Introduction There are two approaches to design coil array for body imaging at ultra-high field: wrapping the transmit-receive elements directly on the body [1, 2] or creating a rigid stand-off array [3]. Placing the elements directly on the body provides better raw B1⁺ efficiency, but creates challenges for distributing the dipoles evenly around the body. Uneven placement can lead to common nulls in both circular polarized (CP) and gradient mode excitations. We propose a flexible 8 channel dipole array with trellis-like substrate which can expand and contract to fit various body sizes and hold electric dipole antenna elements in the optimum distribution around the body for all subjects.

Method Electric dipole elements were mounted on a trellis-like lattice of interlinked slats [4] (Figure 1). By squeezing or extending the structure, the circumference of the substrate can change from 60.5cm to 121cm. An animation of size changes is shown in Figure 2. Eight dipoles were evenly distributed circumferentially. Each dipole board has a hole and two slots which are used to secure it to the trellis with nylon screws. This allows the trellis structure be squeezed and extended without changing the orientation of the dipole. The distance between the dipoles changes from 7.6cm to 15.1cm. The dipole was 20cm long, 8mm wide, with two inductors for tuning the resonant frequency. A lattice balun was used to balance the current distribution on the dipole, with a variable capacitor for fine-tuning the match. Shielded RF traps were provided every 20cm in the cable to reduce the common-mode current and ensure the patient safety. Tuning and match were adjusted in the presence of large body phantom (ε_r=77, σ=0.5S/m) and a small body phantom (ε_r=63.5, σ=0.62S/m). The array was tested in a 7T 8ch pTx system (Siemens, Magnetom). A flexible yet fixed-element-spaced folded dipole body array [1] was used for RF field homogeneity comparison (Figure 1). Flip angle maps were acquired with a turbo-flash with preparation pulse sequence [5]. Phases to the elements were chosen to create constructive interference at the center of the phantom. SNR maps were generated from 2D GRE acquisitions (TR/TE/BW=2000/4.1/260, FOV=400×400mm, Matrix=128×128, slice thickness=5mm) with and without RF excitation. Human subject studies were performed according to institutional IRB. In-vivo liver images were acquired with a 2D FLASH sequence (TR/TE=7.0/2.3, BW=977/, FOV=400×400mm, Matrix=256×256, slice thickness=5mm) using a simplified TIAMO technique [6] with breath holding.

Results S21 between adjacent dipole elements varied from -15.4dB to -24.3dB as the distance between the adjacent dipoles changed. All elements were matched to lower than -20dB. Experimentally acquired flip angle maps of two different body phantoms are shown in Figure 3. With the small phantom the elements of both the folded dipole array and the trellis array were evenly distributed, and both transmit fields were relatively homogeneous. With the large phantom, gaps between the top and bottom sections of the folded dipole array lead to nulls. The trellis array maintains even spacing for the coil elements resulting in a much more homogeneous transmit field and no transmit nulls. To achieve a 90 degree flip angle in the center of ROI with the same RF pulse, the trellis array required 135 volts for the small phantom and 126 volts for the large phantom, while the folded dipole array required 156 volts for the small phantom and 160 volts for the large phantom. SNR maps in transverse and sagittal planes are shown in Figure 4. The trellis array has evenly distributed signal in both small phantom and large phantom, while the folded dipole array has nulls in the image in the large phantom. Figure 5 shows in vivo images acquired with the trellis array using a simplified TIAMO technique. This image verifies that there are no common nulls between the two modes with the trellis array.

Discussion The higher SNR and B1⁺ efficiency achieved with the trellis array is in part due to the shorter dipole elements compared to the folded dipole array. This also leads to a shorter field of view along Z direction, as can be seen in the sagittal SNR maps. Although the dipole array with trellis-like substrate was only demonstrated here for body imaging, at its smallest size it is suitable for head or knee imaging. This allows for the proposed array to be used to image a wide variety of anatomical targets.

Conclusion The flexible dipole array with trellis-like substrate can image different body sizes without creating common nulls between the CP and gradient modes, which is important for large field of view imaging.