Kyle M. Gilbert

Design of field-cycled magnetic resonance systems for small animal imaging

This paper presents a design study for a field-cycled magnetic resonance imaging (MRI) system directed at small animal imaging applications. A field-cycled MRI system is different from a conventional MRI system in that it uses two separate and dynamically controllable magnetic fields. A strong magnetic field is used to polarize the object, and a relatively weak magnetic field is used during signal acquisition. The potential benefits of field-cycled MRI are described. The theoretical dependences of field-cycled MRI performance on system design are introduced and investigated. Electromagnetic, mechanical and thermal performances of the system were considered in this design study. A system design for imaging 10 cm diameter objects is presented as an example, capable of producing high-duty-cycle polarizing magnetic fields of 0.5 T and readout magnetic fields corresponding to a proton Larmor frequency of 5 MHz. The specifications of the final design are presented along with its expected electromagnetic and thermal performance.

Simulation of scattering and attenuation of 511 keV photons in a combined PET/field-cycled MRI system

Mixing the imaging modalities of positron emission tomography (PET) and magnetic resonance imaging (MRI) will offer the best soft tissue contrast (MRI) with information about metabolic function (PET). The high magnetic field environment of an MRI system makes the detection of annihilation photons difficult, as the response of standard photo-multiplier tubes is compromised. An approach using field-cycled MRI is discussed here, as field-cycled MRI makes it possible to have long periods of time available for nuclear imaging when there is no magnetic field present. This work focuses upon the effect of the field-cycled MRI upon the nuclear image due to the added material providing additional attenuation of the PET signal, and additional nuclei for scatter. These effects are studied using a Monte Carlo simulation based upon the GEANT libraries. Attenuation effects are shown to be significant, approximately 6% for the RF shield and coil and approximately 24% for the gradients. No significant effect is seen in image quality due to the scattering of the gammas. With these levels of attenuation it is concluded that open gradient coils and shim coils are required around the imaging volume.

Slice-by-slice B1+ shimming at 7 T.

Parallel transmission has been used to reduce the inevitable inhomogeneous radiofrequency fields produced in human high‐field MRI greater than 3 T. Further improvements in the transmit homogeneity and efficiency are possible by leveraging the additional degree of freedom permitted by multislice acquisitions. Compared to simple scaling of the flip angle to compensate for B1+ falloff along the radiofrequency coil, calculation of B1+ shim solutions on a slice‐by‐slice basis can markedly improve homogeneity and/or reduce transmitted power and global SAR. Performance measures were acquired at 7 T with a 15‐channel head‐only transceive array featuring elements distributed over all three logical axes, facilitating B1+ shimming over arbitrary orientations. Compared to a circularly polarized volume mode of the same coil, shimming to maximize excitation efficiency on a slice‐by‐slice basis yielded improvements in mean B1+ by 12.8 ± 2.4% and a reduction in standard deviation of B1+ of 16.3 ± 6.8%, while reducing relative SAR by 6.2 ± 3.1%. When shimming for greater uniformity, the mean and standard deviation of B1+ were further improved by 15.9 ± 2.6% and 26.2 ± 10.4%, respectively, at the expense of a 135 ± 8% increase in global SAR. Robust multislice‐shim solutions are demonstrated that can be quickly calculated, applied in real time, and reliably improve on volume coil modes. Magn Reson Med, 2012.

A radiofrequency coil to facilitate B shimming and parallel imaging acceleration in three dimensions at 7 T

A 15‐channel transmit–receive (transceive) radiofrequency (RF) coil was developed to image the human brain at 7 T. A hybrid decoupling scheme was implemented that used both capacitive decoupling and the partial geometric overlapping of adjacent coil elements. The decoupling scheme allowed coil elements to be arrayed along all three Cartesian axes; this facilitated shimming of the transmit field, B  1+ , and parallel imaging acceleration along the longitudinal direction in addition to the standard transverse directions. Each channel was independently controlled during imaging using a 16‐channel console and a 16 × 1‐kW RF amplifier–matrix. The mean isolation between all combinations of coil elements was 18 ± 7 dB. After B  1+ shimming, the standard deviation of the transmit field uniformity was 11% in an axial plane and 32% over the entire brain superior to the mid‐cerebellum. Transmit uniformity was sufficient to acquire fast spin echo images of this region of the brain with a single B  1+ shim solution. Signal‐to‐noise ratio (SNR) maps showed higher SNR in the periphery vs center of the brain, and higher SNR in the occipital and temporal lobes vs the frontal lobe. Parallel imaging acceleration in a rostral–caudal oblique plane was demonstrated. The implication of the number of channels in a transmit–receive coil was discussed: it was determined that improvements in SNR and B  1+ shimming can be expected when using more than 15 independently controlled transmit–receive channels. Copyright

A conformal transceive array for 7 T neuroimaging

The first 16‐channel transceive surface‐coil array that conforms to the human head and operates at 298 MHz (7 T) is described. Individual coil elements were decoupled using circumferential shields around each element that extended orthogonally from the former. This decoupling method allowed elements to be constructed with arbitrary shape, size, and location to create a three‐dimensional array. Radiofrequency shimming achieved a transmit‐field uniformity of 20% over the whole brain and 14% over a single axial slice. During radiofrequency transmission, coil elements couple tightly to the head and reduce the amount of power necessary to achieve a mean 90° flip angle (660‐μs and 480‐μs pulse lengths were required for a 1‐kW hard pulse when shimming over the whole brain and a single axial slice, respectively). During reception, the close proximity of coil elements to the head increases the signal‐to‐noise ratio in the periphery of the brain, most notably at the superior aspect of the head. The sensitivity profile of each element is localized beneath the respective shield. When combined with the achieved isolation between elements, this results in the capacity for low geometry factors during both transmit and receive: 1.04/1.06 (mean) and 1.25/1.54 (maximum) for 3‐by‐3 acceleration in the axial/sagittal plane. High cortical signal‐to‐noise ratio and parallel imaging performance make the conformal coil ideal for the study of high temporal and/or spatial cortical architecture and function. Magn Reson Med, 2012.

Transmit/receive radiofrequency coil with individually shielded elements.

A novel method for decoupling coil elements of transmit/receive (transceive) arrays is reported. Each element of a coil array is shielded both concentrically and radially to reduce the magnetic flux linkage between neighboring coils; this substantially reduces the mutual inductance between coil elements and allows them to behave independently. A six‐channel transceive coil was developed using this decoupling scheme and compared with two conventional decoupling schemes: the partial overlapping of adjacent elements and capacitive decoupling. The radiofrequency coils were designed to image the human head and were tested on a 7‐T Varian scanner. The decoupling, transmit uniformity, transmit efficiency, signal‐to‐noise ratio, and geometry factors were compared between coils. The individually shielded coil achieved higher minimum isolation between elements (2.7–4.0 dB) and lower geometry factors (2–14%) than the overlapped and capacitively decoupled coils, while showing a reduction in transmit efficiency (2.8–5.9 dB) and signal‐to‐noise ratio (up to 34%). No difference was found in the power absorbed by the sample during a 90° radiofrequency pulse. The inset distance of coil elements within their shields was then reduced, resulting in significant improvement of the transmit efficiency (1.3 dB) and signal‐to‐noise ratio (28%). The greatest asset of this decoupling method lies in its versatility: transceive coils can be created with elements of arbitrary shape, size, location, and resonant frequency to produce three‐dimensional conformal arrays. Magn Reson Med, 2010.

Evaluation of a positron emission tomography (PET)-compatible field-cycled MRI (FCMRI) scanner

Field‐cycled MRI (FCMRI) uses two independent, actively controlled resistive magnets to polarize a sample and to provide the magnetic field environment during data acquisition. This separation of tasks allows for novel forms of contrast, reduction of susceptibility artifacts, and a versatility in design that facilitates the integration of a second imaging modality. A 0.3T/4‐MHz FCMRI scanner was constructed with a 9‐cm‐wide opening through the side for the inclusion of a photomultiplier‐tube–based positron emission tomography (PET) system. The performance of the FCMRI scanner was evaluated prior to integrating PET detectors. Quantitative measurements of the systems signal, phase, and temperature were recorded. The polarizing and readout magnets could be operated continuously at 100 A without risk of damage to the system. Transient instabilities in the readout magnet, caused by the pulsing of the polarizing magnet, dissipated in 50 ms; this resulted in a steady‐state homogeneity of 32 Hz over a 7‐cm‐diameter volume. The short‐ and long‐term phase behaviors of the readout field were sufficiently stable to prevent visible readout or phase‐encode artifacts during imaging. Preliminary MR images demonstrated the potential of the FCMRI scanner and the efficacy of integrating a PET system. Magn Reson Med, 2009.

First image from a combined positron emission tomography and field-cycled MRI system

Combining positron emission tomography and MRI modalities typically requires using either conventional MRI with a MR‐compatible positron emission tomography system or a modified MR system with conventional positron emission tomography. A feature of field‐cycled MRI is that all magnetic fields can be turned off rapidly, enabling the use of conventional positron emission tomography detectors based on photomultiplier tubes. In this demonstration, two photomultiplier tube‐based positron emission tomography detectors were integrated with a field‐cycled MRI system (0.3 T/4 MHz) by placing them into a 9‐cm axial gap. A positron emission tomography‐MRI phantom consisting of a triangular arrangement of positron‐emitting point sources embedded in an onion was imaged in a repeating interleaved sequence of ∼1 sec MRI then 1 sec positron emission tomography. The first multimodality images from the combined positron emission tomography and field‐cycled MRI system show no additional artifacts due to interaction between the systems and demonstrate the potential of this approach to combining positron emission tomography and MRI. Magn Reson Med, 2011.

Optimized parallel transmit and receive radiofrequency coil for ultrahigh-field MRI of monkeys

Monkeys are a valuable model for investigating the structure and function of the brain. To attain the requisite resolution to resolve fine anatomical detail and map localized brain activation requires radiofrequency (RF) coils that produce high signal-to-noise ratios (SNRs) both spatially (image SNR) and temporally. Increasing the strength of the static magnetic field is an effective method to improve SNR, yet this comes with commensurate challenges in RF coil design. First, at ultrahigh field strengths, the magnetic field produced by a surface coil in a dielectric medium is asymmetric. In neuroimaging of rhesus macaques, this complex field pattern is compounded by the heterogeneous structure of the head. The confluence of these effects results in a non-uniform flip angle, but more markedly, a suboptimal circularly polarized mode with reduced transmit efficiency. Secondly, susceptibility-induced geometric distortions are exacerbated when performing echo-planar imaging (EPI), which is a standard technique in functional studies. This requires receive coils capable of parallel imaging with low noise amplification during image reconstruction. To address these challenges at 7T, this study presents a parallel (8-channel) transmit coil developed for monkey imaging, along with a highly parallel (24-channel) receive coil. RF shimming with the parallel-transmit coil produced significant advantages-the transmit field was 38% more uniform than a traditional circularly polarized mode and 54% more power-efficient, demonstrating that parallel-transmit coils should be used for monkey imaging at ultrahigh field strengths. The receive coil had the ability to accelerate along an arbitrary axis with at least a three-fold reduction factor, thereby reducing geometric distortions in whole-brain EPI.

Diffusion-weighted tractography in the common marmoset monkey at 9.4T

The common marmoset (Callithrix jacchus) is a small New World primate that is becoming increasingly popular in the neurosciences as an animal model of preclinical human disease. With several major disorders characterized by alterations in neural white matter (e.g., multiple sclerosis, Alzheimers disease, schizophrenia), proposed to be transgenically modeled using marmosets, the ability to isolate and characterize reliably major white matter fiber tracts with MRI will be of use for evaluating structural brain changes related to disease processes and symptomatology. Here, we propose protocols for isolating major white matter fiber tracts in the common marmoset using in vivo ultrahigh-field MRI (9.4T) diffusion-weighted imaging (DWI) data. With the use of a high angular-resolution DWI (256 diffusion-encoding directions) sequence, collected on four anesthetized marmosets, we provide guidelines for manually drawing fiber-tracking regions of interest, based on easily identified anatomical landmarks in DWI native space. These fiber-tract isolation protocols are expected to be experimentally useful for visualization and quantification of individual white matter fiber tracts in both control and experimental groups of marmosets (e.g., transgenic models). As disease models in the marmoset advance, the determination of how macroscopic white matter anatomy is altered as a function of disease state will be relevant in bridging the existing translational gap between preclinical rodent models and human patients.NEW & NOTEWORTHY Although significant progress has been made in mapping white matter connections in the marmoset brain using ex vivo tracing techniques, the application of in vivo virtual dissection of major white matter fiber tracts has been established by few studies in the marmoset literature. Here, we demonstrate the feasibility of whole-brain diffusion-weighted tractography in anesthetized marmosets at ultrahigh-field MRI (9.4T) and propose protocols for isolating nine major white matter fiber tracts in the marmoset brain.