A modular 7T high-impedance array for ex-vivo imaging
Shin-Ichi Urayama1, Bei Zhang2, Koji Fujimoto3, Tomohisa Okada3, and Martijn A Cloos2

1Center for Educational Program in Graduate School, Kyoto University, Graduate School of Medicine, Kyoto, Japan, 2Radiology, New York University School of Medicine, New York, NY, United States, 3Human Brain Research Center, Kyoto University, Graduate School of Medicine, Kyoto, Japan


The high SNR provided by whole body 7T also provides enticing opportunities for ex-vivo imaging. However, such MRI systems are usually only equipped with a limited number of coils, rarely optimized for ex-vivo imaging. In this work, we leverage the new-found degrees of freedom provided by High Impedance Coil elements to create a versatile modular array-coil that can be re-configured in seconds to fit each sample optimally.


The high SNR provided by whole body 7T also provides enticing opportunities for ex-vivo imaging. However, such MRI systems are usually only equipped with a limited number of coils, rarely optimized for ex-vivo imaging. Moreover, even when limited to neuro-applications, ex-vivo samples come in many different sizes ranging from 1mL to more than 1L 1-4. When using traditional low impedance coil elements, one might need a variety of dedicated ex-vivo coils to optimally cover such a diverse range of samples. In particular, their mechanical rigidity and sensitivity to electro-dynamic interactions between elements limits their ability to adapt to the sample. In this work, we leverage the new-found degrees of freedom provided by High Impedance Coil (HIC) elements5 to create a versatile modular array-coil. In addition to their mechanical flexibility, HIC elements effectively cloak themselves from electrodynamic interactions with neighboring elements, which means they can be placed freely on any sample. By stitching the elements onto patches of Velcro, a versatile modular array was created that can be re-configured in seconds to fit each sample optimally.


Eight HIC elements resonant at 297MHz were built (Fig. 1). The coaxial structure (∅=3.5cm) was made out of micro-coax. Two pin diodes were used to detune each element during transmission. The interface board, including the matching circuit, was shielded using a 3D-printed housing covered with copper tape. The cable length was adjusted to achieve a “reversed” pre-amplifier decoupling condition5. Each element was stitched to a soft Velcro disk such that they can easily be re-arranged on a Velcro sheet (Fig 2) Five different HIC arrangements were tested on two different phantoms (Fig. 2); the larger of the two approximates a container used to hold a macaque brain (∅=6.5cm, length=14cm, σ=0.5S/m), the smaller one representative of a container for excised tissue samples such as a human spine segment (∅=2.5cm, length=12cm, σ=0.5S/m).

For transmission an 8-rung de-tunable birdcage resonant at 297 MHz was build (∅=15cm, length=15cm). To detune the transmit-coil during receive, PIN diodes were placed on the rungs. The two channels of the birdcage were driven in quadrature, each equipped with a transmit-receive switch and pre-amplifier interface.

Signal to Noise Ratio (SNR) measurements were performed using the Kellmann method6 and B1+ normalized based on the gradient echo signal equation7. Two spoiled gradient echo images were acquired, one with RF on and one without (noise measurement). Transmit sensitivity maps were obtained using a Turbo FLASH based B1+ mapping sequence8.

The coils elements were built at Kyoto University and tested in the 7T (Siemens MAGNETOM 7T, Erlangen, Germany) located at New York University.

Results & Discussion

The Velcro backing of the coil elements enables easy and quick re-arrangement of the array for use with samples of different size. Despite the use of non-critical overlap9, the noise correlation matrix remains relatively clean for all considered HIC arrangements (Fig. 3). However, some residual coupling is observed between elements 1 and 2 (7 and 8) in the “4x2L” (“8x1L”) arrangement. These may be attributed to a possible imperfection in the interface. In particular, we noticed that it is crucial to shield the inductors on the matching circuits as they may start to couple when placed in close proximity (on the small phantom the inductors are more orthogonal to each other).

In the center of the large phantom, the HIC array and birdcage provide a similar SNR. At the periphery, however, the SNR is considerably higher, as may be expected from a surface array9. In the center for the small phantom, a closefitting arrangement of 4 (or more) HIC elements can provide more than 10 times the SNR observed with the birdcage (Fig. 4). This boost in SNR can first and foremost be attributed to the close proximity of the elements, which, in this case, also happen to have a near optimal diameter to maximize the SNR in the center. It should therefore be noted, that a customized solenoid coil, for example, may be expected to provide a more comparable SNR4,10. However, such a rigid coil cannot easily adapt to samples10.

The dual rows configurations expand the field of view at the expense of azimuthal coverage. Due to the relatively small size of the HIC elements at 7T, additional elements may be needed to coverage large samples. Nevertheless, these initial results already show great promise in terms of SNR and versatility (Fig. 5).


The flexible HIC elements enable the creation of a versatile modular array that can adaptively accommodate ex-vivo samples of different size, thus providing a versatile platform for ex-vivo imaging using only a single coil-array while maintaining an optimal fit and high SNR.


This work was supported by SPIRITS 2018 of Kyoto University. This work was performed under the rubric of the Center for Advanced Imaging Innovation and Research (www.cai2r.net), a NIBIB Biomedical Technology Resource Center (NIH P41 EB017183). We thank Dr. Sheppard (M.D. Ph.D) and Mrs. Bruno (MSc) for their help scanning the ex-vivo sample. We thank Yoshihiko Kawabata (Takashima Seisakusyo) for their help in coil construction in Kyoto.


[1] J. Mollink, et al., Dentatorubrothalamic tract localization with postmortem MR diffusion tractography compared to histological 3D reconstruction. Brain Structure and Function, 2016;221:3487-3501.

[2] B. Strotmann, et al., Mapping of the internal structure of human habenula with ex vivo MRI at 7T. Frontier in Human Neuroscience, 2013;7:878.

[3] A.G. van der Kolk, et al., Imaging the Intracranial Atherosclerotic Vessel Wall Using 7T MRI: Initial Comparison with Histopathology. American Journal of Neuroradiology, 2015; 36:694-701.

[4] J. C. Augustinack, et al., Detection of Entorhinal Layer II Using Tesla Magnetic Resonance Imaging. Annals of Neurology, 2005;57:489:494.

[5] B. Zhang, et al., A high-impedance detector-array glove for magnetic resonance imaging of the hand. Nature Biomedical Engineering, 2018;2:570–577.

[6] Kellman, P. & McVeigh, E. R. Image reconstruction in SNR units: a general method for SNR measurement. Magnetic Resonance in Medicine, 2015;54:1439–1447.

[7] M.A. Bernstein et al., Handbook of MRI Pulse Sequences. ISBN-10: 0120928612.

[8] S. Chung, et al., Rapid B1+ mapping using a preconditioning RF pulse with TurboFLASH readout. Magnetic Resonance in Medicine, 2010; 64: 439-446.

[9] P. B. Roemer. et al. The NMR phased array. Magnetic Resonance in Medicine, 1990: 16; 192–225.

[10] J. Mispelter, et al., NMR Probeheads for Biophysical and Biomedical Experiments. ISBN-10: 1860946372.


Figure 1: The schematic design of the high-impedance element (A) and its physical implementation using two PIN diodes (MA4P4002B-402; Macom, Lowell, MA, USA) and microcoax (∅=1.1mm, ODU-USA, Inc.) (B). The birdcage used for transmission (∅=15cm, length =15cm), PIN diodes (UM9415B; Microsemi, Aliso Viejo, CA, USA) (C). The transmit-receive switch and pre-amplifier interface was procured from Stark Contrast ( Erlangen, Germany).

Figure 2: Velcro coil mounting system (A) and the different coil arrangements evaluated in this study (B).

Figure 3: SNR maps and Noise correlation matrixes. The dashed gray line indicates the location of the axial slice. GRE sequence parameters: nominal flip-angle = 10 TR = 500ms, 1x1mm 5mm slice, TE = 3ms, readout-bandwidth 500Hz, FOV= 128x128mm). B1 mapping sequence parameters: TR = 10sec, 1x1mm 5mm slice, FOV is 128x128mm.

Figure 4: Table with measured SNR values in different regions. When evaluating the multi-row arrangements, the imaging slice was centered between the two rows of non-overlapping elements. Hence some drop in SNR is to be expected. However, shielding effects from the interface boards should also be considered. In the current implementation, the shielded interface boards from the top row have substantial overlap with the coil elements on the bottom row. Miniaturization of the interface board could help mitigate such effects.

Figure 5: High resolution image of a human spinal cord sample (in plane resolution = 38μm, slice thickness = 1.5 mm, TE = 35ms,total scan time ~5min.

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