Introduction

There is a number of MRI applications which would benefit from being able to acquire data for several nuclei in a single imaging session^1,2. For example, one could improve hyperpolarized 13C imaging using 23Na coil sensitivity profiles.
Two strategies to achieve multinuclear imaging have been adopted so far¹. One is to use a dedicated coil for each nucleus of interest with integrated trap circuits to prevent coil crosstalk. Another strategy is to use double- or triple-tuned matching networks in a single coil element. Both approaches to a greater or lesser extent lead to a bulky imaging setup and some degradation of SNR compared to standard single nucleus systems.
In this work we propose a different approach, which is more suitable for detecting several nuclei having similar gyromagnetic ratios. It is based on extending the bandwidth of the preamplifier decoupled detector coil at the expense of the noise figure of the preamplifier. The entire circuit is as compact as a single nucleus MRI coil, but can potentially cover 13C (32.1 MHz@3T), 23Na (33.8 MHz@3T) and 129Xe (35.3 MHz@3T) frequencies with controllable SNR tradeoff and improved decoupling levels. The approach is based on further extending the theory of preamplifier decoupling using compact matching networks³. Using this theory, the achievable bandwidth is analyzed as a function of preamplifier noise figure. A design example is shown to demonstrate the approach.

Methods

Our previous work^3,4 has already indicated, that different topologies of matching and decoupling networks can result in different bandwidth of MRI receive coils. For example, for some combination of coil and preamplifier impedances, the matching/decoupling Pi-network can result in a wider bandwidth compared to T-networks even though both networks in the ideal case provide the same level of SNR and preamplifier decoupling. A schematic representation of the analyzed setup with the implemented matching/decoupling Pi-network as an example is shown in Fig. 1(a).
Further extending this theory^3,4, it can be shown, that the preamplifier can be noise matched to an impedance different from optimal noise impedance. This will degrade the noise figure of the preamplifier, however, at the same time, might lead to a wider SNR bandwidth. The resulting trade-off between the 1 dB SNR bandwidth and the preamplifier noise figure is shown in Fig.2.
Using the coil in Fig. 1 and a standard low noise amplifier³ at a mid-band frequency 33.7 MHz, Pi-networks can achieve a bandwidth increase by sacrificing the noise figure by a fraction of a dB. This, however, is not the case for the T-networks evaluated in this example.
One useful feature of this approach is that since the magnitude of the reflection coefficient at the input of the preamplifier is increasing with NF, the preamplifier decoupling also increases. The calculated preamplifier decoupling versus NF is shown in Fig. 3.
As can be observed in Fig. 3, the more noise figure of a preamplifier can be sacrificed, the higher the preamplifier decoupling that can be achieved. This is particularly useful in flexible receive arrays, where ensuring sufficient coil decoupling in different conformations is critical.

Results

To demonstrate the approach, the matching/decoupling network in Fig. 1(a) was designed for NF_min and NF_min+0.2 dB. The coil used is an 8 cm in diameter² loop made of insulated 1 mm copper wire, as shown on the photo in Fig. 1(b). In case of matching to NF_min, the matching circuit elements in Fig. 1(a) are C₁ = 91.1 pF, C₂ = 8.9 pF and L₃ = 1.7 µH. In case of matching to NF_min + 0.2dB, the designed matching circuit elements in Fig. 1(a) are C₁ = 80.1 pF, C₂ = 19.9 pF and L₃ = 1.1 µH. The simulated SNR frequency sweep using Keysight ADS is shown in Fig. 4.(a).
As can be observed, the mid-band SNR when matched to NFmin (blue curve) is 0.2 dB higher than for the NFmin+0.2 dB case (red curve), however, the bandwidth is wider for the later case. The decoupling for the NFmin+0.2 dB case is also higher, as can be seen from the simulation results presented in Fig. 4(b). The used ADS simulation setup is shown in Fig. 5.
To verify the results experimentally, the coil in Fig. 1 has been constructed and is currently being tested. The results of the measurement results will be reported at the conference.

Discussion and Conclusion

An alternative approach to design MRI detectors for multinuclear imaging is to extend the bandwidth of the single coil. This can potentially be achieved in receive coils by spoiling the Q-factor using preamplifier decoupling approach. The simulation results shown that sacrificing a small amount of preamplifier noise figure a stronger Q-factor spoiling can be achieved leading to wider bandwidth and also better decoupling.

Acknowledgements

The authors would like to thank the Danish National Research Foundation (grant DNRF124) for partial support of the activities.