Makoto Tsuda1 and Katsumi Kose1
1University of Tsukuba, Tsukuba, Japan
Synopsis
A
digital MRI system was developed using a digital oscilloscope that had a
waveform generator function. An RF pulse with 43.85 MHz Larmor frequency was
successfully output using this oscilloscope and the NMR signal was captured
using the undersampling technique (144 ns dwell time, 6.94 MHz sampling
frequency). The acquired a set of 3D MR image using a 1.0 T permanent magnet
MRI system demonstrated the validity of our approach.Introduction
Various
approaches to digital MRI systems have been reported [1-4]. Most of them used a
Field Programmable Gate Array (FPGA) to generate radio frequency pulses and
detect NMR signals. However, it is very time consuming and difficult for many
people to develop such a sophisticated digital system. In 2014, we developed a
simple digital MRI system that used a digital oscilloscope, an arbitrary
waveform generator, and three 32-bit small board computers [5]. In this study,
we have successfully further simplified the system using a digital oscilloscope
that can generate the Larmor frequency RF pulse.
Materials
and Methods
Our digital MRI system consists of a
digital oscilloscope (PicoScope 5242B, Pico Technology, St Neots, UK), three
32-bit board computers (Arduino Due, Smart Project, Torino, Italy), and a
laptop PC (CPU Intel Core i5, 2.60GHz) (Fig.1). The digital oscilloscope has
the following specifications: 125 MS(mega sample)/s for two-channel 15-bit mode
input and 200 MS/s for one-channel 14-bit output. This oscilloscope can be used
for both RF pulse generation and RF signal detection (Fig.2).
Figure
3 shows the block diagram of our digital MRI system. One of the board computers
(master system) controls the total timing of the MRI pulse sequence (20 ms
time resolution) and outputs trigger signals to other board computers and the
oscilloscope. The board computers output waveforms of Gx, Gy, and Gz field gradients
synchronously via each DA output to the gradient driver.
We
made the Larmor frequency (43.85 MHz) RF signal by writing the 43.85 MHz
sinusoidal signal to the wave memory of the oscilloscope with 14-bit resolution
and 5 ns dwell time in 1000 word length. The RF pulse was output to the RF
transmitter for excitation. The MRI signal detected by the RF coil is amplified
with a low noise preamplifier and supplied to the oscilloscope through the low-pass
filter (50MHz). To reduce the amount of the sampling data and speed up the data
transfer and processing time, the under sampling technique (dwell time = 144
ns, sampling frequency ~ 6.94 MHz) is used [4]. Imaging experiments were
performed using a 1.0 T MRI system using a yokeless permanent magnet with a 90
mm gap [6]. A 3D gradient echo sequence (TR = 200 ms, TE = 11 ms, image matrix
= 256×256×16)
was used for image acquisition of a water phantom. The phase of the receive
signals were corrected using that of the excitation RF pulses because the external
trigger timing generated by the board computer was not synchronized with the RF
oscillator time base.
Results and Discussion
Figure 4 shows a 2D cross section of a
water phantom selected from a 3D image dataset acquired with a 3D gradient echo
sequence (TR = 200 ms, TE = 11 ms, Image matrix = 256×256×16).
Because the phase correction was successfully performed, no ghosting artifact
was observed.
In
conclusion, because the total cost of the digital instruments is about $2,000
and the time required for the system development is about one year, our system
can be a promising approach to development of a digital MRI system.
Acknowledgements
No acknowledgement found.References
1.
S. Jie, et al, Rev. Sci. Instrum. 76, 105101 (2005).
2.
P. Stang, et al., IEEE Trans Med Imaging. 31(2): 370 (2012).
3.
S. Hashimoto, et al, Rev. Sci. Instrum., 83, 053702 (2012).
4.
P. Pe’rez, et al, Medical Engineering & Physics 26 523 (2004).
5. M. Tsuda, et al., 23th ISMRM, Toronto, Canada. p1843 (2015).
6. T. Shirai, et al, Magn Reson Med
Sci 4,137 (2005).