Whole-body imaging at 7T: Preliminary results

Authors


Abstract

The objective of this study was to investigate the feasibility of whole-body imaging at 7T. To achieve this objective, new technology and methods were developed. Radio frequency (RF) field distribution and specific absorption rate (SAR) were first explored through numerical modeling. A body coil was then designed and built. Multichannel transmit and receive coils were also developed and implemented. With this new technology in hand, an imaging survey of the “landscape” of the human body at 7T was conducted. Cardiac imaging at 7T appeared to be possible. The potential for breast imaging and spectroscopy was demonstrated. Preliminary results of the first human body imaging at 7T suggest both promise and directions for further development. Magn Reson Med 61:244–248, 2009. © 2008 Wiley-Liss, Inc.

Whole-body imaging and its clinical applications have been commercially developed for 3T. Body imaging has been demonstrated in research applications to 4T (1). Since the late 1990s, 7T has been used for human head imaging (2, 3). The major MRI system manufacturers are now supporting 7T whole-body MRI systems for head imaging only. Radio frequency (RF) field nonuniformities have been observed due to destructive interference patterns of the short radio wavelengths, (12 cm in brain tissue). These destructive interferences have presented significant challenges to image homogeneity, contrast uniformity, signal to noise ratio (SNR) contours, and specific absorption rate (SAR) distributions in the body at 300 MHz, the 7T 1H Larmor frequency. New methods such as transmit RF field (Bmath image) shimming are being developed to overcome these problems (4, 5). Human brain images with unprecedented SNR, spatial and temporal resolution, contrast, and spectral resolution are being acquired at 7T (6–11). The SAR is being minimized through optimization methods employing Bmath image shimming from multichannel transmit coils, thus facilitating safe and efficient RF power management as well. Many human body applications will benefit significantly from increased SNR and spectral resolution, if the body can indeed be imaged successfully at the highest field strengths available. This note reports the results of the first feasibility results of human body imaging and spectroscopy at 7T.

MATERIALS AND METHODS

Modeling

The RF field and SAR were modeled by the finite difference time domain (FDTD) method (Remcom, Inc., State College, PA, USA), for the human body–loaded coil at 7T. The body model used was derived from the National Library of Medicine (NLM) Visual Human digital atlas whose segmented anatomy was adjusted to match the conductivity and permittivity of tissues at 300 MHz. This body coil was driven at four ports (45°, 135°, 225°, and 315°) referenced to 0° at the bottom of the coil, as shown in Fig. 1a. The inner diameter of the coil measured 57.5 cm, the outer diameter was 62.5 cm, the length of the 32 resonant elements was 33 cm, and the length of the cylindrical cavity required to shield these elements from the gradients was 100 cm (Fig. 2). The magnitude of the magnetic component of the RF electromagnetic field was designated as “B1.”. The B1 field generated in the coil was circularly polarized and the magnitude was uniform in the unloaded state.

Figure 1.

7T TEM body and surface coils (14). a: Schematic diagram of the TEM body coil connected for four-port transmit and receive (1). b: Body coil mounted on the Siemens bore tube for insertion into the Siemens Sonata gradients. The transmission line “rungs” of the coil occupy the middle third (33 cm) of the 1-m coil, as shown. The cavity or shield component of the coil is slotted for gradient-induced eddy-current attenuation and TEM element segregation. c: Schematic of a section of the TEM surface coil. The capacitances CT, CM, and CD are for tuning, matching, and decoupling the independent elements, respectively. d: An eight-element TEM surface coil (32 cm × 15 cm) built of eight sets of two parallel copper strips separated by a 2-cm-thick polytetrafluoroethylene dielectric, with eight independent drive cables connected. [Color figure can be viewed in the online issue, which is available at http://www.interscience.wiley.com.]

Figure 2.

FDTD models of B1 and SAR contours generated by a body coil in the human body at 7T. a,b: Relative B1 magnitude isocontours, T/m (dB) in center-line transverse and sagittal planes. c,d: SAR, W/kg (dB) isocontours in the same planes, respectively.

Hardware Development

RF coils and drive circuits for human body imaging at 7T were not commercially available, and therefore had to be developed for this investigation. The MRI system used for the study was built around a Magnex 7T, 90-cm bore magnet (Magnex Scientific, Oxfordshire, UK) equipped with Magnex body gradients and a Symphony gradient amplifier (Siemens, Erlangen, Germany). Interfaced to this magnet was a Unity Inova spectrometer (Varian, Palo Alto, CA), together with a custom, 8-kW solid-state RF power amplifier (Communications Power Corporation [CPC], Hauppauge, NY, USA). The whole-body images in this report were acquired with an actively detuneable 300-MHz transverse electromagnetic (TEM) body coil, shown in Fig. 1b, built to the dimensions used in the computer modeling (1). Also developed was an eight-channel TEM surface coil, shown in Fig. 1c and d. For imaging the human trunk, these surface coils were used in pairs, located anterior and posterior to the region of interest in the body. The elements of these multichannel, transmit and receive coils were driven independently from a new 16-channel, parallel transceiver designed in-house and built by CPC (5, 12). This system enabled RF transmit phase and magnitude control of each of the coil array's 16 independent elements to facilitate Bmath image shimming and parallel imaging methods (1, 5, 7, 13–17). The term Bmath image refers to the transmit component of the RF magnetic field vector, in contrast to the RF receive field (Bmath image). A circularly-polarized, unilateral, RF breast coil consisting of two shielded crossed loops was also built and tested, and is presented below (in the “Breast Imaging and Spectroscopy With Quadrature Surface Coil” section; Fig. 5a) (18).

Figure 5.

Breast imaging and spectroscopy at 7T. a: Crossed-pair surface-coil transceiver used to acquire breast images (b), and proton spectra (c). b: A sagittal slice from a fat-suppressed, T1-weighted 3D FLASH image acquired in a normal subject. The box indicates voxel placement in the fibroglandular tissue. A single-voxel spectrum acquired from the voxel indicated in (b) appears in (c). Peaks from taurine and total choline are visible. [Color figure can be viewed in the online issue, which is available at http://www.interscience.wiley.com.]

Data Acquisition

Body Imaging with TEM Body Coil

The abdomen and thorax of a healthy normal, adult human were imaged with the body coil, in an institutional review board approved study at 7T. Due to the limited peak power available (approximately 4 kW at the coil) and the significant high-frequency load losses predicted, the initial experiments employed low flip angle sequences of greater pulse widths. For an initial mapping of the RF “landscape” in the body at 300 MHz, coronal, sagittal, and transaxial images were acquired with the body coil switched to transmit and receive mode, without the use of additional receiver coils. The receive signal was acquired by one of two methods: 1) analog, quadrature-phased recombination (as shown in Fig. 1a) or 2) sum-of-squares combination from four individually received ports. The common parameters used for acquiring the gradient echo, whole-body images of Figs. 3 and 4a were as follows: matrix size = 256 × 256, slice thickness = 3 mm, windowed three-lobed sinc RF pulse = 2 ms, and nominal flip angle = 25°. Specific to the coronal images: repetition time/echo time (TR/TE) = 60 ms/4 ms, field of view (FOV) = 50 × 35 cm, number of excitations (NEX) = 4, and scan time = 110 s. The sagittal images were acquired by: TR/TE = 50 ms/4 ms, FOV = 50 × 35 cm, NEX = 2, and scan time 55 s; and the transaxial images required: TR/TE = 50 ms/4 ms, FOV = 35 × 35 cm, NEX = 4, and scan time 110 s.

Figure 3.

Whole-body abdominal images at 7T. The coronal, sagittal, and transaxial, multislice image sets were acquired with body coil transmit and receive using low-tip, low-power, low-resolution gradient-echo sequences. No intensity correction was applied to the data.

Figure 4.

Whole-body thorax images at 7T. Using the same imaging parameters as in Fig. 3, the heart is visible in coronal, sagittal, and transaxial, multislice planes.

Body Imaging With Multichannel TEM Surface Coil

Employing multichannel transceiver coils closely fitted to the body offers an alternative to whole-body coils for imaging localized regions of interest. Each element in the transceiver array was driven by a dedicated RF power amplifier with independent phase and magnitude control, a transmit/receive switch, and a preamplifier for reception. RF power was monitored for each channel. In the cardiac imaging example shown in Fig. 4b and c, low flip angle, breathheld, electrocardiogram retrogated, gradient echo images were acquired using the parameters: TR/TE = 30 ms/3.0 ms; matrix size = 144 × 192, resolution (in Fig. 4b) = 2.0 × 2.0 × 5.0 mm, resolution (in Fig. 4c) = 1.7 × 1.7 × 5.0 mm. These 7T cardiac images were acquired with an Avanto body gradient interfaced to a Siemens spectrometer, a later upgrade to the 7T system described. To mitigate RF artifacts and to maximize excitation efficiency (reducing SAR), Bmath image shimming was applied. The relative transmit Bmath image phases of the coil elements were calculated and established by acquiring phase maps (4, 19) to determine a set of transmit phases required to maximize the Bmath image phase coherence within a given region of interest.

Breast Imaging and Spectroscopy With Quadrature Surface Coil

MRI and MRS data were acquired from a healthy normal female volunteer using the transmit/receive surface coil of Fig. 5a. To distinguish fibroglandular from adipose tissue, T1-weighted images were acquired using a fat-suppressed three-dimensional (3D) gradient-recalled echo sequence. Acquisition parameters were: TR/TE = 15 ms/5 ms, FOV = 14 cm × 14 cm × 14 cm, and matrix size = 256 × 256 × 64. Single-voxel spectra were collected using the Localization by Adiabatic SElective Refocusing (LASER) pulse sequence (20) and echo time averaging. To quantify fibroglandular choline-containing compounds, the unsuppressed water signal in the voxel was used as an internal reference. The LASER single-voxel spectroscopy was preformed with voxel size = 1.6 ml, TR = 3 s, TE = 43–195 ms in 128 increments, and was processed with 8-Hz line broadening.

RESULTS AND DISCUSSION

7T Models

Results from numerical calculations of the B1 and SAR contours in the NLM human male at 300 MHz are shown in Fig. 2. Figures 2a and b show the calculated B1 distribution in a loaded body coil for this frequency. Interference patterns are shown to create the nonuniformities in the B1 field within and outside of the body at 7T. In the body, these nonuniformities span 40 dB, with the lowest level B1 regions running longitudinally in the body center, and the highest B1 magnitudes in the periphery of the body near the four driven elements. Interestingly, the head, located well outside of the coil elements, experiences some of the highest B1 values. The central blue-colored lines of RF destructive interference in the Figure 2a and b models are observed as dark lines in the body images of Fig. 3. Figure 2c and d show the calculated SAR distribution in the body excited by the 300-MHz body coil described. The SAR contours span a 50-dB gradient, with maximum SAR experienced in the tissues nearest the drive elements of the coil, and the lowest SAR values found near the center-line of the body. The severe B1 and SAR gradients modeled suggest that conventional MRI methods employing a homogeneous, circularly polarized body coil for uniform excitation of the body may not be possible for whole-body, 1H imaging at 7T. Bmath image shimming with a body coil may, however, offer a solution.

7T MRI With Body Coil

The results of the 3D electromagnetic model of a human in a 7T body coil predicted significant RF artifacts running longitudinally through the center of the body, due to the destructive interference of the short (12 cm) wavelengths at 300 MHz, from the high dielectric constants of high water-content tissues. Initial images appear to follow these predictions. Central dark bands seen in the abdominal images of Fig. 3 could not be excited within U.S. Food and Drug Administration (FDA) SAR guidelines. The RF power (SAR) used to acquire the images presented was 11.1 W per slice. For the five interleaved slices, the total power used to acquire these first data sets was 55.5 W, <1 W/kg for the 80-kg subject imaged and well within the FDA guidelines for human torso imaging. Initial images of the human thorax in Fig. 4, however, show some promise for cardiac imaging. The longer wavelengths in the low-density lung tissue of the chest may improve the Bmath image penetration to the pericardium. It was found that receiving the signal from the four independent channels of the body coil, and constructing images by simple sum-of-squares magnitude addition, resulted in images with 20% higher SNR than images acquired by circularly phased analog recombination of the received signal.

7T Body Imaging With Multichannel TEM Surface Coil

Not all “body” imaging is best served with a body coil. Cardiac imaging and spectroscopy can benefit greatly from the inherent temporal, spatial, and spectral resolution at 7T. Compared to the body coil results in Fig. 4a, significant improvements in heart image SNR and Bmath image uniformity were achieved by using the multichannel transceiver of Fig. 1d, together with Bmath image shimming. This improvement is evident in the four-chamber (Fig. 4b) and short axis (Fig. 4c) views acquired with the multichannel surface coil. The local SAR required for these improved surface coil images was approximately the same, 1 W/kg, used to acquire the body images. Use of local, multichannel, transceiver coils with Bmath image shimming has also proven successful in 7T prostate imaging (19).

7T MRI and MRS with Local Transmit and Receive Coil

For local imaging and spectroscopy of more superficial anatomy such as the breasts, conventional transceiver surface coils can be used effectively. Early results from a 7T breast cancer study give examples of the promise of 7T diagnostics (18). Measurement of total choline-containing compounds (tCho) by localized 1H MRS offers a means to distinguish malignant from benign lesions and to predict response to neoadjuvant chemotherapy. The detection of tCho in small lesions having low cellularity is usually limited by insufficient SNR at clinical field strengths. Previous breast MRS studies at 4T have reported higher SNR, enabling the detection of tCho in smaller lesions as well as the occasional detection of other metabolites such as taurine, creatine, and glycine (21). The ability to detect relatively narrow tCho and taurine resonances in normal breasts, shown in Fig. 5b and c, provides motivation to investigate breast cancer at 7T.

CONCLUSIONS

First examples of theoretical models, technology, and methods for investigating the feasibility of 7T body imaging have been demonstrated. Technology developments include a 300-MHz body coil, a multichannel surface coil, and a small, quadrature surface coil. Preliminary imaging results show whole-body imaging at 7T with a body coil, and with locally placed multichannel transceiver coils and Bmath image shimming. Models and measurements indicate that the conventional, uniform, circularly polarized body coil may have limitations for homogeneous excitation at 7T. Locally placed, multichannel transceivers could provide a solution to body-coil shortcomings at 7T. Improvements in both signal intensity and homogeneity appear to be gained when using multichannel transceiver coils with Bmath image shimming for imaging local regions of interest. Local, single-channel, or quadrature surface coils may also be useful, especially for superficial regions of interest. Cardiac and breast imaging and spectroscopy appear to be early applications worth pursing at 7T. With the proper selection of coils and methods such as Bmath image shimming, general body imaging appears to be feasible at 7T.

Ancillary