ⁺ shimming at 7 T include 2D CINE imaging, coronary MR angiography, and four-dimensional (4D) flow MR angiography of the entire aorta, which demonstrated improved excitation efficiency, increased SNR of myocardium and blood, and enhanced blood/myocardium contrast or helped to balance the B ₁ ⁺ uniformity with B ₁ ⁺ efficiency.

Multichannel transmission using spoke RF pulses presents a major breakthrough for pulse sequence developments tailored for UHF-CMR. A juxtaposition of the two-spoke approach with B ₁ ⁺ shimming revealed an enhanced excitation fidelity or a reduced RF pulse energy for the two-spoke technique. This benefit holds the promise to be instrumental for a broader range of CMR applications, including uniform excitation for parametric mapping used for tissue characterization. The pulse designs used for two-spoke parallel transmission RF pulses make use of a single set of B ₁ ⁺ / B ₀ maps. This approach may not be suitable for subsequent scans acquired at another respiratory phase because of organ displacement, which might induce severe excitation profile and B ₀ degradation. To address this issue, a tailored parallel transmission RF pulse design which is immune to respiration-induced B ₁ ⁺ / B ₀ variations has been recently proposed. Multiband spoke pulses afford simultaneous multislice acquisitions with enhanced flip angle homogeneity but minimal increase in integrated and peak RF power.

Ancillary Hardware Tailored for Ultrahigh Frenquency-Magnetic Resonance

In current clinical MR practice, cardiac motion is commonly dealt with using electrocardiographic gating/triggering techniques to synchronize data acquisition with the cardiac cycle. If brought into a magnetic field, ECGs, being an electrical measurement, are corrupted by magnetohydrodynamic (MHD) effects. The MHD effect is pronounced during cardiac phases of systolic aortic flow, which results in a distortion of the S-T segment of the ECG. The susceptibility to EMF interference manifests itself in ECG waveform distortions already apparent in ECG traces acquired in clinical 1.5 T MR scanners. At ultrahigh fields the propensity of ECG recordings to MHD effects is pronounced. Artifacts in the ECG trace and severe T-wave elevation might be misinterpreted as R-waves, resulting in misdetection of cardiac activity or erroneous cardiac gating together with motion-corrupted image quality. Realizing the constraints of conventional ECG, an MR stethoscope has been proposed ( Fig. 13.4 ) for the pursuit of robust and safe clinical gated/triggered CMR. In contrast to ECG-triggering, the MR stethoscope employs acoustic instead of electrical signals. For cardiac gating/triggering, the first heart tone of the phonocardiogram, which marks the onset of the acoustic cardiac cycle, is selected. The phonocardiogram was reported to be immune to interferences with electromagnetic fields and to provide reliable and stable trigger information in UHF-CMR, as illustrated in Fig. 13.4 .

Block diagram (left), clinical setup (center), and four-chamber views of the heart obtained for (top) vector electrocardiogram (ECG) gating and (bottom) acoustic cardiac triggering (ACT) used for synchronization of retrospectively gated two-dimensional gradient echo imaging with the cardiac cycle at 7 T. Interference with magnetohydrodynamic effects cause severe distortion in the vector ECG waveform, resulting in erroneous trigger recognition and cardiac motion-induced artifacts (top right). For comparison, ACT is free of interferences from magnetohydrodynamic effects, provides a reliable trigger signal for ultrahigh frequency cardiovascular magnetic resonance, and images free of cardiac activity-induced motion artifacts (bottom right). MR, Magnetic resonance; MRI, magnetic resonance imaging; MRT, magnetic resonance tomograph.

Radiofrequency Power Deposition

Radiofrequency pulses used in MR deposit RF power in tissue, which can be described by the specific absorption rate (SAR):

At UHF-MR, the wavelength becomes sufficiently short versus the upper torso, which affects the field distribution inside the body and offsets some of the SAR predictions governed by :

RF power deposition is substantially affected by the geometry of the upper torso, by positioning of the RF coil with respect to the torso and the heart, by the RF coil design, and by the number of RF transmission channels, as well as by the RF driving conditions. Here, EMF simulations are essential to provide an insight into the local SAR distribution. Recent simulation studies showed that using the merits of high RF frequencies for multichannel RF antenna array design, the relative SAR increase is even well below a SAR ~ B ₀ dependency up to UHF-MR frequencies as high as 1 GHz (23.5 T) ( Fig. 13.5 ).

Frequency-dependent relative specific absorption rate (SAR) increase covering frequencies ranging from 300 MHz (7 T) to 1 GHz (23.5 T). For the electromagnetic field simulations, multichannel dipole antenna arrays were positioned around a cylindrical phantom (diameter = 180 mm). Discrete radiofrequencies (RF) ( f = 300 MHz to 1 GHz) have been evaluated for two transmit settings: 00 (0 degree phase shift between transmit channels) and CP (phase shift between transmit channels: 360 degrees/number of elements). At higher RF the number of transmit channels can be increased from an 8-channel (8ch) to a 19-channel configuration (maxCh). This improvement in element density increases transmit efficiency, lowers SAR, and reduces the relative SAR increase vs. 300 MHz. Taking advantage of high-density multichannel RF antenna arrays, the relative SAR increase (1) is well below the SAR ~ B ₀ relationship described by the literature for low magnetic field strengths and (2) falls even below a SAR ~ B ₀ dependency up to ultrahigh frequency magnetic resonance frequencies as high as 1 GHz (23.5 T).

(From Winter L, Niendorf T. Electrodynamics and radiofrequency antenna concepts for human magnetic resonance at 23.5 T (1 GHz) and beyond. MAGMA. 2016;29(3):641–656.)

Although SAR is key for RF safety evaluations, temperature distribution—which is the cause of tissue damage based on RF heating—does not follow SAR in a straightforward manner. Thermoregulation and heat transfer inside the body (i.e., blood vessels) need to be considered while moving toward a more realistic scenario. Ongoing research focuses already on temperature modeling instead of SAR models, suggesting a thermal tissue damage threshold CEM43 known from thermal therapies. The thermal dose concept will be included in edition 4 of the IEC 60601-2-33 safety guidelines, which is expected to be finalized by 2018. Here, realistic thermal modeling based on electromagnetic field simulations and MR thermometry is essential to assess thermal distributions in vivo. These explorations will help to advance toward enhanced MR safety assessment for UHF-CMR. This development is expected to include a drive toward controlled E-field steering and targeted SAR reduction technologies in electrically conductive implants, catheters, and guide wires using parallel transmission to create excitations that induce minimal RF current in elongated conductors. An open-minded look reveals that UHF-CMR parallel transmission technology perhaps forms an enabling platform to advance interventional CMR applications at 7 T by using an MR antenna array to deliberately focus RF power and increase the temperature locally in a controlled way, denoted as thermal magnetic resonance. In this context, potential applications could include (1) targeted RF-guided drug release and image-guided monitoring, (2) localized contrast agent release and tracking, or (3) stem cell delivery to the myocardium afforded by local RF heating. In this context new developments of MR thermometry techniques and their validation will be crucial to MR safety and to advance the field of thermal therapies and thermal MR.

Radiofrequency-Induced Heating of Passive Conductive Devices

En route to broader clinical UHF-CMR studies, it is essential to gain a better insight into the interaction of passive conducting implants with RF fields. Considering an ever-increasing population of patients with a history of stent implantation, detailing the interaction of stents with RF fields is of profound relevance for the advancement of UHF-CMR. This is not an easy task because of the many RF coil and stent configurations available, yet becomes even more relevant when moving to many element transceiver arrays or even to large-volume coils covering major sections of the body.

At UHF-MR the RF wavelength ( λ ) in myocardial tissue and blood is sufficiently short ( λ approximately 12 cm) to induce resonance effects at λ /4 to λ /2 in coronary stents with a length of up to approximately 4 cm, a dimension common in clinical practice. EMF simulations and RF heating experiments showed that temperature increase due to coronary stents does not exceed baseline RF heating if IEC standards for local and global SAR limits are strictly obeyed. Although being very important, these early studies did not employ a setup that resembles a clinical scenario. The need for RF heating assessment at B ₀ ≥ 7 T also prompted valuable research into RF-induced heating of metallic arterial stents. The conclusions drawn are valuable but constrained to the very specific experimental setup used.

Recognizing the opportunities for broader clinical UHF-CMR studies, RF-induced heating of coronary stents was examined at 7 T. For this purpose, EMF simulations were performed for a broad range of coronary stent configurations to detail electric fields and local SAR as a function of stent diameter, stent length, stent position with respect to the RF transmission source, and stent orientation versus the E-field. To translate these results to arbitrary coronary stent geometries, arbitrary stent positions, and arbitrary RF coil configurations, a transfer function for the assessment of EM fields induced in coronary stents was proposed. SAR _1g induced by a stent (SAR _{1g stent} ) can be described by:

Electromagnetic field (EMF) simulations using the human voxel model “Duke” from the virtual family and an 8-channel transmit/receive bow-tie electric dipole array. (A) Schematic views of the positioning of the anterior bow-tie radiofrequency antennas on the anterior chest of the human voxel model and coronal view of specific absorption rate (SAR) _{1g baseline} distribution for a plane through the target position without the stent equivalent being present. (B) Surface plot of SAR _1g stent derived from the transfer function for baseline SAR _1g for an input power of 1 W/kg. A stent length ranging from 10 to 40 mm and a stent rotation versus the main E-field vector ranging from 0 to 90 degrees was applied. (C) Coronal view of SAR _{1g stent} distribution for a plane through the target position with the stent equivalent being present. (D) Simulated maximum baseline SAR _{1g baseline} and SAR _{1g stent} for the stent equivalent using eight randomly generated phase settings compared with the SAR estimation deduced from the analytical approach (see Eq. 13.3 ). Although conservatively overestimating SAR, SAR estimation using Eq. 13.3 was able to predict the induced SAR _1g levels without the need to perform extra time-consuming EMF simulations with a stent being present in the simulation model.

(From Niendorf T, Paul K, Oezerdem C, et al. W(h)ither human cardiac and body magnetic resonance at ultrahigh fields? Technical advances, practical considerations, applications, and clinical opportunities. NMR Biomed. 2016;29(9):1173–1197.)

The transfer function provides a fast way to estimate induced stent SAR levels. EMF simulations performed with a stent integrated in a human voxel model prove to be very time consuming. This is a particular practical obstacle if many-element RF TX/RX arrays and a set of phase settings mimicking B ₁ ⁺ shimming or other parallel transmit applications are included. Benchmarking the analytical approach versus EMF simulations showed a good conservative estimation of induced SAR _1g peak levels. A practical example is shown in Fig. 13.6 using an 8-channel bow-tie antenna RF coil array optimized for CMR at 7 T (see Fig. 13.1E ). EMF simulations including the stent consumed more than 2 hours central processing unit (CPU) and graphics processing unit (GPU) time per phase setting (not including RF shim field combining and SAR calculations) while the analytical approach can be calculated in real time. This example underscores the value of the proposed analytical approach based upon a transfer function for obeying regulatory RF power limits. The approach of using a transfer function is conceptually translatable to other implant configurations, which is in particular suitable for short implants (≤λ _tissue /3) where the influence of the phase distribution of the RF transmit field is negligible. Furthermore, it can be conveniently incorporated into state-of-the art SAR prediction concepts to provide SAR estimations induced by coronary stents and other conductive implants for arbitrary RF pulses used for transmission field shimming or parallel transmission. To generalize, this approach provides a novel metric or design criteria for RF coils: for example, to design RF coils (1) with low SAR levels in the vicinity of the stent or (2) with SAR levels during the presence of the stent which do not exceed SAR levels without the presence of the stent. The transfer function approach should not end at SAR estimation, but should move toward temperature or CEM43 estimations. In particular, for short implants like coronary stents, maximum induced SAR levels in the implant are much more restrictive to the allowed input power than maximum temperature or CEM43 values would be. It was shown that even though maximum induced local SAR at the stent tip was by a factor >2 higher than in surface regions, the eventual temperature after 6 minutes of RF heating was by >6°C lower at the stent tip region. This temperature saturation effect at the stent tip originates from a strong temperature gradient toward colder tissue regions at the stent center and is even more pronounced by physiologic parameters such as blood flow in the stented vessel. These findings are heartening; they will accelerate further research on safety assessment of coronary stents and other implants common in clinical practice and will help to relax current UHF-CMR limitations.