Healthy volunteer subjects were safely imaged in a 3.0 T MRI system with an ultra-high performance head-gradient coil. The MAGNUS coil operates at 200 mT/m and 500 T/m/s, simultaneously on all 3-axes, and has a patient (head) bore diameter of 37-cm. Substantial reduction in pulse sequence TE, TR, and echo-planar imaging echo spacing was achieved with observable improvement in image quality. The design enabled the achievement of higher peripheral nerve stimulation thresholds compared to that for whole-body gradient coils.
Slew rate in whole-body gradient coils is typically limited to a maximum of 200 T/m/s by peripheral nerve stimulation (PNS). However, gradient amplitudes on the order of 200-300 mT/m are needed for improved brain microstructure connectivity [1], and slew rates >500 T/m/s are essential for improved functional connectivity [2]. Smaller, head-sized gradient coils can achieve substantially higher slew rates because of 2-3x higher PNS thresholds [3-6]. The Compact 3T (C3T) scanner [5], achieves a slew rate of 700 T/m/s and Gmax of 80 mT/m using a 1 MVA driver. In comparison, the Connectome gradient [1] achieves 300 mT/m using an 8 MVA driver but its slew rate of 200 T/m/s is no higher than a whole-body clinical MRI system. Initial in-vivo results using the 42-cm inner diameter MAGNUS gradient coil that can simultaneously achieve 200 mT/m with 500 T/m/s [7] is described.
The MAGNUS
gradient coil was installed in a whole-body 3T (WB3T) MRI (GE SIGNA MR750). A custom patient bore tube (“trumpet”)
(Figure 1) was used to mount the RF shield and 16-rung birdcage RF coil. With the same form factor as the C3T gradient,
the PNS limits of the rheobase and chronaxie time were set to 52.2 T/s and 611 μs, respectively.
Acoustic measurements were made using a Brüel & Kjaer Model 2250 sound level meter. Imaging protocols used for testing the C3T system [8] were used for measuring the peak (dB) and A-weighted (dBA) sound pressure levels (SPL). The sequence TE and TR times were compared to that of WB3T MRI, and also the C3T scanner. A 32 -channel head coil (NOVA Medical, Wilmington, MA) and a standard 8-channel brain coil were used in all experiments. Four healthy volunteers (3 males, 1 female; mean mass = 76 +/- 11 kg, and mean age = 44 +/- 9 years) were consented and scanned under IRB-approved protocols.
Figure 2 shows the results of the acoustic SPL tests at full amplitude and slew rate (200 mT/m and 500 T/m/s). Initial measurements were made with progressive addition of acoustic treatments to remain within safe operating limits with hearing protection: 99 dBA without hearing protection, and 140 dB peak. When compared with the C3T system (80 mT/m and 700 T/m/s), the mean SPLs were about 3 dBA and 3 dB higher for the A-weighted and peak levels, respectively. Additional acoustic treatments are being pursued to reduce the noise level further.
In terms of spatial coverage, across the four volunteers, the cervical spine C4/C5 junction was well visualized without distortion in a 26-cm FOV (Figure 3) after correction for gradient non-linearity using 10th order 3D spherical harmonic correction [9]. As such, the pre-frontal cortex and the cerebellum are well visualized without distortion. In addition, as seen in Figure 4, gradient non-linearity was well corrected in the axial, sagittal, and coronal planes as seen with the ACR phantom. For all subjects, the scans were well tolerated with 2 of 4 subjects reporting mild but not painful sensation in the area of the face.
In Figure 5, 3D-MPRAGE images demonstrate
clear depiction of the anatomical structures and qualitatively appear to be
sharper than those acquired in whole-body MRI systems, as previously observed using
the C3T scanner [5]. With higher performance gradients, the MPRAGE TR for a 1-mm isotropic acquisition was reduced 10%
from 6.1 ms to 5.5 ms compared to the C3T scanner, and reduced 25% compared to WB3T
MR (from 7.4 ms). The reduction in TR resulted in a sharper point spread
function for the FGRE readout. Figure 6
shows a SE-EPI image that has reduced geometric distortion and
signal drop-off compared to WB3T MRI. This was the result of the
echo spacing reduced from 856 μs to 508 μs. Cusp-artifact (“Annefact”)
[10] was noted only in the MPRAGE images. The artifact was minimal and did not affect any part of the brain.
1. Setsompop K, et al. Neuroimage 2013; 80: 220-33.
2. Feinberg D, et al. Neuroimage 2017; 164: 155-63.
3. Lee SK, et al. Magn Reson Med 2016; 76: 1939-50.
4. Tan ET, et al. J Magn Reson Imag 2016; 44: 653-64.
5. Foo TKF, et al. Magn Reson Med 2018; 80: 2232-45.
6. Weiger M, et al. Magn Reson Med 2018; 79; 3256-66.
7. Foo TKF, et al. Proc. 26th ISMRM, p. 839.
8. Weavers PT, et al. Med Phys 2016; 43: 1259-64.
9.
Tao S,
et al. Phys Med Biol 2017; 62:
N18-31.
10. Rangwala N, et al. Magn Reson Med 2010; 64: 220-8.