Synopsis

Keywords: Physics & Engineering: RF Safety

This educational talk on Parallel Transmit RF safety gives an overview of the challenges and potential benefits of parallel transmit RF coils with respect to RF heating. The concept of local SAR supervision in contrast to conventional global SAR supervision is introduced. The basic process of obtaining and implementing a safe local SAR model for parallel transmit coils is outlined. This includes RF field simulations, SAR model compression, validation and online supervision. Finally, potential benefits of parallel transmit RF coils to control and even reduce SAR are discussed.

Abstract

The introduction of parallel transmit RF coils to mitigate B1+ inhomogeneities especially in ultra-high field MRI significantly increased the complexity of SAR assessment and limitation strategies required to comply with regulatory standards [1].

Several publications assist users who build experimental RF hardware to properly validate their RF coils and obtain a safe SAR model for patient protection [2,3].

Typical conventional single channel transmit coils are only required by [1] to comply with global SAR limits. In this case, SAR can be calculated based on power measurements:
$$ SAR(t) \propto (P_{fwd}(t) -P_{ref}(t)) $$
For these coils, the spatial distribution of local SAR is constant.

For multi-channel RF coils, this approach does no longer allow safe subject scanning. The fields generated by different transmit channels that are driven independently superimpose at each location in different ways and thus create a situation where the local SAR distribution is changing depending on how the RF coil is driven. Local hot spots that damage tissue may occur. Therefore, [1] requires local SAR supervision for parallel transmit RF coils.
Local SAR is now calculated as a function of the measured RF voltage vector $$$U(t)$$$:
$$ SAR(x,t) = U(t)^HQ(x)U(t) $$
Because the local SAR distribution cannot be measured easily, a pre-calculated SAR model based on electro-magnetic field simulations is used to convert measured voltages to SAR. In the equation above, the matrix $$$Q(x)$$$ represents this model.
This matrix needs to be obtained for every 10g volume of tissue. For a typical SAR model, $$$10^6 - 10^7$$$ Q-matrices are required.
To allow fast online supervision of local SAR, this SAR model can be compressed using virtual observation points (VOP) [4].
Furthermore, these compressed models allow the user get a better understanding of where a given RF coil may create hot spots. It also allows to assess required safety margins to compensate for measurement errors.
Finally, VOP compressed SAR models not only allow the supervision of local SAR, but also allow controlling local SAR by employing specialized RF pulses.

Acknowledgements

No acknowledgement found.