An asymmetrical whole‐body birdcage RF coil without RF shield for hyperpolarized 129Xe lung MR imaging at 1.5 T

This study describes the development and testing of an asymmetrical xenon‐129 (129Xe) birdcage radiofrequency (RF) coil for 129Xe lung ventilation imaging at 1.5 Tesla, which allows proton (1H) system body coil transmit–receive functionality.


| INTRODUCTION
MRI of hyperpolarized gases xenon-129 ( 129 Xe) and helium-3 ( 3 He) 1,2 is established as a clinically sensitive functional imaging modality for assessment of the airways and lungs. The safety of the technique 3,4 and its inherent sensitivity to regional lung ventilation and function make it an ideal imaging tool for the assessment of lung diseases. 5,6 The method has been evaluated in clinical studies of different lung diseases and has been shown to have high sensitivity to early-stage lung diseases such as emphysema 7 and early obstruction of the airways in pediatric cystic fibrosis. 8 Due to its moderate cost and natural availability, 129 Xe has recently become the most employed hyperpolarized gas for MR imaging. Studies have shown that 129 Xe provides similar image quality and quantitative information, as can be achieved with 3 He MRI. 9 129 Xe is also soluble in blood and tissue 10 and thus can be used to image lung ventilation-perfusion 11 and evaluate alveolar-capillary gas exchange. 12 In addition, it has been established as a useful tool for the assessment of interstitial lung disease. 13 Recently, the feasibility of 129 Xedissolved phase imaging of other perfused organs such as kidneys [14][15][16] and brain [17][18][19] has also been demonstrated. In all of these hyperpolarized 129 Xe applications, image quality is highly dependent on the delivered flip angle (FA); therefore, high-quality transmit RF coils capable of uniformly exciting the signal from 129 Xe in various organs are desirable.
In this work, we demonstrate the design and implementation of an elliptical asymmetrical thorax birdcage RF coil initially proposed for 3 He, 20 which has since been adopted for various applications such as sodium-23 21,22 and hyperpolarized gases. 23,24 The 129 Xe birdcage here was designed for the following: 1) to operate without an RF shield between the 129 Xe birdcage RF coil and the proton ( 1 H) system body coil to enable both 1 H-129 Xe multinuclear lung MR imaging, and 2) for use in conjunction with a dedicated receive-only RF coil array for 129 Xe imaging for enhanced SNR and accelerated imaging. The electromagnetic interaction of the 129 Xe coil with the 1 H body coil is an important consideration for coil performance and patient safety; thus, specific absorption rate (SAR) was evaluated with electromagnetic simulation software. In addition, transmit efficiency of the 1 H body coil with and without the 129 Xe birdcage is another necessary consideration that was optimized with electromagnetic simulation and then verified experimentally in the MR scanner. We go on to experimentally compare the 129 Xe birdcage RF coil when used as a transceiver and as a transmit-only RF coil in conjunction with a 129 Xe receive-only 8-channel RF coil array for lung ventilation MR imaging. We also demonstrate the ability to acquire coregistered 1 H lung images from the system body coil along with accelerated 129 Xe lung images.

| 129 Xe birdcage RF coil design and simulations
The 129 Xe birdcage RF coil is a 12-legged asymmetrical elliptical structure of band-pass topology, as shown in Figure 1A, similar to the topology used in an earlier 3 He RF coil. 25 The birdcage RF coil has length (head-feet direction) of 47 cm, common elliptical diameter (left-right direction) of 50.8 cm, asymmetrical diameter (anteroposterior direction) of 51.6 cm (top half) and 21.2 cm (bottom half), as shown in Figure 1A. The design was optimized to maximize the use of the available space within the magnet (1.5 T GE HDx scanner, GE Healthcare, Waukesha, Wisconsin), with a 60 cm inner-bore diameter used along with a custom-made patient tabletop insert. Capacitance values were calculated as described in De Zanche et al. 20 Referring to capacitors on the legs and endrings, as shown in Figure 1A,E, the capacitor values are C 1 = C 12 = 450 pF, C 2 = C 11 = 630 pF, C 3 = C 10 = 780 pF, C 4 = C 9 = 680 pF, C5 = C8 = 910 pF, C 6 = C 7 = 580 pF, C 12-1 = C 1-2 = 1700 pF, C 11-12 = C 2-3 = 4400 pF, C 10-11 = C 3-4 = 2220 pF, C 9-10 = C 4-5 = 750 pF, C 8-9 = C 5-6 = 860 pF and C 7-8 = C 6-7 = 640 pF. Figure 1B illustrates the simulation environment showing the RF coils setup in the scanner. The 1 H system body coil has a length of 60 cm and diameter of 60.5 cm, with an outer RF shield at 63.5 cm. The proposed 129 Xe asymmetrical birdcage was nested inside the 1 H body coil. Full-wave EM simulations were performed with Ansys HFSS (Canonsburg, PA) to evaluate the following: 1. B + 1 field homogeneity of the 129 Xe birdcage, for which the 129 Xe birdcage was loaded with human body model (σ = 0.14 S/m and ε r = 81) with air spaces for lungs. 2. The 129 Xe birdcage RF coil was simulated for its 1  3. The isolation s-parameters between the RF coils in all of the 4 configurations were simulated over a frequency span of 10 to 100 MHz. The 1 H body RF coil was tuned and the 129 Xe birdcage RF coil was detuned as per the diode configurations mentioned earlier. In addition, SAR was simulated in MatLab (MathWorks, Natick, MA) as described in Collins et al. 26 using the EM fields exported from Ansys HFSS 18.0 for all the 4 diode configurations mentioned.

| 129 Xe Birdcage coil construction
The mechanical structure of the birdcage is made from machined fiberglass composite. The conductors are solid copper bars with a length of 47 cm, width of 1.5 cm and thickness of 0.3 cm. This mechanical arrangement and its assembled geometry (shown in Figure 1A,D,E) optimally utilizes the available space and accommodates a dedicated 129 Xe (or 1 H) receiver array. The structure of the coil is split in 2 parts to allow subjects to lie down on the posterior part and the anterior part is then fixed on top. Four copper beryllium contact pins make the connection between the 2 halves. Unlike flexible coils, 27 the rigid chassis ensures a more consistent homogeneous magnetic field with loading with a range of subject sizes. Both halves of constructed RF coil are shown in Figure 1E.
The transmit RF line from the system is split into 0° and 90° channels using hybrid coupling. Each output port of the coupler was fed to the 129 Xe birdcage coil at quadrature using a lattice balun that also matches the RF coil to 50 Ω. The RF feeds are on the end-ring across capacitors C 6 and C 9 , as shown in Figure 1E. Two cable traps tuned to 1 H resonance were placed on both transmit and receive cables. Active detuning was achieved by fixing 4 PIN diodes in series with capacitors C 2-3 , C 6-7 , C 7-8 and C [11][12] (Figure 1A,D,E), using RF chokes (10.4 µH chip inductance, parallel resonance notch circuits using 1 µH, as shown in Figure 1D).
Diode detuning was assessed by s-parameter transmission loss using magnetic flux probes with diodes turned on and off. Matching and tuning were evaluated with a human load (28-year-old male, 75 kg, 177 cm) and positioning within the scanner. Q factor was measured with the coil in the loaded and unloaded conditions. The receiver RF coil was a repurposed 1 H 8-channel GE cardiac coil (GE Healthcare) retuned to resonate at the 129 Xe Larmor frequency of 17.66 MHz at 1.5 T ( Figure 1C). The array is divided into 2 separate halves, anterior and posterior, with 4 channels each.

| Hyperpolarization
Experiments were performed on a healthy male volunteer (26 years old, 70 kg) with informed written consent and approval from the UK national research ethics committee. 129 Xe (87% enriched 129 Xe) was polarized to ~ 30% by spin-exchange optical pumping 28 using a homebuilt regulatory-approved polarizer system. 29

| FA mapping
To assess the 129 Xe birdcage coil homogeneity, in vivo B + 1 maps were acquired (inhaled dose of 500 mL 129 Xe gas topped up with 500 mL N 2 ). The sequence was a 2D spoiled gradient-echo with imaging parameters: matrix resolution 32 × 32, slice thickness 200 mm, coronal/axial plane, field of view (FOV) 40 × 40 cm 2 , receiver bandwidth (BW) ±4 kHz, TR = 10.2 ms and TE = 4 ms. Images from the same slice were acquired 10 times repeatedly in a short acquisition time of 4 s, thereby depolarizing the hyperpolarized 129 Xe gas with a known transmit RF power. The pixel-wise rate of depolarization was then used to calculate the B + 1 maps using the relation S n = S 0 sin ( ) cos ( ) (n − 1) , where S is the signal amplitude at nth RF pulse, α is the FA and S 0 is initial amplitude. To assess the RF transparency of 129 Xe birdcage RF coil when positioned within the 1 H body coil, a FA map of the 1 H body coil was acquired with and without the detuned 129 Xe birdcage RF coil in situ using a large cylindrical phantom filled with 3.6 g/L NaCl and 1.96 g/L CuSO 4 ⋅5H 2 O salt solution (radius = 15.5 cm, height = 42 cm). The imaging sequence was a 2D spoiled gradient-echo, axial and coronal plane, matrix 128 × 128, TR = 300 ms, TE = 30 ms, FOV = 40 × 40 cm 2 , 8 mm slice thickness and BW = 15.63 kHz. To obtain the FA map, acquisition was repeated with incremental transmit RF power and the acquired image was fitted pixelwise as described elsewhere. 30

| 129 Xe ventilation imaging
The sequence used for xenon ventilation imaging was a 3D balanced steady-state free precession sequence with parameters: 100 × 82 × 24 matrix resolution, 10 mm slice thickness, BW of ±8 kHz, FOV of 40 × 40 cm 2 , TR = 6.4 ms, TE = 3.1 ms, FA = 10°, and with 750 mL 129 Xe dose. FA was calculated with a series of pulse-acquires at a particular transmit RF power, as described earlier 28 and the average FA from all the channels were used. SNR was calculated as a ratio of average signal from the region of interest of the lungs to the square root times the SD of the background noise. 31 For the receiver array, all channels were combined with root sum of square reconstruction 32 before calculating SNR. Performance of the 2 RF coil combinations were compared using the mean SNR ( Figure 4C) in a slice.

| Accelerated imaging
Array sensitivity was estimated through autocalibration by fully sampling the 20 lines at the center of k-space, 33 then under sampling the rest of the k-space, 34 zero-padding the unacquired k-space and reconstructing after filtering (3D lowpass Kaiser window (β = 3) 23 to prevent truncation artefacts). The optimal flip angle of the steady-state free precession varied with the number of RF encoding steps depending on the acceleration factors 35 , calculated as described elsewhere, 36,37 assuming a T 1 of 20 s 38 and no off-resonance effects. The sequence used for accelerated imaging was a 3D steady-state free precession with 80 × 80 × 20 matrix, slice thickness 10 mm, BW = ±8.06 kHz, FOV = 40 × 40 cm 2 , TR = 6.4 ms, TE = 3.1 ms and acceleration factor of R = 2 in the phase encode direction.
Coregistered anatomical proton MR images of the lungs were also acquired with the 129 Xe coils in situ; 1 H imaging parameters were 3D coronal spoiled gradient echo pulse sequence with matrix 80 × 80 × 20, BW = ±83 kHz; FOV = 40 × 40 cm 2 , TR = 1.5 ms, TE = 0.6 ms, 5 Averages, and total scan times 12 s. Using the same parameters, a separate noise scan was performed without the inhaled gas to compute SNR. The transmit RF power for 1 H was limited for patient safety in accordance with the estimated SAR and RF coupling with 129 Xe RF coil.

| Simulations
At 17.66 MHz B + 1 field simulation showed a homogeneous field within the lung region with a SD of 4%. For RF F I G U R E 2 Simulated B + 1 field on the axial plane of the 1 H body coil at 63.8 MHz without and with 129 Xe birdcage RF coil nested inside for various detuning circuit configurations. The mean and SD of B + 1 uniformity of the 1 H body coil within the 129 Xe RF coil as outlined by the red rectangle for all the configurations are 0.85% and 1.9% (without 129 Xe RF coil), 0.38% and 3.8% (no diodes), 0.55% and 14.3% (4 diodes), 0.85% and 20.1% (8 diodes) and 0.11% and 9.5% (12 diodes), where mean values being normalized to an obtained maximum regional value. Location of the detuning circuits on the legs in series with capacitors are marked as boxed numbers and its equivalent circuit is also shown. The RF isolation between 129 Xe and 1 H RF coils measured as transfer s-parameters (S 21 ) for all the configurations are indicated. SAR simulation for 1 H body coil with detuned 129 Xe birdcage RF coil in situ is also shown transparency of the 129 Xe birdcage RF coil for 1 H imaging, the detuning configuration with 4 diodes was optimal when compared to the other configurations despite some residual coupling that impaired the 1 H B + 1 magnitude (reduction of 36%) and homogeneity (SD 14.36%), as shown in Figure 2. Simulated isolation between the 129 Xe birdcage coil and the 1 H body coil was −15 dB. Average SAR was 1.23 W/Kg and local SAR (near the shoulders) was 9.39 W/Kg simulated for 1 H body coil with the detuned 129 Xe birdcage coil in situ, as shown in Figure 2.

| 129 Xe Birdcage coil evaluation
Measured active detuning effectiveness (measured as difference) with diodes on and off was −23 dB at 17.66 MHz; the return loss of the 129 Xe transmitted birdcage was below −22 dB for both ports. The isolation between the 2 quadrature ports of the 129 Xe birdcage RF coil was −30.8 dB. Q factor was 197 and 100 in the unloaded and loaded conditions, respectively.
The in vivo FA delivered across the lungs with the 129 Xe birdcage coil had a SD of 10% and 9% for coronal and axial plane, respectively, as seen in Figure 3A,B. The FA for the 1 H body coil for a particular transmitted RF power with the 129 Xe birdcage coil in situ was 4 lower when compared without it present, as seen in Figures 3C-F. The SNR of the birdcage when used as a transceiver versus as a transmit-only coil with 8-channel receiver RF coil array was 31, 26,18 and 46, 25, 32 in posterior, central and anterior slices, respectively ( Figure 4A,B). An approximate twofold net increase of SNR was observed using the birdcage as a transmitter with receive-only RF coil array ( Figure 4C). This is due to better sensitivity of the array in the anterior and posterior regions of the lungs, as seen in the SNR profile in Figure 4C. 1 H images acquired with 1 H system body coil with 129 Xe birdcage RF coil in situ enabled spatially coregistered multinuclear MR lung images ( Figure 5A). HP 129 Xe steady-state free precession sequence signal simulations found an optimal FA of 10.3° and 14.1° for fully sampled encoded and accelerated (R = 2), respectively, as demonstrated by coronal and axial lung images reconstruction shown in Figure 5. Total scan times were 11 s and 8 s and for full sampled and R = 2 sampled, respectively. Image reconstruction for R = 2 ( Figure 5B) shows no visible reconstruction artefacts with a g-factor of 1.4 ( Figure 5C).

| DISCUSSION
In this note, we have demonstrated an asymmetrical band pass birdcage for hyperpolarized 129 Xe MRI that works without an RF shield enabling in situ 1 H imaging. In vivo FA mapping demonstrated B + 1 similar to values reported in the literature. 24,39 Of the diode configurations evaluated for detuning, the configuration with 4 diodes showed least distortion of the 1 H B + 1 field. Although the 129 Xe transmit birdcage coil enables proton imaging, the obtained FA with the 129 Xe coil in situ was lower by a factor of 4 for the same transmit power, which is likely due to residual undesired coupling between the RF coils. Increasing the transmit power of the 1 H RF amplifier is not an optimal solution to mitigate this because the local SAR values (9.23 W/kg) were close to the International Electrotechnical Commission standard for normal level controlled scans. 40 Therefore, future work should focus on improving the isolation between the 2 RF coils in the magnet, for example, using passive trap circuits tuned to the 1 H resonance frequency on the 129 Xe RF coil. 41 The obtained results show some improvement when compared with F I G U R E 4 Ventilation images obtained with (A) the asymmetrical birdcage in transceiver configuration and (B) 8-channel receiver RF coils array. (C) Comparison of averaged SNR of each coronal slice for the 2 RF configurations the B 1 field homogeneity obtained with flexible vest coil designs similar to results previously reported in literature. 24,25 When combined with a separate 8-channel receive array, the increased SNR was evident despite some dropoff in the sensitivity toward the central slices. The receiver RF coil array permitted acceleration of 2 with low g-factor values (maximum values ~1.4).

| CONCLUSION
We have demonstrated an RF coil setup for 129 Xe human lung imaging at 1.5 T using an asymmetrical elliptical 129 Xe birdcage RF coil both as a transceiver and with a receive-only RF coil array. This 129 Xe birdcage RF coil, designed without an RF shield and with receive-only RF coil array compatibility, has enabled both 1 H MR imaging and accelerated imaging, functionality which has not been previously demonstrated together.