3119

Comparison of SNR between a low-field (0.26T) Tabletop-MRI and a clinical high-field (3T) scanner
Robert Kowal1, Enrico Pannicke2, Marcus Prier1, Ralf Vick2, Georg Rose3, and Oliver Speck1
1Department of Biomedical Magnetic Resonance, Otto von Guericke University, Magdeburg, Germany, 2Chair of Electromagnetic Compatibility, Otto von Guericke University, Magdeburg, Germany, 3Chair in Healthcare Telematics and Medical Engineering, Otto von Guericke University, Magdeburg, Germany

Synopsis

The SNR-performance of a low-field (0.26T) Tabletop-MRI-system was experimentally compared to a high-field (3T) clinical scanner. The SNR of simple FID sequences were evaluated for RF-units with identical coil and sample geometries as well as T/R-switch designs. The SNR was calculated from time signals and put into perspective to compute the relative SNR-performance of the Tabletop-system which was measured at 11.3%. Additional compensations for several differing experimental conditions were carried out. The full comparison suggests limiting factors in the equivalent measurement in the clinical scanner which can result in more SNR loss than expected.

Introduction

The use of small scale, low-field MR-systems has the potential to offer new affordable and very flexible measurement setups1. Not only as a means of students being able to do their first MR-experiments2,3,4, the capabilities of such MR(Tabletop)-systems also allow for biopsy analysis5, X-nuclei measurements and hyperpolarized samples with less wastage due to frequency adjustable RF modules and low sample volumes2,6,7. Additionally low-field systems benefit from lower susceptibility artifacts and higher T1 contrasts8-10. The biggest drawback in using small low-field systems however is the generally associated decrease in signal strength11-13. In this study we experimentally investigated this decrease in terms of comparing the signal-to-noise ratio (SNR) from FID time signals in such a Tabletop-system3,4 and a high-field clinical scanner.

Methods

The systems were an in-house built Tabletop-MRI operating at a static magnetic field strength B0 of about 0.26T (see Fig. 1), and a clinical scanner with B0 of about 3T (Skyra, Siemens). To create comparable measurements in both systems, two similar RF-units were developed for their respective Larmor frequencies (f0: 11.16MHz and 123.26MHz), based on a design originally ment for operation in the Tabletop-system. Each RF-unit consists of an RF-coil (2-turn solenoid) and a Transmit-Receive-(T/R)-switch, serving as the interface to the corresponding MR-system. The coils were wound with 0.8mm-diameter copper wire and loaded with the same water sample of about 3.5cm3 in a test tube (see Fig. 2). Coil efficiencies were computed as quality factors Q using a vector network analyzer14. After isocentric placement, FID-signals were acquired. In the Skyra-system for this purpose the sequence "FID10Hz" for transmit adjustment was used. From the measurements, the SNR was calculated in the time domain based on the quotient of the initial signal amplitude, and the underlying noise floor. This leads to a measure of the homogeneity independent SNR of the system corresponding in the frequency domain to the quotient of the signal integral to noise. The maximally reached values were compared to state the relative SNR-performance, which were obtained when the highest transverse magnetization was achieved. Despite identical coil and sample geometries as well as their approximate temperatures, the experiments conducted feature several differing measurement parameters with an influence on the systematic SNR. Well known is that $$$SNR\propto\frac{Q}{F\Delta f}^{\frac{1}{2}}$$$ as the resulting effects of the coil quality factor Q, preamplifier (PHA-13LN+, Mini-Circuits) noise figure F and receiver bandwidth Δf 15-17. The influence of B0, however, is described in the literature with varying degrees of proportionality, from slightly sub-linear to quadratic dependence18,19. These vary depending on the dominance of the noise sources, such as the sample, the coil and the further electronics20. By incorporating the different experimental conditions listed in Figure 4, the FID-responses were evaluated.

Results

The FID signals recorded in the Tabletop-system reached a maximum SNR of 114.8. Using the same sample in the clinical scanner the maximally achieved SNR-value was 1019. From these values one can conclude the Tabletop's relative SNR-performance of 11.3%. This 11.3% represents the raw measured performance of the low-field system. From a theoretical point of view, due to the proportionalities of the differing measurement parameters in Figure 4 the measured SNR of the Tabletop-measurement was systematically lower. Hence the lower quality factor, higher noise figure and higher receiver bandwidth account for a systematic factor of 8.1 towards the expected relative SNR-performance of the Tabletop-system leading to an actual relative value of 90.8%. This shows an almost identical SNR despite more than 10-fold difference in field strength and has to be caused by further factors limiting the measured SNR of the FID-response in the clinical-system.

Discussion

The experiments performed show a limiting behavior of the relative SNR in the large scale system. This loss can be estimated with a factor of at least 10 due to differences in field strength. Reasons for the low value may stem from the fact that the RF-unit used was optimized for the Tabletop-system and side effects result from the adaptation to nearly 3T. The sample size used does not correspond to the one the clinical-system was optimized for. This is associated with a substantially different noise regime, which is no longer shaped by the sample, as is the case with humans6,18,20. For such small samples, a much lower voltage level is induced than usual which can lead to a loss of SNR if not taken into account. When examining small objects such as the test tube but possibly also small animals21,22, less SNR may be obtained if no appropriate considerations are made, or a scanner optimized for the sample should be used. One should also be mindful of the assumptions made for the comparison, such as the independence of effective coil geometry with field strength or slight differences in the sequences used like deadtimes. Nevertheless, the use of low-field scanners offers even further possibilities to increase the SNR, which can be exploited at these low frequencies onwards or above all, when investigating other nuclides or hyperpolarized samples6,23,24. With litz-wires25 due to higher Q an SNR gain of 13%6 or 31%26 is feasible and by using hyperpolarisation Coffey et al.6 report an SNR drop to only 40% despite a 100-fold B0 difference .

Acknowledgements

Funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation)- Project-ID 422037413 - TRR 287. This work was supported by the Federal Ministry of Education and Research within the Research Campus STIMULATE (Grant Number 13GW0095A) and was supported by the FLEXtronic ego.-Inkubator (FKZ IK 05/2015). Additional credit goes to Ivan Fomin and David Schote for their work on establishing the Tabletop-system.

References

  1. Mathieu Sarracanie and Najat Salameh. "Low-Field MRI: How Low Can We Go? A Fresh View on an Old Debate". In: Frontiers in Physics 8 (2020), page 172. doi:10.3389/fphy.2020.00172.
  2. Clarissa Zimmerman Cooley et al. "Design and implementation of a low-cost, tabletop MRI scanner for education and research prototyping". In: Journal of Magnetic Resonance 310 (2020), page 106625. doi: https://doi.org/10.1016/j.jmr.2019.106625.
  3. Marcus Prier et al. "Educational Tabletop MRI System using the Open-Source Console for Real-time Acquisition (OCRA)" submitted to ISMRM 2021.
  4. Thomas Witzel, Marcus Prier, Ivan Fomin, and David Schote. "OCRA  Open-source Console for Real-time Acquisition". 2020. url: https://zeugmatographix.org/ocra/.
  5. Juan Rigla et al. "Tabletop MRI system development for intraoperative biopsy analysis". In: Oct. 2016, pages 1-4. doi: 10.1109/NSSMIC.2016.8069627.
  6. Aaron Coffey, Milton Truong, and Eduard Chekmenev. "Low-field MRI can be more sensitive than high-field MRI". In: Journal of magnetic resonance (San Diego, Calif.: 1997) 237C (Oct. 2013), pages 169-174. doi: 10.1016/j.jmr.2013.10.013.
  7. OSI². "Open source imaging". 2020. url: https://www.opensourceimaging.org/projects/.
  8. Albert Macovski and Steven Conolly. "Novel approaches to low-cost MRI". In: Magnetic Resonance in Medicine 30.2 (1993), pages 221-230. doi: https://doi.org/10.1002/mrm.1910300211.
  9. Seung Kyun Lee et al. "SQUID-detected MRI at 132 µT with T1-weighted contrast established at 10 µT300 mT". In: Magnetic Resonance in Medicine 53.1 (2005), pages 9-14. doi: https://doi.org/10.1002/mrm.20316.
  10. Whittier Myers et al. "Calculated signal-to-noise ratio of MRI detected with SQUIDs and Faraday detectors in fields from 10µT to 1.5T". In: Journal of Magnetic Resonance 186.2 (2007), pages 182-192. doi:https://doi.org/10.1016/j.jmr.2007.02.007.
  11. José P. Marques, Frank F.J. Simonis, and Andrew G. Webb. "Low-field MRI: An MR physics perspective". In: Journal of Magnetic Resonance Imaging 49.6 (2019),pages 1528-1542. doi: https://doi.org/10.1002/jmri.26637.
  12. Antoine J. Maubon et al. "Effect of Field Strength on MR Images: Comparison of the Same Subject at 0.5, 1.0, and 1.5 T". In: RadioGraphics 19.4 (1999), pages 1057-1067. doi: 10.1148/radiographics.19.4.g99jl281057.
  13. Victor D. Schepkin. "Sodium MRI of glioma in animal models at ultrahigh magnetic fields". In: NMR in Biomedicine 29.2 (2016), pages 175-186. doi: https://doi.org/10.1002/nbm.3347.
  14. Giulio Giovannetti et al. "Conductor geometry and capacitor quality for performance optimization of low-frequency birdcage coils". In:Concepts in Magnetic Resonance Part B: Magnetic Resonance Engineering 20B(Feb. 2004), pages 9 -16. doi: 10.1002/cmr.b.20005.
  15. Anatole Abragam. "The Principles of Nuclear Magnetism". Oxford University Press,1961.
  16. H D W Hill and R E Richards. "Limits of measurement in magnetic resonance".In: Journal of Physics E: Scientic Instruments 1.10 (1968), pages 977-983. doi:10.1088/0022-3735/1/10/202.
  17. D.I Hoult and R.E Richards. "The signal-to-noise ratio of the nuclear magnetic resonance experiment". In: Journal of Magnetic Resonance 24.1 (1976), pages 71-85. doi: https://doi.org/10.1016/0022-2364(76)90233-X.
  18. Christopher Collins, Pierre-Marie Robitaille, and Larry Berliner. "Radiofrequency Field Calculations for High Field MRI". In: Dec. 2007, pages 209-248. doi: 10.1007/978-0-387-49648-1_8.
  19. Riccardo Lattanzi and Daniel Sodickson. "Ideal current patterns yielding optimal signal-to-noise ratio and specific absorption rate in magnetic resonance imaging: Computational methods and physical insights". In: Magnetic resonance in medicine: official journal of the Society of Magnetic Resonance in Medicine / Society of Magnetic Resonance in Medicine 68 (July 2012), pages 286-304. doi: 10.1002/mrm.23198.
  20. D.I Hoult and Paul C Lauterbur. "The sensitivity of the zeugmatographic experiment involving human samples". In: Journal of Magnetic Resonance (1969) 34.2(1979), pages 425-433. doi: https://doi.org/10.1016/0022-2364(79)90019-2.
  21. Olivier Beuf, Franck Jaillon, and Hervé Saint-Jalmes. "Small-animal MRI: Signal-to-noise ratio comparison at 7 and 1.5 T with multiple-animal acquisition strategies". In: Magma (New York, N.Y.) 19 (Oct. 2006), pages 20-28. doi: 10.1007/s10334-006-0048-9.
  22. Mark Difrancesco et al. "Comparison of SNR and CNR for in vivo mousebrain imaging at 3 and 7 T using well matched scanner congurations". In: Medical physics 35 (Oct. 2008), pages 3972-8. doi: 10.1118/1.2968092.
  23. T. Grafendorfer et al. "Can Litz Coils benefit SNR in Remotely Polarized MRI?" In: Proc. Intl. Soc. Mag. Reson.Med.13 (2005).
  24. T. Grafendorfer et al. "Optimized Litz Coil Design for Prepolarized Extremity MRI". In: Proc. Intl. Soc. Mag. Reson. Med.14 (2006).
  25. A. W. Lotfi and F. C. Lee. "A high frequency model for Litz wire for switch-mode magnetics". In: Conference Record of the 1993 IEEE Industry Applications Conference Twenty-Eighth IAS Annual Meeting. 1993, 1169-1175 vol.2. doi: 10.1109/IAS.1993.299045.
  26. William Viqueira, Warren Berger, Juan Parra-Robles, and Giles Santyr. "Litz Wire Radiofrequency Receive Coils for Hyperpolarized Noble Gas MR Imaging of Rodent Lungs at 73.5 mT". In: Concepts in Magnetic Resonance Part B: Magnetic Resonance Engineering 37B (Apr. 2010), pages 75-85. doi: 10.1002/cmr.b.20155.

Figures

Measurement setup on the Tabletop-system. Left side: OCRA-console for controlling the system, supplying the components and connecting to the computer. Right side: Opened system with the sample in the carrier between the neodymium magnets with connected signal and shimming lines.

Coil with sample and TM-board in printed carrier. The sample was placed in the solenoid coil as shown on the left, which was adjusted to the Larmor frequency via the capacitors on the tuning-matching(TM)-board. To create this figure, the top of the RF-shield was removed. On the right side the relative position of the water sample to the coil is shown.

Measurement setup on the Skyra-system. The test tube sample was positioned with the coil perpendicular to the static magnetic field fixed by wooden planks on the patient table. The connection to the MR system was established via the T/R-switch and TIM-connector. Shown on the bottom is an image taken with the patient camera.

Measurement parameters influencing the systematic SNR.

Proc. Intl. Soc. Mag. Reson. Med. 29 (2021)
3119