Continuous SWIFT on a whole-body 7T system: Initial phantom and in vivo images
Florian Maier1, Manuela Rösler1,2, Armin M. Nagel1,3, and Reiner Umathum1

1German Cancer Research Center (DKFZ), Heidelberg, Germany, 2Institute for Biomedical Engineering, University of Zurich and ETH Zurich, Zurich, Switzerland, 3Diagnostic and Interventional Radiology, University Medical Center Ulm, Ulm, Germany

Synopsis

Simultaneous excitation and acquisition is a promising approach to acquire MR images using very low transmission power. This concept is also well-suited for imaging tissues and materials with ultrashort T2* relaxation times since no echo time exists. In this work, a cSWIFT hardware setup, pulse sequence, and reconstruction were designed for and implemented on a whole-body 7T system. Spectroscopy and imaging of phantoms and in vivo imaging of the human forearm were successfully performed. Feasibility of cSWIFT with the implemented setup was demonstrated.

Purpose

Simultaneous excitation and acquisition is a promising approach to acquire MR images using very low transmission power. Due to the instantaneous acquisition of the spin signal during excitation, no echo times exist (TE = 0). Therefore, this concept is also well-suited for imaging tissues and materials with ultrashort T2* relaxation times. Additionally, pulse sequences with low acoustic noise can be designed.

Two simultaneous excitation and acquisition methods were proposed recently: Continuous Swept Imaging with Fourier Transform1 (cSWIFT) and Concurrent Excitation and Acquisition2 (CEA).

In this work, a cSWIFT hardware setup, pulse sequence, and reconstruction were designed for and implemented on a whole-body 7T system. To demonstrate the feasibility of cSWIFT with the presented setup, spectroscopy and imaging of phantoms and in vivo imaging of the human forearm were performed.

Methods

All experiments were performed on a whole-body 7T MR system (Magnetom 7T, Siemens Healthcare, Erlangen, Germany).

Hardware. Fig. 1 shows the schematic of the setup. From the low-power TX output of the scanner system, a low-noise amplifier supplies about 65 mW to a small birdcage coil (∅shield = 140 mm; ∅coil = 100 mm; lcoil = 100 mm) via a carefully adjusted quadrature hybrid (QH). To minimize the unwanted leakage power, the coil input is coarsely adjusted by trimmer capacitors and finely adjusted by the reverse voltages of varactor diodes. A sample-and-hold circuit (S/H) keeps these voltages constant and isolates the coil during the sweeps. A high-dynamic-range, low-noise amplifier (NF = 0.6 dB; P-1dB = 100 mW; gain = 22 dB) amplifies the spin signal RX1, while RX2 represents leakage and RX3 enables registration of the exciting sweep. These RX-signals are processed by the scanner receiver. All RF hardware was built in-house.

Software. The implemented pulse sequence (Fig. 2) used a swept RF pulse with variable sweep span (for spectroscopy: 10 kHz; for imaging: 100 kHz) and duration (for spectroscopy: 100 ms; for imaging: 10 ms). TR was 10ms longer than the sweep to allow for slow gradient switching and an additional trigger signal to reload the S/H of the varactor diodes. To keep the excitation pulse bandwidth small, the pulse amplitude was ramped up and down slowly within 20% of the sweep time. Every second TR, a leakage measurement was performed using an additional gradient pulse (amplitude: 10.0 mT/m; direction: left-right) during the sweep to avoid resonant spins in the range of the swept pulse. For imaging, 16384 spokes were acquired in 11 min using a radial imaging scheme based on golden ratios3. During the acquisition of each spoke, an imaging gradient of 3.76 mT/m was applied, corresponding to a spherical FoV with a diameter of 50 cm. Images were reconstructed with 2×2×2 mm3 nominal resolution similar to a recently published algorithm1 using Python (Python Software Foundation).

Experiments. A spectrum was acquired using a phantom filled with Vodka (40 % ethanol C2H5OH). Imaging was performed on a latex phantom and a human forearm of a healthy volunteer.

Results

In the cSWIFT spectrum of Vodka (Fig. 3) the three resonances are clearly distinguishable at the expected positions. Signal was received from the latex structure (Fig. 4), albeit with an actual resolution that was lower than the nominal resolution. Fig. 5 shows in vivo images of a human forearm in which the ulna and radius are visible.

Discussion and Conclusion

A setup to perform simultaneous excitation and acquisition on a 7T whole-body system was implemented. The initial results show the feasibility of the concept. An in vivo image using the cSWIFT approach was acquired. However, image quality and resolution needs to be improved. In particular, a method to reconstruct images containing tissues with short T2* relaxation times (< 1 ms) in the presence of tissues with long relaxation times (> 1 ms) needs to be implemented. Due to the direct measurement of the leakage signal, this approach is robust against RF pulse inaccuracies.

In conclusion, the feasibility to acquire cSWIFT images on a 7T whole-body MR system was demonstrated.

Acknowledgements

The authors thank Dr. Richard R. Bouchard for proofreading and Barbara Dillenberger and Christian Kindtner for their support.

References

[1] Idiyatullin D et al., J. Magn. Reson. (2012) 220: 26-31.
[2] Özen AC et al., Magn. Reson. Mater. Phy. (in press, DOI: 10.1007/s10334-015-0497-0)
[3] Chan RW et al., Magn. Reson. Med. (2009) 61(2): 354-363.

Figures

Fig. 1: Schematic of setup: The birdcage coil is driven in quadrature to enable separation of exciting pulse from spin signal.

Fig. 2: Pulse sequence diagram. Swept RF pulses (orange) ramped up and down. Simultaneous acquisition (green). Spokes were acquired with imaging gradients in all directions. The leakage measurement was performed with a large gradient (10.0 mT/m) in RL direction. S/H reloading in gaps without active gradients or RF.

Fig. 3: Spectrum of Vodka (40% ethanol, C2H5OH). Three peaks at expected positions and with expected peak ratios (2:1:3).

Fig. 4: Sagittal, coronal, and axial slices of 3D image of latex phantom and photo of phantom. Thin latex layer wrapped around a hot-gun glue stick. Piece of natural rubber included off center.

Fig. 5: Sagittal, coronal, and axial slices of 3D dataset of human forearm. Ulna and radius visible in axial slice.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
3555