Design, evaluation and application of a 16-channel Frequency Synthesizer Module for Thermal Magnetic Resonance
Haopeng Han1, Shuailin Wang2, Thomas Wilhelm Eigentler1, Lukas Winter3, Eckhard Grass4,5, and Thoralf Niendorf1,6,7

1Berlin Ultrahigh Field Facility (B.U.F.F.), Max Delbrück Center for Molecular Medicine in the Helmholtz Association, Berlin, Germany, 2Beijing Deepvision Technology Co., Ltd., Beijing, China, 3Physikalische Technische Bundesanstalt (PTB), Berlin, Germany, 4IHP – Leibniz-Institut für innovative Mikroelektronik, Frankfurt (Oder), Germany, 5Institute of Computer Science, Humboldt-Universität zu Berlin, Berlin, Germany, 6Experimental and Clinical Research Center (ECRC), a joint cooperation between the Charité Medical Faculty and the Max Delbrück Center for Molecular Medicine, Berlin, Germany, 7MRI.TOOLS GmbH, Berlin, Germany


Thermal Magnetic Resonance makes use of the physics of the radio frequency fields applied at ultrahigh field magnetic resonance imaging. While UHF-MRI enables measuring temperature in vivo, highly localized power deposition can be achieved by interfering RF waveforms with a shortened wavelength. This constitutes a means for supervised in vivo temperature modulation. The number of RF signals and the signals’ properties greatly affect the heating performance. In this work, a 16-channel frequency synthesizer module was developed as an RF signal source for Thermal MR. Preliminary experiments were conducted to demonstrate that the proposed module is suitable for Thermal MR.


Thermal Magnetic Resonance (Thermal MR) has unique potential to provide focal temperature manipulation, high resolution diagnostic imaging and MR thermometry1 in an integrated device. Electromagnetic field simulations showed that we can adapt an ultrahigh field MR device to generate heat in highly focused regions of tissue by using high-density radio frequency transmitter arrays2. Previous experimental work on thermal intervention operated at a fixed frequency3 and has a limited number of channels4. While lower RF frequencies focus energy to larger regions and have lower RF energy losses in and outside tissue5, EMF simulations showed that the power absorption in the target as compared to regions outside the target can be improved by applying higher RF frequencies and a higher number of transmit channels6,7. It is therefore beneficial to investigate RF heating over a wider frequency range4. For this reason this work proposes a 16-channel modular frequency synthesizer with high phase and frequency tuning resolution that operates between 0.06~3.5GHz and that can be used as the signal source for Thermal MR.


The hardware is an AXIe-18 (Advanced Telecommunications Computing Architecture (ATCA) Extensions for Instrumentation and Test) compliant modular design. Figure 1 shows its block diagram. It was designed around the phase-locked loop chip ADF4356 (Analog Devices, MA, USA). This PLL chip generates an output frequency in a range of 53.125~6800MHz. It provides a very fine frequency resolution with practically no residual frequency error due to its 52-bit modulus. The output phase can be adjusted with a theoretical resolution of 360°/224. There are 16 PLL chips on the module that lock to the same reference signal. A low jitter 2-input selectable 1:16 clock buffer CDCLVP1216 (Texas Instruments, TX, USA) is used to fan out the reference signal to 16 PLL chips. Filters are added to filter out the harmonics. RF switch chips HMC245A (DC~3.5GHz, Analog Devices, MA, USA) are used to select different filter options. The whole system is managed by a field programmable gate array chip ZU3EG (Xilinx, CA, USA) which is the core of a system-on-module module AES-ZU3EG-1-SOM-I-G (Avnet, AZ, USA).

A 4-hour continuous testing was conducted for 500MHz, 1GHz, and 1.5GHz with a spectrum analyzer (ZVL, R&S, Munich, Germany) to evaluate the frequency drift. Various phase settings with 4 channels running at 300MHz were tested using an oscilloscope (MSOS054A, Keysight, CA, USA). Heating experiments were conducted at 300MHz, 400MHz and 500MHz. The output RF signal (Pout=-4dBm) was fed to a power amplifier. After amplification, a ~50dBm signal was fed to an ultra-wide-band antenna. The antenna was applied to a phantom placed into the isocenter of a 7.0T MR scanner (Magnetom, Siemens Healthineers, Erlangen, Germany). The heating paradigm was applied for 2 minutes for each frequency. MR thermometry using the PRFS approach (TR=102ms, TE1=2.26ms, TE2=6.34ms, Voxel size=1.5x1.5x4mm³) at 298 MHz was conducted before and after the heating for each frequency. Fiber optic temperature sensors (Neoptix, Québec, Canada) were used to validate the MR thermometry results.


Figure 2 shows the hardware module operating inside an AXIe chassis. The tested three frequency points are shown in Figure 3. No frequency error was observed. There was no measurable frequency drift during the 4-hour continuous testing. Four 300MHz signals with different phase settings are highlighted in Figure 4. The measured phases comply with the programmed phase settings in the PLLs with an average error of 0.3306°. The heating setup and MR thermometry results are displayed in Figure 5. The applied heating paradigm (Pin=~50dBm, t=120s) induced a temperature change of 15°C in an area underneath the antenna. The MR thermometry results agree with the measurements obtained from fiber optic sensors.

Discussion and Conclusion

This work demonstrates that the proposed 16-channel modular frequency synthesizer is suitable for Thermal MR. The AXIe modular frequency synthesizer has many advantages over a conventional RF signal source. The benefits include its large module size which enables hosting a large number of channels on one module, its modularity which makes it possible for easy up scaling to a larger number of RF channels (e.g. 80 channels with 5 modules inside one chassis). The availability of commercial-off-the-shelf chassis saves the effort of designing the power supply and cooling system.

The wide frequency range and ultra-fine frequency and phase resolution provide high flexibility in the construction of the heating pattern. The high quality RF signals generated by the module provide a fundamental basis to RF heating and suit the needs of Thermal MR applications.


This project was funded in part by an advanced ERC grant (EU project ThermalMR: 743077).


1. Winter, L., et al., Design and evaluation of a hybrid radiofrequency applicator for magnetic resonance imaging and RF induced hyperthermia: electromagnetic field simulations up to 14.0 Tesla and proof-of-concept at 7.0 Tesla. PLoS One. 2013 Apr 22;8(4):e61661. doi: 10.1371/journal.pone.0061661. Print 2013.

2. Winter, L., et al., Thermal magnetic resonance: physics considerations and electromagnetic field simulations up to 23.5 Tesla (1GHz). Radiat Oncol. 2015 Sep 22;10:201. doi: 10.1186/s13014-015-0510-9.

3. Bakker, J.F., et al., Design and test of a 434 MHz multi-channel amplifier system for targeted hyperthermia applicators, Int J Hyperthermia, DOI: 10.3109/02656730903341191.

4. Trefná, H.D., et al., Design of a wideband multi-channel system for time reversal hyperthermia, Int J Hyperthermia, DOI: 10.3109/02656736.2011.641655.

5. Wust, P., et al., Hyperthermia in combined treatment of cancer. The Lancet Oncology 2002;3(8):487-497.

6. Winter, L. & Niendorf, T., Electrodynamics and radiofrequency antenna concepts for human magnetic resonance at 23.5 T (1 GHz) and beyond. Magn Reson Mater Phy (2016) 29: 641. https://doi.org/10.1007/s10334-016-0559-y.

7. Guerin, B., et al., Computation of ultimate SAR amplification factors for radiofrequency hyperthermia in non-uniform body models: impact of frequency and tumour location, Int J Hyperthermia, DOI: 10.1080/02656736.2017.1319077.

8. AXIe Consortium, AXIe Specifications, http://axiestandard.org/, Accessed October 20, 2018.


Figure 1: System block diagram. Various connections to the backplane were implemented according to the AXIe standard. An intelligent platform management controller (IPMC) module can be populated on the board to communicate with the chassis controller. The output of the PLL is square wave when the output dividers are used. There are 3 filtering options for each channel: a 400MHz low-pass filter, a 1200MHz low-pass filter and a direct connection. The reference signal comes either internally from an ultra-low jitter programmable oscillator or externally from a high performance RF source. All the components are controlled by the FPGA.

Figure 2: Pictures of the hardware module. left) 16-channel AXIe Frequency Synthesizer module. There are 16 RF output SMB connectors, a pair of reference clock input SMA connectors, an RJ45 Ethernet port, a micro USB port for serial connection, a micro SD card connector and a SMA connector for trigger signal on the front panel. Ethernet, PCIe and LVDS signals are connected through high density connectors to the backplane. right) The 16-channel AXIe Frequency Synthesizer module working inside a 5-slot AXIe chassis (M9505A, Keysight, CA, USA).

Figure 3: Example frequency outputs. The center frequency of the spectrum analyzer is set to 1GHz with a span of 2GHz. Three frequencies: 500MHz, 1GHz and 1.5GHz are tested. The reference signal for PLLs comes internally from the on-board oscillator. The power level for each frequency is set to 5dBm. The observed power losses (2.05dBm for 500MHz, 2.14dBm for 1GHz and 3.64dBm for 1.5GHz) are due to the value of the RF chokes, the value of the AC coupling capacitors and the insertion loss of the RF components (switches & filters).

Figure 4: Four 300MHz signals with different phase settings. left) the initial phase measurements. right) The phase measurements after the phase shift. Channel 1 was used as the reference signal. After initialization, the four channels were locked to the same clock signal. The initial phase differences were induced by the output skew of the clock fan-out chip and the differences of the clock signal length. Then phase shift commands were issued to shift the phase of Ch2, Ch3 & Ch4 for 100°, 200° & 300° respectively. An average error of 0.3306° was measured with a maximum error of 0.5172°.

Figure 5: The heating experiment setup and MR thermometry. top) Two identical RF signals were generated. One signal was connected to an oscilloscope for monitoring. The other signal was connected to a power amplifier to drive the ultra-wide-band antenna. The experimental setup is shown in the middle figure. The antenna was placed on top of an agarose phantom. Three fiber optic sensors were inserted into the middle of the phantom. bottom) MR thermometry results and thermometer readings obtained for heating using three frequencies. The results show the temperature increase in the phantom for a transversal slice underneath the antenna.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)