Design and Implementation of Split-Leg Type Elliptical Whole-Body Birdcage RF Coil at 1.5 T MRI

Featured Application

An elliptical RF coil configuration can be implemented for whole-body imaging in a magnetic resonance imaging (MRI) system. Additionally, this elliptical configuration can be modified, and the space between the RF shield and the RF coil can be utilized for interventional radiological equipment systems such as PET detectors and HIFU transducers. The PET detector can be placed between the horizontal space between neighboring legs and the vertical space between the RF coil and the RF shield. Additionally, the HIFU transducer can be implemented for breast as well as prostate cancers, where the transducer is placed below the patient.

Abstract

The feasibility and the development of a four-port elliptical birdcage radio frequency (RF) coil for generating a homogenous RF magnetic (B₁) field is presented for a space-constrained narrow-bore magnetic resonance imaging (MRI) system. Optimization was performed for the elliptical birdcage RF coil by adjusting the position and the structure of the legs to maximize the B₁⁺-field uniformity. Electromagnetic (EM) simulations based on RF coil circuit co-simulations were performed on a cylindrical uniform phantom and a three-dimensional human model to evaluate the B₁⁺-field uniformity, the transmission efficiency, and the specific absorption rate (SAR) deposition. An elliptical birdcage RF coil was constructed, and its performance was evaluated through network analysis measurements such as S-parameters and Q-factor. Quadrature transmit and receive MRI experiments were conducted using both phantom and in vivo human for validation. The EM simulation results indicate reasonable B₁⁺-field uniformity and transmission efficiency for the proposed elliptical birdcage RF coil. The signal-to-noise ratio and the flip angle maps of the uniform phantom and the in vivo human MR images acquired using an elliptical birdcage (62 cm × 58 cm) were similar to those of a commercial circular birdcage (diameter, 58 cm), thereby indicating acceptable performance. In conclusion, the proposed split-type asymmetric elliptical birdcage RF coil is useful for whole-body MRI applications and can be used for imaging larger human subjects comfortably in a spacious imaging space.

1. Introduction

Over the past few decades, magnetic resonance imaging (MRI) has become an essential imaging modality in modern bioimaging research and clinical applications. In MRI, radio frequency (RF) coils are crucial for transmitting and receiving RF signals in the target object, where a highly homogenous RF magnetic (B₁) field, a large field-of-view (FOV), and a high signal-to-noise ratio (SNR) are required. The birdcage RF coil introduced over four decades ago [1] remains as the preferred option for transmitting RF coils in clinical MRI systems (1.5 T, 3.0 T, and 7.0 T) owing to its capability in generating a uniform B₁-field over a large portion of the imaging volume. The birdcage RF coil driven in quadrature mode generates a circularly polarized B₁-field, which reduces the RF power requirement by a factor of 0.5 to 0.7 depending on the shape and the size of the subject; additionally, it increases the receiver sensitivity by √2 over the linear mode [2,3]. Birdcage RF coils can be categorized into low-pass (capacitors placed on legs), high-pass (on end-rings), and hybrid (both on legs and end-rings) configurations. Many different variants of birdcage RF coils were designed, such as circular, elliptical [4], and asymmetric shapes [5,6], to closely match the RF coil outline to the subject outline as well as the space constraint for improved B₁-field homogeneity and overall performance.

Numerous analytical and theoretical solutions were reported previously for both circular [7,8,9,10,11,12] and noncircular [13,14,15,16,17] birdcage RF coils to optimize the B₁-field and the capacitor values for the desired resonance at a specified Larmor frequency. These numerical solutions were based on equivalent circuit analysis by which the optimum capacitor values of the birdcage RF coil and their frequency response spectrum were calculated. However, these calculations yielded substantial errors in the optimum capacitor values for large-sized shielded birdcage RF coil designs, resulting in degraded B₁-field performance. In the aforementioned numerical solutions, quasi-static conditions were assumed in calculating the B₁-field, and the electromagnetic (EM) interaction between the RF coil and the target object placed inside the RF coil was disregarded [18]. Many studies were conducted to solve problems pertaining to EM interactions between the RF coil and the target object by solving Maxwell’s equation using the EM toolbox based on various solvers, such as finite element method (FEM), finite difference time domain method (FDTD), method of moments (MOM), hybrid approaches, and others [19,20,21,22,23,24]. Through these EM simulations, EM parameters such as EM fields (e.g., B₁- and E-fields) and the specific absorption rate (SAR) related to safety could be calculated to evaluate the RF coil performance [25,26,27]. However, these EM simulation studies did not account for the RF circuitry, such as tuning, matching, and decoupling conditions at the excitation port of the RF coil, which can alter the total power balance of the RF coil as well as B₁-field and SAR distribution [28]. A two-way link between RF circuitry and EM simulations was demonstrated for the RF coil co-simulation approach, which offered significant time reduction in RF coil analysis with a human model via tuning/matching as a post-simulation adjustment [29]. To obtain the optimal lumped element value for the RF coil, a network co-simulation approach was used [30]. This method yielded variations in the SAR estimates based on the decoupling between the excitation ports. In particular, while designing an RF coil for RF transmission, an accurate SAR distribution is crucial for ensuring RF safety standard guidelines [31].

Apart from numerical and EM simulation considerations, physical constraints such as conductor length, conductor width, and ratio of RF shield diameter to RF coil diameter affect the homogeneity of the B₁-field and the transmission power efficiency of the RF coil [32]. RF shields are used in every MRI system to prevent the coupling of the RF coil with other RF electronic hardware, specifically those with gradient coils [33]. According to various commercial MRI system specifications, the minimum distance between the RF shield and the RF coil is typically at least 5 cm. Hence, a circular birdcage RF coil placed in close proximity to an RF shield (less than 5 cm) becomes extremely difficult to implement because of the difficulty in tuning/matching and the poor isolation between ports, which result in poor transmission efficiency. Considering the outline of the average human body, an elliptical birdcage whole-body RF coil can be advantageous for improving the B₁-field homogeneity and its relative SNR [34]. However, large-size symmetric elliptical birdcage RF coils have lower efficiency than circular birdcage RF coils, which can be improved through geometry optimization to generate an acceptable homogeneous B₁-field and lower SAR depositions.

In this study, a split-leg type asymmetric elliptical birdcage RF coil for whole-body MRI was proposed to generate a homogenous B₁-field within a larger imaging space horizontally compared with that of a circular birdcage RF coil. An extended RF coil circuit co-simulation approach was also proposed to calculate the optimal capacitor for the proposed RF coil. Both quadrature (same amplitude, 90° phase difference) and elliptical (different amplitudes and phase differences) driving approaches were evaluated through EM simulations in this study [35]. To implement the elliptical driving (or multiport driving) approach in an MRI system, more than one RF amplifier is required; however, this is not available in every MRI system. Therefore, we suggested a semi-elliptical driving approach, where the set of vertical (#1, #3) and horizontal (#2, #4) opposite ports had the same amplitude with a 180° phase difference, and adjacent ports (#1, #2, #3, #4) had different amplitudes with a 60° phase difference. High-power RF phase shifters and RF attenuators were used to adjust the phase and the power attenuation at the ports, respectively. A practical design of a four-port elliptical high-pass birdcage with asymmetric leg distribution was adopted for narrow-bore MRI (magnet bore inner diameter, 64 cm). EM simulations were performed to evaluate the following parameters: (1) optimal end-ring capacitor value for the desired resonance mode, (2) decoupling between the excitation ports, (3) transmitted RF field (B₁⁺-field) uniformity, (4) transmission efficiency, and (5) SAR field distribution.

2. Materials and Methods

2.1. EM Simulation and Optimization

The objective of the EM simulation study was to optimize the elliptical birdcage design by evaluating the parameters presented in the introduction section. The initial values of the capacitors were selected from the birdcage builder toolbox [36]; however, they failed to generate the desired resonance mode and homogenous B₁⁺-field for the large-sized shielded birdcage RF coil. Several numerical approaches have been reported to calculate the optimum capacitor values; however, they require extensive numerical calculations and precise accuracy in self- and mutual inductance calculations. The accuracy of the inductance calculation degrades in asymmetric and curvy structures because of ideal assumptions in the theoretical analysis. Thus, an RF coil co-simulation approach was implemented in this study to calculate the optimal capacitor values. First, all the lumped elements were replaced with excitation ports, and the full port S-parameters were generated via EM simulation and then exported to MATLAB (The MathWorks, Natick, MA, USA) to calculate the optimal capacitor by minimizing the cost function for the desired mode at the resonant frequency (further details are provided in Supplementary Materials Section A), as Equation (1):

$$arg min\left\{\left|S_{i=j}^{port}+\lambda \left(min\left(S_{i\ne j}^{port}\right)\right)\right|\right\}$$

(1)

where $S_{i=j}^{port}$= S-parameter of the individual i port; lambda ($\lambda$) = trade-off parameter between decoupling and fine tuning; $min\left|S_{i\ne j}^{port}\right|$ = S-parameter between the i, j port.

In circular birdcages, same-value capacitors at the end-rings cause horizontal and vertical resonance modes to occur at the same frequency. However, in an elliptical birdcage, different-value capacitors are required at the end-rings such that the horizontal and the vertical resonance modes occur simultaneously at the same frequency. After the optimal capacitors were calculated, tuning and matching were performed through the RF circuit matching toolbox of the simulation software. All RF coil modeling and full-wave EM simulations for this study were performed using the commercially available software Sim4Life based on the FDTD solver (ZMT, Zurich MedTech AG, Zurich, Switzerland). The simulations were performed on a cylindrical uniform phantom (diameter, 50 cm; length, 50 cm) with a low dielectric constant (electrical conductivity of 0.066 S/m; relative permittivity of 13.69) for B₁⁺-field uniformity evaluation; furthermore, a multi-tissue three-dimensional (3D) human model, “Duke”, obtained among virtual population models [37] was used to evaluated the SAR and the transmission efficiency. The B₁⁺-field uniformity was evaluated in a 3D region of interest (ROI) measuring 45 cm × 41 cm × 35 cm to calculate the worst case uniformity in the cylindrical uniform phantom. The total input power for the B₁⁺-field and the SAR fields was normalized to 1 µT. Figure 1 shows the modeled circular and various elliptical birdcages evaluated in this study.

2.2. RF Coil Design and Construction

A modified version of an elliptical birdcage (named split-type elliptical birdcage) with 16 legs was constructed for the 1.5 T (63.87 MHz) MRI system with the following dimensions: major axis, 62 cm; minor axis, 58 cm; length, 60 cm. The legs and the end-rings were prepared using copper strips with a thickness of 0.15 mm and widths of 3.5 cm and 5 cm, respectively. Four split legs were placed such that the distance between the neighboring leg and the gap between the leg splitting were identical, and the width of the split legs was 1.75 cm. Acryl was used to fabricate an RF coil supporting frame, which comprised a 7.5 mm-thick elliptical surface (major axis, 62 cm; minor axis, 58 cm; length, 100 cm) to place the RF coil. High quality factor (Q) and high RF power capacitors (DLC70E, Dalicap Tech. Corporation, Dalian, China) were attached at the end-ring to tune the RF coil to the desired resonance mode. Instead of using a single-valued capacitor, the three parallel capacitors were used to ensure no capacitor breakdown due to high RF power. The tuning and the detuning of the RF coil operating frequency was controlled using an MRI controller by supplying forward and reverse DC bias voltages in the detuning circuit [38].

The detuning circuit utilized at each leg comprised a high-voltage pin-diode (MA4P4000 Series, MACOM Technology Solutions, Lowell, MA, USA) with a parallel resistor to ensure equal bias currents (4.9 kΩ CMF65, Vishay Intertechnology, Malvern, PA, USA) in a series network with RF chokes (4611-RC, BOURNS, Riverside, CA, USA) to obstruct high RF power. To prevent the flow of common-mode currents on the shield of the coaxial cable [39], shielded coaxial cable traps or ground breakers (design process is described in Supplementary Materials Section C) were attached after the matching circuit followed by a high-power coaxial cable (RG-214). The scattering (S) parameters and the Q measurement (at 3 dB bandwidth) using two weakly coupled sniffer probes placed across the RF coil were measured for those unloaded and loaded with light (1 g NaCl per 1000 mL of distilled water) and heavy (6 g NaCl per 1000 mL of distilled water) phantom loads inside the MRI bore using a network analyzer (E5063A/N9913A, Keysight Technologies, Santa Rosa, CA, USA). After tuning and matching, they were connected to the two-port hybrid (90°) coupler through a splitter/combiner, as shown in the schematic illustration of the RF coil (Figure 2). Figure 2 shows the asymmetric leg distribution in the elliptic circumference, the unfolded view of the proposed split-type elliptical birdcage, the constructed prototype, and the schematic illustration of transmission and reception (TX–RX) implementation of a four-port birdcage RF coil through a two-port hybrid coupler TX/RX system.

2.3. Magnetic Resonance (MR) Experiment

The prototype elliptical whole-body birdcage RF coil was tested in TX–RX mode for a narrow-bore whole-body 1.5 T MRI system (Chorus, Coretech Co. Ltd., Seongnam-si, Korea) equipped with a 20 kW RF power amplifier (AN8102, Analogic Corporation, Peabody, MA, USA). The ports of the RF coil were connected to a hybrid coupler integrated with a TX/RX switch for quadrature excitation. B₁⁺-field mapping experiments were performed on a cylindrical phantom (inner diameter of 400 mm; length of 140 mm) comprising a copper sulfate solution (1 g CuSO₄·5H₂O per 1000 mL of distilled H₂O) to measure the B₁⁺-field uniformity, and in vivo experiments were performed on a healthy male human volunteer (30 years old, 70 kg) using a composite RF pulse B₁ mapping method (further details are provided in Supplementary Materials Section B) comprising two acquisitions with different RF pulses (#1–90°y–180°x–90°y and #2–90°y–0°x–90°y) in a single scan [40]. T₁-weighted spin-echo MR images were acquired for SNR measurement in both the phantom and the in vivo human volunteer. The sequence parameters are shown in

Table 1. MR sequence parameters.

	Uniform Phantom		In Vivo Human
Sequence	Composite Spin-Echo	T1-Weighted Spin-Echo	Composite Spin-Echo	T1-Weighted Spin-Echo
Nominal FA (°)	180	90	180	90
TE/TR (ms)	25/1600	15/500	15.4/600	13.2/600
FOV (mm2)	500 × 500	500 × 500	420 × 420	420 × 420
Acquisition matrix	256 × 256	128 × 128	128 × 128	256 × 256
Slice thickness (mm)	5	5	8	8
Average	2	2	1	1
Scan time (min)	10 min 38 s	4 min 16 s	4 min 1 s	2 min 41 s

.

For reference, all of the MR sequences were performed on a commercial two-port circular birdcage RF coil (20 legs; diameter, 58 cm; length, 60 cm). The study was approved by our institutional review board, and written informed consent was obtained from all volunteers before the experiments commenced.

FA maps, SNR maps, and other post-processing visualizations were reconstructed from raw MR data using MATLAB. The FA maps were smoothed using a 3 × 3 median filter for each image. The SNR maps were calculated in the acquired phantom and in vivo MR images. The mean SNR was calculated in the ROI using Equation (2), as follows [41]:

$$SN{R}_{mean}=0.66\times \frac{Mean signal of RO{I}_B}{Average standard deviation of RO{I}_s }$$

(2)

where the mean signal of ROI_B represents the mean signal values in a two-dimensional (2D) ROI (diameter, 30 cm) within the phantom, and the average standard deviation of the ROI_S represents the average standard deviation of noise in the four small ROIs (diameter, 6 cm) placed diagonally outside the phantom at the corner of the MR image. For the MR images acquired from quadrature volume RF coils, a factor of 0.66 (Rayleigh distribution correction factor) was necessarily multiplied to obtain the true SNR [42]. The FA uniformity was evaluated in a 2D ROI (diameter, 30 cm) within the acquired phantom MR images using Equation (3), as follows [41]:

$$Uniformity (\%)=\left[1-\frac{I_{max}-I_{min}}{I_{max}+I_{min}}\right]\times 100$$

(3)

where I_max and I_min represent the maximum and the minimum pixel intensity values in the FA maps, respectively.

3. Results

Using the RF coil circuit co-simulation approach, the optimal end-ring capacitors were calculated such that the reflection coefficients (S_ii) at each port and the isolation coefficients (S_ij) between adjacent ports were at least −15 dB before including the matching circuit. Figure 3 shows B₁⁺-field distributions (Figure 3a,b, respectively) in a uniform cylindrical phantom and a human model as well as the SAR distribution (Figure 3c) in the coronal (upper row) and the axial (bottom row) planes.

Table 2. Summary of uniformity, input power, transmit efficiency, and maximum B₁⁺ at 20 kW (most commercial MRI systems are equipped with 20 kW RF amplifier) in cylindrical uniform phantom; mean 10 g average SAR and peak 10 g average SAR in human “Duke” model for simulated RF coils in optimal driving mode, as shown in Figure 1.

	Case 1	Case 2	Case 3	Case 4	Case 5
Cylindrical Uniform Phantom
${\mathrm{B}}_1^{+}$ Uniformity (%)	74.29	70.21	69.68	73.28	76.25
Total input power (watt) normalized to 1 µT	24.41	33.03	36.52	37.16	29.89
Transmit efficiency $\left(\frac{{\mathrm{B}}_1^{+}}{\sqrt{P}}\frac{\mathsf{\mu }\mathrm{T}}{\sqrt{\mathrm{watt}}}\right)$	0.202	0.174	0.166	0.164	0.187
${\mathrm{B}}_1^{+}$ at 20 kW (µT)	28.63	24.61	23.40	23.20	26.47
Human “Duke” Model
Total input power (watt) normalized to 1 µT	21.57	31.46	25.96	30.37	24.21
Mean 10 g-avg SAR (watt/kg)	0.093	0.11	0.13	0.11	0.093
Peak 10 g-avg SAR (watt/kg)	0.78	0.89	1.44	1.16	0.97
Transmit efficiency $\left(\frac{{\mathrm{B}}_1^{+}}{\sqrt{P}}\frac{\mathsf{\mu }\mathrm{T}}{\sqrt{\mathrm{watt}}}\right)$	0.22	0.18	0.20	0.18	0.20
${\mathrm{B}}_1^{+}$ at 20 kW (µT)	30.45	25.21	27.76	25.66	28.74

summarizes uniformity, transmission efficiency, and maximum B₁⁺ at 20 kW in the cylindrical uniform phantom as well as mean 10 g average SAR and peak 10 g average SAR in the human “Duke” model, normalized to 1 µT in terms of B₁⁺ magnitude. The B₁⁺-field uniformities for a circular birdcage (diameter, 58 cm) driven with quadrature driving and the split-type elliptical birdcage (62 cm × 58 cm) driven with semi-elliptical driving were 74.29% and 76.25%, respectively. The simulated SAR values obtained for all the birdcage RF coils were within the IEC SAR limits (IEC 60601-2-33). The mean 10 g average SAR for the circular and the split-type birdcages was identical; however, the peak 10 g average SAR of the split-type birdcage was 19.56% higher than that of the circular birdcage.

The S-parameter and the Q measurements of the split-type elliptical birdcage performed inside the MRI bore are listed in

Table 3. Experimental S-parameters and Q-factor measurement for constructed split-type elliptical birdcage in unloaded and loaded conditions. S-parameter was measured at each splitter/combiner port, and Q-factor was measured using two weakly coupled sniffer probes placed at opposite ends of the RF coil.

	Unloaded	Light Loader Phantom	Heavy Loader Phantom
Reflection Coef. (S11)	−7.73 dB	−18.09 dB	−24.62 dB
Reflection Coef. (S22)	−5.05 dB	−16.03 dB	−20.35 dB
Isolation Coef. (S12)	−17.96 dB	−22.69 dB	−22.17 dB
Impedance (Z11)	35.07–j34.33 Ω	39.69–j4.33 Ω	44.45–j0.05 Ω
Impedance (Z22)	20.54–j31.6 Ω	38.61–j8.23 Ω	47.17–j8.93 Ω
Quality factor (Q)	173.91	79.07	67.43

. The unloaded-to-loaded Q ratios (Q_U/Q_L) were 2.2 and 2.6 for light and heavy loader phantoms, respectively, indicating sample dominance. One of the main difficulties encountered in achieving better isolation was the coupling between the coaxial cable traps and the RF shield, which improved significantly when cable traps were shielded compared with when they were unshielded. After the tuning and the matching of each port, the vertical ports (#1 and #3) and the horizontal ports (#2 and #4) were connected to a −3 dB splitter/combiner, and their respective S-parameters were measured. The reflection coefficient (S₁₁) and the isolation coefficient (S₂₁) at the splitter/combiner ports are listed in Table 3, indicating acceptable isolation between the horizontal and the vertical ports. The constructed prototype included a tuning/detuning circuit, which was tuned at a forward DC bias of +30 V and detuned at a reverse DC bias of −30 V.

Figure 4 shows the SNR map and the FA maps of the uniform phantom acquired in axial, sagittal, and coronal orientations using the commercial two-port circular birdcage and the proposed split-type elliptical birdcage. For each orientation, the RF pulse power was calibrated during pre-scan, as presented in Table 3. Furthermore,

Table 4. Pre-calibrated RF pulse power in cylindrical uniform phantom and in vivo human (min–0/max–1) for commercial circular birdcage and proposed split-type elliptical birdcage.

Loading	Coil	Axial	Sagittal	Coronal
Cylindrical Phantom	2-port commercial circular birdcage	0.14	0.13	0.12
	4-port proposed split-type elliptical birdcage	0.16	0.17	0.15
In Vivo Human	2-port commercial circular birdcage	0.14
	4-port proposed split-type elliptical birdcage	0.16

shows that the split-type elliptical birdcage required a moderately higher RF power (average, 18.68%) than the circular birdcage, which is consistent with the simulation analysis (18.33%).

The SNR measurement shows that the mean SNR of the circular birdcage was higher than that of the split-type elliptical birdcage in the axial and the sagittal slices by 10% and 14.85%, respectively; however, the mean SNR of the split-type elliptical birdcage in the coronal slice was higher by 9.45%. The FA uniformity measurement shows that FA uniformity of the circular birdcage was higher than that of the split-type elliptical birdcage in the axial and the sagittal slices by 1.77% and 3.66%, respectively; however, the FA uniformity of the split-type elliptical birdcage in the coronal slice was higher by 0.86%. The mean SNR and the FA uniformity of both coils in a large FOV were similar, demonstrating the acceptable performance of the proposed split-type birdcage RF coil.

Figure 5 shows the SNR and the FA maps of the in vivo human volunteers acquired in the axial and the coronal orientations. The SNR maps showed the reasonable performance of the proposed coil in comparison with that of the commercial birdcage coil. The FA maps of the in vivo human volunteers were obtained after B₀ correction, similar to the original composite RF pulse B₁ mapping method [40]. The FA maps in the axial orientation showed similar results for both coils; however, in the coronal orientation, the FA maps indicated a slightly lower sensitivity of the proposed coil, which was due to the slight mismatch in the slice position. The proposed coil for the MRI experiments was only used in transmit and receive modes; however, if operating only in the TX mode, a dedicated multichannel RX RF coil can be utilized to further increase SNR and receiver sensitivity.

4. Discussion and Conclusions

The aim of this study was to present an RF coil circuit co-simulation approach to calculate the optimal capacitor for four-port split-type asymmetric elliptical birdcage RF coil and its construction procedure, which can generate a homogenous B₁⁺-field and an acceptable transmission efficiency for the narrow bore 1.5 T MRI system. The RF coil circuit co-simulation approach previously reported was implemented for a multi-channel surface loop RF coil [29], which did not have characteristic B₁ resonance modes, as observed in the birdcage RF coil. In this work, the RF coil circuit co-simulation approach was extended to calculate the optimal capacitor of the proposed birdcage RF coil in the presence of loading as well as coupling between the excitation ports, which required more careful consideration of the optimal capacitor value to generate the desired resonance mode. The decoupling between the excitation ports was very crucial, as they had significant effects on the input power and the SAR deposition (effect of decoupling is shown in Supplementary Materials Section A). Additionally, the circular birdcage RF coil resonance modes occurred at the same frequency when using identical-value end-ring capacitors, however, the elliptical birdcage RF coil had horizontal and vertical resonance modes which needed to be maintained at the same resonant frequency using different-value end-ring capacitors. During the evaluation of the elliptical birdcage, it was discovered that the homogeneity of the B₁⁺-field could be increased further if the leg positions became asymmetric around the circumference and the position of the legs shifted away from the shield in the horizontal axis. However, asymmetric legs had less B₁⁺-field distribution near the empty region of the birdcage coil. To improve this, the legs were split, and the gaps between the splits were maintained identical to the gap between adjacent non-split legs. This method improved the B₁⁺-field distribution in the aforementioned regions with a moderately higher peak 10 g average SAR compared with the circular birdcage. To further improve the B₁⁺-field distribution, elliptical driving was implemented because the horizontal and the vertical modes had different B₁⁺-magnitudes at the center of the RF coil. During EM simulation post-process analysis, it was discovered that the 60° phase difference between adjacent ports and a 180° phase difference between opposite ports generated the maximum uniform B₁⁺-field distribution. Similarly, the different amplitudes of the input power at the horizontal (#2 = #4) and the vertical (#1 = #3) ports were investigated. It was also discovered that the B₁⁺-field uniformity increased with the amplitude difference between the horizontal (#2 and #4) and the vertical (#1 and #4) ports up to 12 dB in the uniform phantom and up to 6 dB in the human “Duke” model. It is noteworthy that the input power of the vertical ports was lower than that of the horizontal ports. Furthermore, the four-port implementation in the closely shielded elliptical birdcage could be implemented if the ports were located diagonally, as in the case of the two-port circular birdcage. However, the maximum B₁⁺-field uniformity achieved in this state was 5% lower than that achieved by horizontal–vertical port feeding.

Based on the EM simulation, an experimental prototype of a split-type asymmetric elliptical birdcage was constructed and verified through MRI experiments. The calculated and measured end-ring capacitor values varied from 10% to 15% owing to the change in the ideal condition in the EM simulation to real complex conditions in the MRI system. During construction, one of the main problems encountered was the effect of the unshielded coaxial cable traps on the frequency response due to its coupling with the coaxial cable and the RF shield as well as the different loading conditions, which were solved by shielding of cable traps (further details are provided in Supplementary Materials Section C). The MRI experiments were performed with quadrature driving because of incomplete resources for elliptical driving, such as additional RF amplifiers, TX waveform generators, high-power isolators, and high-power RF dummy loads, to allow precise handling of the reflected power for guaranteeing the safety of both the MRI hardware and the attendants. The experimental analysis showed that the SNR of the acquired MR image through the birdcage coil can be further improved using a dual-preamplifier at reception [43]. Compared with the commercial two-port circular birdcage, the proposed elliptical birdcage requires approximately 20% higher RF input power, which is reasonable.

In context with the birdcage builder toolbox [38], the birdcage builder only performs the numerical calculations with symmetric circular and elliptical configurations. It was found that the birdcage builder had higher uncertainty than the proposed method, especially for the larger-body size birdcage RF coil. The absolute errors in calculated capacitance for the circular birdcage coil using the birdcage builder and the proposed method were 32.3% and 11.86%, respectively, in comparison to the practical capacitance value used in our commercial system. The birdcage builder cannot be used for the split-leg elliptical birdcage RF coil due to unavailability of asymmetric configuration, however, the absolute error between the proposed method and the practical values was 12.05%. The difference between the calculated and the practical values was due to the effect of unspecified configuration of the RF shield and other components such as patient bed, gradient coil, and magnet of the MRI scanner. During the construction of an elliptical birdcage RF coil, it was realized that there could be an effect from the number of legs on the magnitude of isolation between the ports, which was not investigated in this work. Additionally, if the proposed RF coil is implemented in wide-bore MRI systems with an optimal RF shield geometry, then the B₁⁺ field uniformity and the transmission efficiency as well as the isolation between ports can be further improved, because the B₁⁺ magnitude increases when the shield-to-coil-diameter ratio decreases [44,45].

In conclusion, we report an RF coil circuit co-simulation approach for the optimization of a four-port split-type asymmetric elliptical whole-body birdcage coil and its construction design. We evaluated the coil performance through phantom and in vivo human experiments using a 1.5 T MRI system. The elliptical configuration permitted a larger imaging space horizontally without degrading the B₁⁺-field uniformity and the transmission efficiency considerably compared with the commercial smaller circular birdcage RF coil. The RF coil circuit co-simulation approach showed efficient strategy for tuning, matching, and optimization of the RF coil; hence, it can be applied to any desired resonant frequencies and any arbitrary geometries of the birdcage coil. The proposed split-type elliptical birdcage coil MRI experiments demonstrated acceptable SNRs and B₁⁺-field uniformities in an adequate FOV for whole-body applications in MRI.