In contrast to lower field strengths, MRI systems at high field strengths of 7 Tesla and above are not equipped with an integrated body transmit coil. Due to the severe problems with B1 inhomogeneity, volume resonators are not a good choice for body applications at ultra-high fields, and local multi-channel arrays are commonly used for transmission. In this work we present a 32-channel transmit/receive body array for 7 Tesla integrated between the bore liner and the gradient coil.

Figure 1a shows the array, which consists of 32 micro strip line elements with meanders arranged in 3 interleaved rings. The inner ring consists of 12 elements, while the outer rings consist of 10 elements each. The mechanical structure is made of a rigid polycarbonate frame (Fig. 1b) onto which the PCBs are glued. The front PCBs with the micro strip structure are 0.8 mm RO4003 boards with a length of 25 cm, while the ground plane is made from 0.127 mm RO3010 with 18 µm copper plating on both sides. The ground plane is slotted to reduce eddy currents induced by the gradient coil, the slots being place in such a way that the resulting copper strips on the front and the backside overlap. In this way, lumped capacitors only had to be used at the connection points of the end-capacitors. The overall length of the ground plane is 60 cm. The arrangement of the elements can be seen in Fig. 1c, where one half of the array is shown from the inside. The array is divided into an upper and a lower half to ease maintenance. The distance between head-on elements is 5 cm. The width of the front side PCBs is 86 mm.

All elements are connected to custom-built T/R switches¹ with an appropriate length of low-loss cable (Aircell 5, SSB Electronic GmbH, Lippstadt Germany) to ensure pre-amp decoupling. Detuning of the elements is performed with dedicated detuning boards containing PIN diodes that produce an RF short circuit of the transmit cable to ground when a forward current is applied.

While side-by-side neighboring micro strip lines with meanders are intrinsically well decoupled², they couple strongly when placed head-to-head or when they are next to each other but shifted in the longitudinal direction. Thus, decoupling networks were used to decrease coupling between some of the neighboring elements. Fig. 2 shows a schematic of the elements including the decoupling networks with the values of the lumped elements. There are 50 of these networks for the complete array. The connecting lines are semi-rigid cables (Huber+Suhner AG, Herisau, Swiss). There are no decoupling networks connecting the two halves of the array, where the adjoining elements are side-by-side.

To enable usage of the 32ch array together with local receive coils, a custom built PIN-diode controller was used to provide forward current and reverse voltage for the PIN diodes, and the receive chain of the MRI system was extended with 32ch switches and a 32ch 2nd-stage receive amplifier with bias tees for power supply of the pre-amps³.

A custom-built 32ch pTx system including custom RF amplifiers and RF modulators was used to drive the array during transmit.

The highest coupling between elements occurs where the two halves of the array connect. Here the coupling is -14 dB when loaded with a volunteer centered on the abdominal region.

Figure 3 shows the noise correlation for the array. Overall noise correlation is low; the maximum value is 22%. This value occurs between the elements that border the array halves.

Figure 4 is a table of the in vivo g-factors in an axial slice through the abdomen of a male volunteer (1.72m, 65kg), showing the means as well as maximums.

Figure 5 shows the first in vivo dataset. The entire body is covered in 4 stations. The images appear fairly homogeneous over a 500mm field-of-view.

The g-factors are quite high for an array with 32 channels. The reason for this is most probably the large distance between the array elements and the body, leading to high correlations between the sensitivities of the elements. To enhance the receive performance, a dedicated receive array could be used, which is feasible since the changes in the receive chain allow for computer-controlled switching between reception with the body array or a local receive array.First tests with the presented 32ch array show promising results. The entire body of a human volunteer could be covered quite uniformly in only 4 stations.