Dipole antennas are being used increasingly for body imaging at 7T [1-3]. Although local SAR levels underneath the elements are already relatively low, the antenna designs without a dielectric substrate can further reduce their SAR levels by increasing the antenna-subject spacing. However, this will increase inter-element coupling. Although arrays of dipole antennas have intrinsically low inter-element coupling without adding additional circuitry, the increased inter-element coupling may cause considerable scattering losses which will result in reduced efficiency. In this study we investigate the relationship between antenna-subject spacing, inter-element coupling and maximum local SAR levels for the fractionated dipole antenna [1]. We introduce with an improved antenna design with larger antenna-subject spacing and demonstrate its imaging performance and local SAR levels.

We investigated inter-element coupling of two fractionated dipole antennas positioned at varying inter-element distance on a large elliptical ASTM phantom filled with saline water. Coupling was measured without matching circuitry using the formula $S_{12norm} = S_{12}^2/(1-S_{11}^2)$. The fractionated dipole antennas have a polycarbonate placeholder that guarantees 20 mm distance from the imaging subject. Graphs of coupling vs inter-element distance were obtained with three additional layers of spacing of 25, 40 and 55 mm. The same setup was simulated (Sim4life, ZMT, Zurich).The simulation setup is depicted in Figure 1 (c). In the electromagnetic model there are two printed meandered dipole antennas on a plexiglass substrate, positioned on a water phantom. Simulations were performed for varying inter-element distances (0 – 60 mm) and for 3 additional layers of foam (25, 40, 55 mm).

The same simulations were used to acquire the local SAR distributions underneath the active element for the various antenna-phantom spacings.

As a result of these investigations, a new antenna array has been realized with increased subject-antenna spacing (Fig 4a). In addition to a 17 mm plexiglass layer, 23 mm of foam was attached to the elements resulting in a total thickness of 40 mm. To demonstrate the reduction in SAR_10g, the new array and the original fractionated dipole antenna array were simulated for a prostate imaging setup on the Virtual Family model ‘Duke’ [4] (Fig. 3a). SAR_10g distributions were evaluated by calculating the SAR for a sum-of-magnitude electric field distribution which is a severe over-estimation of the real SAR levels but serves well to compare the two array setups (Fig. 3b,c).

Finally, the new array with thicker spacers has been used for prostate imaging on a healthy volunteer (informed consent). The in-situ coupling matrix has been measured using directional couplers. B1 maps and T2w images (TR/TE=2500/90 ms, 0.5x0.5x3 mm3, TSE-factor 9) have been performed using the original array of fractionated dipole antennas and the new array with thicker spacers on the same volunteer.

Figure 1 (a,b) shows the resulting graphs of inter-element coupling vs inter-element distance for various thicknesses of the additional foam layer between the antennas and the phantom. As expected, coupling increases with additional spacing. Figure 2 shows the local SAR distributions for the investigated spacings. As a result of the additional spacing, SAR levels are reduced. However, B1+ measurements indicate that the efficiency of (isolated) antennas remains the same. Question remains whether the increased coupling does not result in increased scattering losses resulting in lower B1+ for an array setup. Therefore, an array setup has been realized and simulated. Again, simulations indicate that B1+ values inside the prostate are the same for both the original array and the array with thicker spacers (Fig.5). However, the maximum SAR_10g level for the array with thicker spacers is reduced by 45 %.

The measured S-parameter matrix for the array with thicker spacers is indicated in Figure 4(c). Clearly, the inter-element coupling levels are not dramatic (-12 dB for nearest neighbors, much less for next-nearest neighbors and beyond). The reflection levels of element 2 and 3 are a bit too high. But this does not result in dramatically less B1+ as can be seen in the B1+ maps for the original fractionated dipole antenna array and the new array with thicker spacers (Fig.5). These results show that actually even more B1+ is reached for the array with thicker spacers. T2w images for both setups are indicated in Figure 5.

The original array of fractionated dipole antennas for body imaging at 7T uses 20 mm spacing. This study shows that the spacing can be increased without compromising image quality but with the advantage of a 45 % lower SAR level. Although the inter-element coupling increases with the increased spacer thickness, coupling parameters for the investigated subject remain below -12 dB.