Imaging subjects with implanted devices presents challenges for safe imaging as strong induced radio-frequency (RF) electric fields around metallic implants may lead to excessive local heating¹ quantified with the specific absorption rate (SAR). We have previously shown that it is possible to electromagnetically decouple a metallic prosthetic hip implant using an 8-channel parallel transmit (PTx) array². In this study, we explore the efficacy of using one, two, four and eight channels for RF shimming to ensure safety around the hip implant.

Our previous work demonstrated that it is possible to use B$$$_1^+$$$ map measurements to find operating modes of the PTx array that minimally couple to an implant². This used the mathematical formulation outlined by Etazadi et al³ and assumed that the lower, longer part of the prosthetic hip (see Figure 1) behaved as a wire. For an N-channel system, this yields (N-1) minimally coupled modes – otherwise known as null modes.

An electromagnetic simulation of an 8-channel PTx array⁴ loaded with the Duke voxel model⁵ with a hip prosthesis embedded within was performed in CST 2013. The generic prosthesis modelled was a Cobalt-Chromium implant with conductivity = 1.16x10⁶ S/m and density = 8425 kg/m³. The 8-channel array was combined into two and four channel systems using the reduction matrices described by Childs et al in Reference 6 which enabled the assessment of the behaviour for different channel numbers.

RF shimming was performed using just the null modes found from fitting radial field variations of the B$$$_1^+$$$. This is technically feasible in reality as these modes can be found from B$$$_1^+$$$ maps and just using these should ensure lower SAR around the implant. A constrained RF shimming optimisation was also set up using all the modes to see if more optimal solutions existed that traded off some of the high coupling mode to allow greater efficiency in terms of B$$$_1^+$$$/SAR. This latter optimisation, which would not be feasible in vivo, took as input the local SAR Q-matrices⁷ located in the region 1cm around the prosthesis surface and these were compressed down into a smaller set of virtual observation points⁸ with a 1% overestimate bound for speed.

The left side of Figure 2 shows the results of simply shimming with the null modes for the different numbers of channels. It is clear that as the number of channels is increased, the possibility to reduce SAR around the implant is also increased. However for two and four channels, the coefficient of variation in the B$$$_1^+$$$ in the imaging slice actually increases. The corresponding modes are not able to combine to produce better homogeneity.

The right side of the figure shows the results of the optimal constrained shimming using all the modes and it is clear that the added degrees of freedom improve the efficacy of the RF shimming, particularly for the two and four channel cases.

We have assessed the efficacy of RF shimming for decoupling a metallic hip prosthesis for different numbers of transmit channels. As expected, with increasing numbers of channels the potential to reduce SAR is increased. However, with fewer channels, the null modes produced may not be sufficient to ensure a homogenous B$$$_1^+$$$ field in the imaging slice.