Towards low power EPR Imaging using Frank poly-phase pulse sequence
Nallathamby Devasahayam1, Randall H. Pursley2, Thomas J. Pohida2, Shingo Matsumoto3, Keita Saito4, Sankaran Subramanian5, and Murali C. Krishna4

1Radiation Biology Branch, National Cancer Institute, Bethesda, MD, United States, 2Center for Information Technology, Bethesda, MD, United States, 3Graduate School of Information Science and Technology, Sapporo, Japan, 4National Cancer Institute, Bethesda, MD, United States, 5Indian Institute of Technology Madras, Chennai, India

Synopsis

Electron Paramagnetic Resonance (EPR) imaging is suited well for small animal physiological imaging with its unique capability of generating in vivo quantitative oxygen maps. The main bottleneck in scaling up pulsed EPR imaging to human anatomy is that the required RF power of US federal food and drug administration (FDA), specific absorption rate (SAR) limits. In Frank Sequence we are using power levels on the order of 250 microwatts in a crossed coil resonator with ~35 dB isolation. Using a 256 pulse polyphase Frank Sequence, it was possible to obtain images with good SNR.

Purpose

Considerable progress is being made in combining EPR-based quantitative in vivo oximetry in small animal cancer research with co-registered hyperpolarized MR spectroscopic imaging in our understanding of tumor physiology and the action cancer drugs (1). Frank pulse sequence is being developed to reduce the power required to do EPR animal imaging and as well as the feasibility of scaling up RF EPR imaging of larger objects.

Methods

The schematic of the low power spectrometer is shown in Fig. 1. The signal source for the system is a Tektronix 7121B Arbitrary Waveform Generator (AWG). The maximum clock rate of 12 GB/s is used to generate the sequences of RF pulses of 300 MHz with a 50 ns pulse width. In order to be able to rapidly pulse the system with low energy pulses and to record the response in between pulses it is important to have a large isolation (at least 40 dB) between the transmit coil and the receive coil. A saddle coil of 20 mm diameter and a length of 30 mm is used as the Excitation coil. The receive coil a surface coil of 10 mm Diameter and is inserted in parallel to the B1 field of Saddle coil and is shown in Fig. 2a. The surface coil is inserted through a slit in the Lucite former in the middle of the saddle coil. Though the surface coil is inserted vertically, it may not be very much vertical in respect to the point of isolation. To tune the position of the surface coil to achieve the higher order of isolation a position tuning mechanism is introduced. The tuning assembly drawing is given in Fig 2b. Since the surface coil is inside the slit, the position can be changed by tuning the assembly. By doing tuning, the isolation achieved is 38dB. The total assembly is shown in Fig. 2C A Frank sequence, by definition, must contain N2 elements. A length of 256 elements was chosen for the sequence. The sequences are generated using a National Instruments LabVIEW software program as an array of phases. The RF pulses sequences are also generated using LabVIEW. For each phase represented in the Frank sequence (2), a 300 MHz RF pulse with that phase is generated by LabVIEW and stored. Each RF pulse is 50 ns in width with an additional 10 ns of zero-padding for a total of 60ns or 720 sampled points. The LabVIEW program then constructs the entire pulse sequence based on the phase designated for each element of the sequences. The final pulse sequence waveforms consist of 184,320 total points each. These are stored in a format that the AWG can import and implement. One of the available AWG digital signals is used to generate a trigger at the beginning of each pulse sequence. LabVIEW is also utilized to acquire the EPR signals from the Signatec PX14400 data acquisition board. The Signatec board is configured to acquire at a clock rate of 400 MHz and to start acquisition upon receiving a trigger signal from the AWG. It will then acquire a complete pulse sequence of 184,320 points. These sequences are averaged and then passed on to perform band-pass sampling. That extracts the baseband real and imaginary waveform data resulting in a complex waveform. This waveform is then broken up into 60 ns segments, each representing a 50 ns RF pulse and the following 10 ns of time until the next pulse. The data is now represented as a 2D array where each row represents an RF pulse and each column is a coincident time within each RF pulse. A column of this data that represents somewhere in the 10 ns zero-padded region is then correlated with the Frank sequence (3) that the original data was encoded with. This results in a reconstructed FID signal and stored as files.

Results

In order to test the feasibility of using low power Frank pulse sequence for EPR imaging a phantom consisting of a Lucite cylinder that fits snugly in the receiver coil with four holes of different diameters filled with TCNQ was imaged. Fig. 3a, Fig. 3b,c & d, summarizes the results which shows that images with good SNR can be obtained using a cross coil resonator with >35 dB isolation at power levels of 0.25 mW. Further progress by way of designing larger crossed resonators with larger volume will enable extending in vivo EPR imaging and oximetry to larger subjects

Acknowledgements

We thank Mr. Frank Harrington of Radiation Oncology Branch for making all the mechanical assemblies.

References

1. S. Matsumoto et al. Low-field paramagnetic resonance imaging of tumor oxygenation and glycolytic activity in mice, Journal of Clinical Investigation 118, 1965-1973 (2008)

2. B. Blumich B et al. NMR with excitation modulated by Frank sequences, Journal of Magnetic Resonance, 199,18-24 (2009)

3. Mark Tseitlin et al. Use of the Frank sequence in pulsed EPR, Journal of Magnetic Resonance, 209, 306-309 (2011)

Figures

Fig.1 Schematics of the low power spectrometer

Fig. 2A Cross coil assembly schematic 2B. The cross coil tuning assembly . 2C The actual resonator assembly.

Fig. 3A Sketch of the TCNQ phantom used to test the Frank sequence. Fig. 3B 2D image using Frank sequence at 0.25 mW RF power. Fig. 3C, 2D image obtained using conventional single pulse . All images are obtained using phase encoding of a single time point at 600 ns delay from the excitation start. 10000 FIDs were summed. The orientation of the images have been rotated to match with the phantom orientation. The fourth spot with a diameter of 1 mm did not have enough TCNQ to give a signal.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
2378