Stephen Ogier1, Mary McDougall1,2, and Steve Wright1,2
1Electrical and Computer Engineering, Texas A&M University, College Station, TX, United States, 2Biomedical Engineering, Texas A&M University, College Station, TX, United States
Synopsis
Frequency translation is a technique that uses radiofrequency mixers to convert the received signal from one nucleus to another. This technique can be used to adapt 1H array receivers for use with other nuclei, such as 13C. This facilitates the use of arrays with less sensitive nuclei, which will benefit greatly from the SNR enhancement arrays provide.
Frequency translation has been shown to provide a flexible means of adapting receivers for use with other nuclei without signal degradation or corruption.
The recent surge in interest in the
in vivo NMR of nuclei such as
13C and
31P has led to an increased interest in developing array coils for these species. The low sensitivity of these nuclei makes the benefits of array receive coils attractive, but most array receivers are limited to use with
1H. Frequency translation with radiofrequency mixers is a promising method to adapt narrow-band
1H receivers to work with other nuclei. This technique is compatible with
1H decoupling, which is of great importance to
13C spectroscopy.
Materials and Methods
Frequency translation modifies the receive pathway of the system, as shown in figure 1. After preamplification, a radiofrequency (RF) mixer is used to convert the received signal to the 1H frequency. This translated signal is filtered and then fed into the system's 1H receiver.
In order to mix the received signal to the 1H frequency, it is necessary to generate a local oscillator (LO) at a frequency equal to the difference in frequencies of the two signals. 13C is the nucleus under study in figures 2 and 3; it is about 50 MHz at 4.7 T. To translate up to the 1H frequency, 200 MHz, a LO of 150 MHz is needed. In order to be able to acquire multiple signal averages, it is necessary that the LO phase be the same for each acquisition. This is accomplished by using a direct digital synthesis (DDS) IC (AD 9915) to generate the LO. The LO is reset by a system trigger to the same phase immediately before each acquisition, ensuring consistent phase.
To accommodate 1H decoupling, it is necessary to transmit and receive on 1H simultaneously. The high-power decoupling signal can easily bleed through and overwhelm the received signal, so it is advisable to mix to a frequency slightly removed from the 1H frequency. This is possible because MRI receiver bandwidths are typically at least 1 MHz, and the bandwidth of the decoupling signal is at most a few tens of kHz. By mixing to a frequency 100 kHz away from the decoupling signal, we can avoid interference while remaining within the receiver bandwidth.
We have constructed a 16 channel frequency translation system, which can be seen in figure 4. The top unit is the frequency translation unit, which contains the mixers and filters used for frequency conversion. Instead of traditional passive mixers and ferrite baluns, active mixers and fully differential op amps are used to eliminate magnetic components from the frequency translation unit. These components have a broad bandwidth and can operate at least up to 400 MHz, corresponding to 9.4 T.
The second unit from the top is the LO source, which uses an AD9915 DDS IC to generate the LO. This unit requires a 10 MHz reference from the system to ensure the LO is stable as well as a trigger to reset the phase before each acquisition. The LO is controlled by a Raspberry Pi in the bottom unit, which also contains the power supplies for the translation unit. Additionally, a Software Defined Radio (SDR) is included in this bottom unit for monitoring stability of the LO output, which can be viewed with the Raspberry Pi.
Results and Discussion
Performance of the frequency translation system has been verified on a 4.7 T 40 cm horizontal bore magnet with a Varian Unity/Inova spectrometer. Test on a 7T human scanner are ongoing.
Figure 2 shows a comparison of 1H decoupled 13C spectra acquired with an unmodified Varian system and a frequency translation receiver. The increase in SNR for the translated spectrum is attributed to improved sensitivity of the Varian receiver at the 1H frequency. The only artifact visible is a DC offset.
Figure 3 shows the combined magnitude spectra from a 3 element 13C planar array of 4 cm loops. Excitation was performed with a 13C birdcage, which was geometrically decoupled from the array. Because the Varian system only has one receive channel, spectra were acquired one at a time, but through different channels of the translation unit.
Conclusion
Frequency translation is a promising technique to adapt
1H receivers for use with other nuclei. The effectiveness of frequency translation has been shown with
13C. This technique has no negative effect on SNR and introduces no distortions into the spectrum. Additionally, the broadband nature of the devices used makes the system versatile with respect to nuclei and field strength.
Acknowledgements
The authors would like to acknowledge support of NIH
R21EB016394
and Cancer Prevention Institute of Texas (CPRIT) projects RP100625 and RP150456.References
Bottomly, P.A. Radiology, 1989, 170, 1-15
Roemer, P.B., Magn Reson Med, 1990, 16, 192-225
Wright, S.M., NMR Biomed., 1997, 10, 394-410
Lee, R.F, Magn Reson Med, 2000, 43, 269-277