Fig 1.
A) Prototype of the two-element electromagnet, as well as a diagram showing the axis orientations and FOV of interest (20cmx20cm). The system is suitable for many different clinical applications including B) breast or liver MRI, as well as spinal imaging. C) 3D illustration of the magnitude B0-field created by the NuBo system at max power to the coil. D) The magnetic field gradient in the yz-plane overlaid with a vector field representing the direction of the magnetization. The z-component is the major contributor to the magnetic field. E) The magnetic field gradient in the xy-plane. F) Graph showing the magnetic field change with depth, along 3 different points in the FOV (dotted lines in E).
Fig 2.
Field-cycling and prepolarization.
A) Plot of the magnetization Mz of the curved plane over time for 5 evenly spaced slices, from a 20cm height (slice 1) to the surface of the magnet (slice 5). B) Plot showing how current to the electromagnet is adjusted in steps to achieve slice selection. C) Figure showing the location of the 5 evenly spaced slices within the FOV.
Fig 3.
A) An example of a nonplanar imaging slice selected from the inhomogeneous B0-field. The rampable NuBo magnet adjusts the location of the readout slices such that the entire volume can be scanned. B) and C) show a subset of 9 different nonlinear encoding patterns generated by transmitting the Bloch-Siegert off-resonance pulse on 2 or 8 coil elements, respectively. These patterns are analogous to the spatial encoding fields used in nonlinear projection imaging.
Fig 4.
A) Spin echo sequence with Bloch-Siegert encoding pulses. The 90 and 180 pulses are applied in the low-field phase of the field-cycling, and the Bloch-Siegert pulses serve as readout gradients generating in-plane spatial encoding. B) The multi-echo encoding CPMG pulse sequence used in experiments, is designed in such a way that the Bloch-Siegert pulse acts much like a blipped phase encode gradient in turbo spin echo imaging. To collect a full k-space trajectory, the Bloch-Siegert pulses were alternatively applied to odd and even echoes.
Fig 5.
A) A large Tx-only volume coil placed inside the NuBo system was used to transmit the excitation and refocusing pulses. A 3x3 RF planar array was positioned at the base of the magnet to perform spatial encoding exploiting the Bloch-Siegert shift and receive the MR signal. Data was acquired using a single channel in the center of the FOV. B) The two large water-filled phantoms with a diameter of 9cm and 5cm that were individually imaged. C) Photograph of the 3x3 RF planar array used for spatial encoding. Each element in the array is switched-tuned to an off-resonance frequency (870kHz) to transmit the high-power Bloch-Siegert pulse, and the resonance frequency (1MHz) to receive the NMR signal. D) Diagram of the array coil with the transmit channels used in experiments indicated in blue, and the receive channel in green.
Fig 6.
The custom built NuBo electromagnet and supporting hardware systems. A spectrometer with 9 Tx channels and 16 Rx channels serves as the MRI console, and controls the gradient, RF, and pre-amplifiers. The necessary T/R switches for all 9 channels were built in-house. The path labeled “1CH” from the RF amplifier is for the single channel used to transmit the 1MHz excitation and refocusing pulses to the volume coil. The path labeled “9CH’’ is for the 9 channels that are used to transmit the 870kHz off-resonance Bloch-Siegert pulse and receive the 1MHz MR signal from the RF planar array.
Fig 7.
B0 field-cycling characterization.
A) Plot showing the command signal used to drive the electromagnet (blue), and the current feedback (red). Maximum power is applied to the magnet for prepolarization, and then incremented stepwise for slice selection. B) Magnified plot at peak current (blue box), indicating a settling time of 0.8ms. C) Magnified plot at stepwise ramp down of input current (green box), indicating a settling time of 1.1ms.
Fig 8.
A) Graph showing the peak amplitude for 650 echoes collected using a CPMG pulse sequence, and the experimental setup described in Fig 5. The plot shows the signal amplitude at the echo peaks when the experiment was conducted with (blue) and without (red) a large water phantom placed inside the imaging FOV. B) Plot of the first 5 echoes in the echo train.
Fig 9.
A) The first 10 echoes out of 320 echo train from a single encoding pattern. Each echo was acquired with 94 points, with a dwell time of 10us. The midpoint of each echo is used as the k-trajectory sample. B) Example of a 1D time domain MR signal formed by rearranging the midpoint of each echo.
Fig 10.
From left to right: digital phantom representing a cross section of each bottle phantom across the imaging slice, the corresponding simulated, and experimental reconstruction results.
Fig 11.
The resolution achievable in the A) x-direction and B) z-direction with the encoding patterns generated by a planar 3x3 RF array coil, from applying the Bloch-Siegert pulse for 5.88ms. The black box indicates a 20cmx20cm region in the center of the FOV.
Fig 12.
Simulation results demonstrating that image resolution with depth can be improved by increasing the duration of the Bloch-Siegert pulse. First column shows the high-resolution liver scans that were used as phantom images in the simulation experiments, at a 1.4cm depth (top row), and 6.4cm depth (bottom row). Middle column shows the results when the images were encoded using a Bloch-Siegert pulse with a 5.88ms length, and the last column shows the results when this pulse length was doubled to 11.76ms.