Limin Li1 and Alice M. Wyrwicz1,2
1Center for Basic MR Research, Northshore University HealthSystem, Evanston, IL, United States, 2Department of Biomedical Engineering, Northwestern University, Evanston, IL, United States
Synopsis
We report the early development of a multifunctional
digital transceiver built on a Field-Programmable Gate Array (FPGA). The multifunctional
transceiver not only offers the functionalities of a conventional transceiver and
pulse sequencer, but also is capable of combining frequency detection and
compensation, inline data correction and image reconstruction. We describe the
design and implementation of the multifunctional transceiver and demonstrate
its capabilities of image acquisition and reconstruction with real-time
frequency and phase compensation.
Introduction
In human and animal MRI,
physiological motions can cause detectable B0 field fluctuations in
the region of interest, which, if not corrected, can lead to severe image
artifacts. Real-time B0 field monitoring and correction is highly
desirable because the expected data integrity should be preserved prior to
image reconstruction. Several methods1,2 have been employed to accomplish
field stability by setting up a feedback loop in which the B0 field
is monitored and corrected during MRI
scans with a minimum feedback latency in 20 -100 ms.
However, a fundamental shortcoming of these methods is that the components of the
feedback loops are distributed over different electronic devices, boards and
even subsystems with feedback data processing performed on CPUs. The unnecessary inter-device data routing and slower processing
increase the feedback latency. To address the feedback
latency, in this abstract, we describe the design for a multifunctional digital
transceiver and demonstrate its potential applications to real-time zero-order B0
field stabilization.Methods
Our strategy for decreasing the feedback latency
is to integrate a digital transmitter and receiver, a pulse sequencer, signal
and image processors on a single device to constitute a multifunctional
transceiver. An MRI electronic
system built on such a multifunctional transceiver will not only have a greatly-simplified
architecture, but also make it possible to establish real-time interactions
among the signal and image processors, the transmitter and the receiver so to
enable real-time B0 field monitoring and correction. In this work we
developed a new transmitter module and integrated it with a previous digital
receiver3,4 with a newly-designed multiple-slice image
reconstruction module. We used a NI USRP-2940R board (National Instruments,
Inc., Austin, TX, USA)
as a hardware platform. Analog signal processing is performed with the analog
front end; while the multifunctional transceiver architecture is implemented on
an on-board FPGA (Fig. 1). On the receiver portion, an input baseband signal,
consisting of in-phase (I) and quadrature (Q) components, is digitized and
delivered to the digital demodulator controlled by a numerically-controlled
oscillator (NCO). New modules (green) are developed and implemented to perform
frequency detection and correction, data correction and image reconstruction. The
data flow within the receiver is controlled by Rx Control module. On the
transmitter portion, the I/Q components of a digital pulse are generated by a
pulse sequencer module and mixed with a second NCO, resulting in a baseband
analog pulse which is routed to the analog front end for up-conversion
modulation to generate an RF pulse. The pulse attributes are specified by the
pulse sequencer which is essentially a FIFO buffer storing a pulse envelope. The
data flow within the transmitter is controlled by Tx Control module. We
used LabView (National Instruments, Inc., Austin,
TX USA)
software package as a development platform. All the modules were designed
around an FPGA chip XC7K410T (Xilinx Inc., San Jose, CA USA). IF signals were sampled at
120 MS/s with 14-bit resolution; digital signals were processed at 80 MHz.Experiments and Discussion
The multifunctional transceiver was tested with
a series of gradient-echo (FLASH) imaging experiments when incorporated with
our Bruker BioSpec 94/30USR spectrometer. The input port of the FPGA board was
directly connected to the output at the pre amplifier of the spectrometer;
while the output port was directly connected to the input of the RF power
amplifier of the spectrometer so that the RF route completely bypassed the RF transmitter,
RF receiver and digitizer boards installed on the spectrometer. A typical FLASH
sequence was slightly modified to allow for acquiring a FID signal for
frequency detection and a gradient-echo signal for imaging. To evaluate the
effectiveness of the real-time B0-field correction, we performed the
following imaging experiments. At the beginning of every MRI scan, the receiver frequency was offset
randomly from one phase-encoding step to the next to simulate zero-order B0
field fluctuations. The Frequency Detection module detected the frequency shift
from the FID signal, which was immediately used to adjust the receiver NCO. Data
Correction module performed signal phase correction. With this setup, the
frequency values can be determined in 75 ns; while the receiver needs ~0.6 ms
to update the local oscillator frequency. Some of the resulting images of
rabbit brain shown in Fig.2 demonstrate that image ghosting artifacts due to
the random frequency offsets are eliminated with frequency compensation and
phase correction. These results verify that the multifunctional transceiver functions
properly and can perform image reconstruction with real-time frequency and
phase compensation. It is important to note that the frequency detection and
compensation, data correction and image reconstruction were all performed on
the same FPGA. The correct images were generated before they were transferred
to a host computer. With the integration strategy, the multifunctional transceiver
can eliminate unnecessary inter-device data routing, and thus dramatically reduce
the feedback latency, making it possible to perform real-time system
adjustments or data correction in response to unpredictable and irreproducible system
or field perturbations in intra scans. Although this single-channel multifunctional
transceiver is suitable only for the zero-order B0 field
stabilization, it could be expanded to accommodate multiple channels suitable
for the higher-order B0 field stabilization.Acknowledgements
This work was supported by NIH grant
R21EB024852.References
1. Boer VO., et. al. Direct B0
field monitoring and real-time B0
field updating in human breast at 7 Tesla. Magn. Reson. Med. 2012;67:586-591. 2. Duerst Y, et al. Real-time feedback
for spatiotemporal field stabilization in MR systems. Magn Reson Med. 2015;73:884-893. 3. Li
L., et. al. Parallel 2D FFT implementation on FPGA suitable for real-time MR
image processing, Rev. Sci. Instrum.
2018;89:093706. 4. Li L., et. al. A multifunction digital receiver suitable for
real-time frequency detection and compensation in fast magnetic resonance
imaging. Rev. Sci. Instrum. 2019;90:053707.