Local gradient coils have the advantage to be highly efficient; maximum gradient strengths and slew rates are typically much higher compared the built-in whole-body gradient coils. Sites that employ insert gradients usually disable the regular gradient system by disconnecting the built-in gradient from the amplifiers. An alternative approach is to leave the standard gradient system as-is, and treat the insert gradient as an additional gradient axis1. This way, the regular gradient system can compensate any eddy currents (ECs) from the gradient insert induced in conducting parts of the system. Consequently, active shielding of the gradient insert is not required anymore, making the gradient design much more versatile. This can result in smaller, lighter and more efficient gradient inserts. In this feasibility study we demonstrate the design of a high power Z-gradient insert for the human breast at 7T and investigate if a body gradient coil can compensate the ECs of this asymmetric gradient insert coil, connected to a fourth gradient amplifier.

A breast gradient coil producing a linear field in the z-direction was designed with an efficiency of 1mT*m^-1*A^-1 and constructed from hollow copper wire. The coil was placed in a 7 tesla whole body MR system (Philips Healthcare, Cleveland, OH). Via a force compensated power cable, the coil was connected to an additional gradient amplifier (Copley Controls, Canton, MA). A dynamic shimming unit (Resonance Research, Inc., Billerica, MA) was connected to the amplifier to create trapezoidal waveforms. Field characterization was done using a stand-alone field camera system (skope magnetic resonance technologies, Zürich, CH). See figure 1 for overall setup.

Eight field probes were positioned such that five different field components could be distinguished: a B0 offset, the X, Y and Z gradients of the system, as well as the Z gradient generated by the gradient insert.

After switching the current going through the gradient insert coil, short and long term ECs were measured with field cameras. Short ECs were measured for 2ms with a resolution of 1μs, and long term ECs were measured for 5s with a resolution of 100ms.

A B0 map of a spherical phantom was acquired with a 0.01% drive of the gradient amplifier (constant field of 0.1mT/m) to the insert gradient. To assess force behavior of the insert, the gradient was driven with 10ms pulses at 50% drive (300mT/m). Heating of the gradient coil was assessed by placing a temperature probe in a small hole in the casing of the coil, near the conductor.

The heating was assessed with and without active cooling, with a constant field of 30mT/m and subsequently 60mT/m generated by the coil for approximately 7 minutes. Active cooling was achieved by pumping glycol through the copper wire, which in its turn was cooled in a heat exchanger (Thermo Electron Corporation).

Figure 2 shows a B0 map reflecting the efficiency of the coil of 1mT*m^-1*A^-1. When driven at 50% of the gradient amplifier drive, 300mT/m was realized. As the coil is force compensated, the coil was driven at high power reproducibly in the field of 7T.

Figure 3 shows the short term ECs (<2ms). These ECs are the self-term and a B­0 offset, while the other terms (x, y, z) are negligible.

Figure 4 shows the long term ECs. In contrast to the short term ECs, the long term ECs are a Z gradient EC and a B0 offset, while the ECs of the insert gradient are absent.

Figure 5 shows the heating curves of the gradient coil with and without active cooling. With active cooling, the temperature of the gradient coil could easily be controlled within safety guidelines.

The non-shielded gradient insert could easily provide a high field of 300mT/m driven with only 50% of the amplifiers performance. Using the hollow conductors, temperature was managed to remain within patient acceptable values.

Excellent characterization of the spatiotemporal field patterns was achieved with the field cameras. The short term ECs of the non-shielded gradient insert are in the order of 2ms, which can easily be compensated with pre-emphasis. The long term ECs, most probable from the magnet cryostat, are predominantly reflected as a Z gradient field. The built-in (Z-)gradient coils can compensate this, and therefore the EC can be treated as a cross term for the insert gradient coil.

An efficient high power gradient insert coil has been constructed without intrinsic shielding. Using the in bore gradient setup, long term ECs can be compensated facilitating versatile gradient designs (i.e. reduced spaced requirements, light weight and high efficiency).