Introduction

A wide variety of different coil array designs have allowed significant progress for body imaging at ultrahigh-field (UHF) (e.g., using local Tx/Rx loop^1-4 and dipole antenna arrays^5,6). Significant advancement of both $$${B_1^+}$$$-shimming and parallel imaging capabilities were demonstrated for in-vivo cMRI in pigs using an antisymmetric 8Tx/16Rx loop array⁷. To obtain the best image quality for different pigs (25-90kg), three dedicated rigid coil arrays were implemented. These coil arrays have a rigid form and are thus optimized for MR-experiments on pigs with a certain pig weight and thorax shape. However, building an individual coil for each weight range is both time and cost-consuming. The purpose of this work is to design, simulate, implement, and get initial usage experience with elastic coil array, which can be used potentially for cMRI in pigs of a wide range of thorax dimensions and weights (25-90kg).

Methods

In this work, we introduce a dedicated transceiver array printed on a flexible 100-µm thick Kapton polyimide (PI) substrate (ε_r=3.5 and $$$tanδ$$$=0.001). The copper (Cu) trace width is 4mm with 35-µm thickness. The array is composed of 16-elements distributed anti-symmetrically around the central two elements⁷ [Figure 1]. The sizes of the elements 1, 2, 7, 8, 9, and 10 were 4.5×9.8cm². The decoupling between the central two elements was accomplished using a common conductor and shared decoupling capacitor (SDC) ($$${C_1^d}$$$). The size of elements 3 and 4 was 5×8cm². Elements 3 and 4 were decoupled from elements 1 and 2 using SDC ($$${C_2^d}$$$). The size of elements 5 and 6 was 5.6×12.2cm². The sizes of elements 7 and 9 and the identical elements 8 and 10 are 4.5×9.8cm². They were decoupled using a shared decoupling capacitor ($$${C_3^d}$$$). Elements 1, 2, and 3 were decoupled from the neighboring elements 5 and 6 using capacitive decoupling ($$${C_5^d}$$$, $$${C_6^d}$$$, and $$${C_7^d}$$$) in addition to a decoupling gap of 2cm. The external dimension for the array was 57.9×20.8cm². EM-simulations were performed using CST-Microwave-Studio for coil design and $$${B_1^+}$$$-field optimizations. For matching, tuning, and decoupling, RF-circuit co-simulation was employed in CST-Design-Studio to get an initial guess for the optimal lumped elements⁸. Static phase-only $$${B_1^+}$$$-shimming was performed within a pig phantom using two different optimization cost functions ($$${FC_1}$$$ for $$${B_1^+}$$$-homogeneity and $$${FC_2}$$$ for weighted combination of $$${B_1^+}$$$-homogeneity and transmit efficiency⁷. To characterize the Tx performance of the array, $$$CoV=\frac{std(B_1^+)}{mean(B_1^+)}$$$ and $$$Tx_{eff}=\frac{mean(B_1^+)}{\sqrt{P_{acc}}}$$$ were computed. To form an 8Tx/16Rx array compatible with the pTx system, every two neighboring elements were combined in one Tx-channel [Figure 1a]. All measurements were performed on a 7T whole-body MAGNETOM Siemens Terra scanner. The array was tested in phantom with the measurement of SNR-maps, $$${B_1}$$$-shimming, g-factor with high parallel imaging acceleration factors. The reconstructed experimental g-factor and SNR-maps were post-processed in MATLAB. Female pigs (German Landrace) were used after approval by the responsible local animal welfare committee (project 55.2 DMS 2532-1134-16, Regierung von Unterfranken). All in-vivo measurements were done using a GRE cardiac sequence (BEAT) provided by the scanner vendor. To visualize RF-excitation coverage of the heart volume, 10 coronal, sagittal, and transversal slices positioned within the heart were acquired with prospective triggering. Imaging parameters were: TR=44ms, TE=2.9ms, matrix size=150x120, FOV=300x300mm², slice thickness 6mm, inter-slice distance 1.2mm, FA=20°. Then CINE-imaging with a retrospective reconstruction of 30 heart phases was performed for standard cardiologic views of the heart: four-chamber, two-chamber, and SA.

Results

Figure 2 shows the simulated $$${B_1^+}$$$-field distribution for the 16-element flexible array after static phase-only $$${B_1^+}$$$-shimming within a dedicated pig phantom with PV1 and PV2 compared to the zero phases. With PV1, $$$CoV$$$ was improved by 70% in the central transversal slice compared to zero-phases. With PV2, $$$Tx_{eff}$$$ was improved by 3-times compared to zero-phases. Figure 3 shows the measured g-factor maps (using acceleration factors R=2, 3, 4, and 6), the SNR-maps, the noise correlation matrix of the flexible array loaded with a pig phantom, and in-vivo pig. The mean SNR within the heart region from the novel design was 70. With the implemented hardware phases and even prior $$${B_1^+}$$$-shimming, the coil shows good Rx properties with high SNR. Figures 4 and 5 show in-vivo images of a 45-kg pig with normal and high resolutions acquired using the flexible array. The developed coil demonstrates sufficient $$${B_1^+}$$$-field homogeneity within the pig heart without significant destructive interference.

Discussion

The initial experience of using the coil in the animal study has demonstrated that the fully elastic array provides sufficient $$$Tx_{eff}$$$ with $$${B_1^+}$$$-penetration on the full depth of the pig heart and coverage of the whole heart volume. The reasonable imaging quality with SNR moderately lower than delivered by the dedicated rigid array optimized for specific weight range was shown. Further optimizations of the tuning/matching circuits should be performed to make possible application of the fully elastic array for the wide range of the animal’s thorax shapes.

Conclusion

A dedicated flexible pig array with the best bendability to conform to the thorax shape of several pigs was introduced. All fixed capacitors, cable traps, and semi-rigid cables were fixed on the flexible PI-substrate without getting broken from several bending. Further optimizations of the tuning/matching will be performed to optimize the array for a wide range of pig weights.

Acknowledgements

This project was funded by the Federal Ministry of Education and Research (BMBF), Grant/Award Number: 01EO1004 & 01EO1504.