Design, construction and performance testing of a 1.5 T cryogen-free low-temperature superconductor whole-body MRI magnet

This work describes the design, construction and testing of a 1.5 T cryogen-free low-temperature superconductor niobium–titanium whole-body magnet with a bore diameter of 850 mm that is suitable for clinical magnetic resonance imaging (MRI) applications. The magnet is actively shielded and passively shimmed to achieve confinement within a 0.5 mT stray magnetic field at 2.5/4 m radial/axial positions and 12.1 ppm field inhomogeneity over a 45 mm diameter of spherical volume. The magnet is conductively refrigerated using a two-stage Gifford–McMahon cryocooler and can be maintained at a steady temperature below 5.7 K during MRI scanning. The MRI scanner assembled with the designed cryogen-free magnet demonstrated stable performance and provided MRI images with good quality and high contrast.


Introduction
Magnetic resonance imaging (MRI) has emerged as one of the most important medical modalities, and it can non-invasively 6 Yulin Wang, Sucong Wang and Ping Liang contributed equally to this manuscript. * Author to whom any correspondence should be addressed.
Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
image internal structures and hidden lesions in the human body. Superconducting MRI magnets generate strong magnetic signals, maintain a homogeneous and stable magnetic field and provide sufficient compensation for spatial harmonic errors by shimming [1]. Most superconducting magnets immerse niobium-titanium (NbTi) coils in liquid helium at 4.2 K. Typically 1500-2000 l of liquid helium are required for clinical MRI [2,3]. The Siemens 1.5 T Prototype MRI magnet decreased the liquid helium volume to less than 50 l using a thermosyphon cryogenic system [4]. GE 3 T Freelium and Phillips 1.5 T BlueSeal MRI scanners further reduced the helium inventory to 20 l and 7 l, respectively [5]. This takes advantage of Gifford-McMahon (GM) two-stage cryocooler conduction cooling [6,7]. A design for extremity MRI examination uses a three-bore magnet of compact size that eliminates the requirement for liquid helium [8].
A cryogen-free solution reduces the cost of operation by removing the need for liquid helium, which has a limited supply and continuously rising price [9]. Removing the liquid helium cooling system significantly decreases the scanner bulk (volume and weight) and obviates the need for cryogen storage and quench ducting [10][11][12]. Additionally, the cryogen refilling process, switching and quench courses are not required, resulting in simpler operation and a reduced safety risk [13,14]. The economic benefits are multiple: less material is used for production, system maintenance, personnel training and operational costs are reduced, and the cycle is faster [10][11][12].
Helium-free magnets still require cooling. This poses some technical limitations, such as high cryocooler power consumption. Most liquid helium-free magnets are applied in smallbore scanners but not in large-bore ones due to the energy requirements. Here we report the design and manufacturing process of a 1.5 T cryogen-free NbTi magnet with an 850 mm bore diameter for clinical MRI applications. We present examples of clinical MRI use and the results of performance tests to demonstrate the feasibility of our magnet.

Magnet design
This magnet is designed as a three-bore cryogen-free magnet. The system consists of several components (figure 1). The main components are the vacuum vessel, radiation shield, double supporting tube, superconducting coils and GM cryocooler.
The structural layout of our cryogen-free superconducting magnet is shown in figure 1, and the main design parameters are listed in table 1. The magnet bore has a diameter of 850 mm to match the gradient coil sizes of the standard MRI scanner bore. The length is 1550 mm. The mass of the structure is 3.1 tons.

Coil design
Nine low-temperature superconductor (LTS) NbTi coaxial coils are wound on the main aluminum alloy bobbin to form the superconducting magnet. Wet winding technology is chosen for its lower cost, higher thermal conductivity, better gap filling and decreased fiber wear. Vacuum-impregnated coils are also a good choice since they have the advantages of higher immunity to impurities and larger mechanical strength, but their high curing temperature is not suitable for the formvar-insulated conductor we used [15]. With the relatively large required diameter of spherical volume (DSV) of 45 cm for imaging, the ratio between DSV and coil radius is relatively high (approaching 1) at the cost of an increased amount of conductors. The number of main magnetic field coils is designed to be seven (coils A, B-D and F-H in pairs) for high field homogeneity. Their locations are optimized to minimize ampere-meters [16]. The other two coils (E and I)  The simulated radial and axial magnetic forces of each coil are listed in table 2. The finite element model (FEM) simulation using the Ansys software (Ansys, Inc., Canonsburg, PA, USA) iterative solver indicates that the stress of the magnet is less than 79 MPa, which is within the tolerance of the coils, and Lorentz force deformation is relatively small (below 0.1741 mm).
NbTi is chosen as the coil superconductor due to its low cost, high commercial application maturity, and good performance robustness [2,17]. The conductor wire-4a (Luvata, Inc., Delaware, OH, USA) insulated by formvar has a rounded shape with an enamelled monolithic configuration and an outer diameter of 1.55 mm. Its superconducting filament diameter is less than 120 µm and the cable is composed of 35 filament strands. This conductor type with 4.8 Cu/NbTi is selected for a  The magnetic forces are the curve linear integral of the total forces on all the coil turns and represented in unit of tons.
suitable trade-off between high current density and good heat conduction. As listed in table 3, the operating current is 300 A to generate a 1.5 T central magnetic field at 4.2 K with 52.49 H inductance and 2.36 MJ stored energy. A quite safe and conservative ratio between the operating current and critical current is set to be less than 30% under a 5.7 K operating temperature, which also provides a simpler cryogenic strategy and larger temperature margin. The peak magnetic field of the coils is 3.73 T.

Magnetic field
Of the nine coils, the two located on the outside, shown as component 7 in figure 1, have a current flowing in the opposite direction to the rest of the seven main coils for active shielding. The stray field simulation suggests that the 0.5 mT field is at 2.5/4 m from the magnet center radially and axially.  The main magnetic field should be highly uniform or there might be image artifacts or signal-to-noise ratio loss, for example [18,19]. The homogeneity of our magnetic field is optimized by the passive shimming method [20]. The field strength deviation is simulated and is shown in figure 3, which indicates that the peak-to-peak inhomogeneity can be less than 9.85 ppm over the 45 cm DSV, covering a suitable imaging volume of interest [6,21].

Cooling
The magnet is conductively refrigerated by a 1.8 W two-stage GM cryocooler (RDE-418D4, Lake Shore Cryotronics, Inc., Westerville, OH, USA) to maintain the superconductor temperature below a safe one of 6.4 K. This safe temperature is set to provide a sufficient temperature margin around the critical temperature, measured to be around 6.7 K-7.5 K. Once 6.4 K is reached, the scanning will be stopped. A temperature of 40 K is achieved at the first stage cold head for the radiation shield and 4 K at the second one for the cold mass. The total heat loads at these two stages are about 30 W and 1.7 W, respectively, based on theoretical simulations. The GM cryocooler was used due to its lower price and more flexible operating  direction than pulse-tube ones, while the vibration is acceptable. The cold head itself contains around 20 l of helium gas. The system is closed and does not require refilling with helium in case of quench. One of the key requirements is to decrease the thermal radiation from the outside environment to the coils. To achieve insulation, the outer vacuum container isolates convectional heat transfer while an aluminum radiation shield is also placed between the outer and inner thermally insulated supporting tubes. Simulation of the temperature distribution of the 3D magnet by the FEM approach is shown in figure 4. The simulation results indicate that the second stage system temperature is in the range 4.09 K-4.92 K, which is appropriate under the safe temperature requirement.
The low-temperature environment of the cryogen-free magnet is completely based on the cold source provided by the GM cryocoolers, which can be divided into first-and second-stage cold heads. The cooling network is illustrated in figure 5. In the first cooling stage, the heat is conducted from the upper and lower end plates of the thermal shield to the first stage cooling plate base, cooling plate, and cooling tube in succession, and then reaches the cold head. As for the second stage, the 4 K upper and lower end plates transmit heat to the second stage cooling plate base, cooling plate, cooling flange column and to the second stage cold head via conduction.

Magnet construction
Nine coils were wound on the bobbins by the winding machine. CTD-521 epoxy (Composite Technology Development, Inc., Lafayette, CO, USA) was used. Stycast 2850 FT is a potential choice for its higher thermal conductivity, but CTD-521 has a smaller specific gravity, lower viscosity, and longer curing time at room temperature, making it lighter in weight and more suitable for the wet winding technology we adopted. Additionally, an alkaline hardener was used. The coils were clamped in series by superconducting joints of NbTi matrix medium using interconnected filaments. The fastened design helps to realize high field stability [22]. After assembling the coils, the cryocooler was connected through the copper tape heat conduction path to the magnet. The magnetic field strength within the magnet was measured and verified by the gauss meter. The superconducting coils were positioned in the low-temperature compatible epoxy resin and encapsulated. The magnet was cooled down within a period of about 1 month to the operating temperature from room temperature and required around 4 h to ramp up. The assembled magnet is shown in figure 6.

Magnet test performance
Our cryogen-free magnet has been operating smoothly in the persistent mode for 3 years. In active quench tests, the recovery time after full-current quench is about 20 h. The ride-through time before quench for the magnet is 20-30 min if electric power is lost. The main magnetic field strength, stray field, spatial and temporal field inhomogeneity, and thermal stability all met the design requirements in a series of experiments. The results of simulations and practical tests are shown in table 4. The magnetic field strength after passive shimming was measured by a Hall probe on 24 axial planes with 24 radial points on each plane. The central field strength was 1.50167 T, satisfying the design specification. The measured radial and axial distances from the 0.5 mT stray field to the magnet center were consistent with the simulation. The field inhomogeneities on the surfaces of different DSVs were verified by measurements at 30-50 cm, as listed in table 5. The achieved 45 cm DSV uniformity was 12.1 ppm, which was close to the simulated value of 9.85 ppm. Afterwards, the magnet was assembled into an MRI scanner for further tests.
To investigate thermal stability, we performed temperature measurements in the assembled MRI system during continuous fast spin echo (FSE) scanning for 90 min [23]. The temperatures of the coil and switch were measured by a temperature sensor (RX-202 A, Lake Shore Cryotronics, Inc.) mounted on the coils and the temperature data were acquired by a real-time monitoring circuit. The results are presented in figure 7. The experimental operating temperature of the magnet is 5.2-5.7 K. The coil and switch temperatures saw a slight rise of about 0.5 K and 0.3 K, respectively, during the first 60 min and then they maintained at a steady state. The temperatures were shown to stay beneath the safety level of 6.4 K and had a safety margin of around 0.7 K.
The imaging performance was tested on this 1.5 T cryogenfree MRI scanner and compared with that of a traditional 1.5 T one (MagicScan, Xingaoyi Medical Equipment Company, Inc., Ningbo, China). The brain and abdomen were scanned with the sequence parameters listed in table 6.
Their images were evaluated independently by two experienced registered radiologists on a single display. The images acquired from the cryogen-free scanner and the helium-bathed one were randomly ordered. The image quality was ranked on five levels based on a Likert scale (5, excellent and can be used for diagnosis; 4, good and can be used for diagnosis; 3, fair and small defects do not affect diagnosis; 2, poor and the diagnosis is influenced; 1, bad and non-diagnostic) [24].
The scorings of the two clinical radiologists are shown in table 7. The cryogen-free MRI scanner can provide images with high clinical diagnosis applicability, comparable to or better than the conventional one. The brain images display good contrast between the gray matter, white matter and  cerebrospinal fluid, and the structure can be seen clearly without artifacts ( figure 8). The internal organs on the abdomen images also display high contrast and a homogeneous signal ( figure 9).

Discussion
The results of performance tests demonstrated the feasibility and stability of our new MRI scanner based on a 1.5 T cryogenfree magnet. The leakage, inhomogeneity of the magnetic field, and thermal variation during scanning met the design requirements. The images achieved high scores that were as good as or better than those from conventional MRI scanners for clinical diagnosis. Safety has been satisfied in the tests we performed. Regarding temperature requirements, the test with the FSE pulse sequence produced satisfactory results. We plan to perform experiments with other pulse sequences for testing multiple possible operating modes. Despite the numerous benefits, such as low cost and convenient operation, the ride-through capacity of the cryogen-free magnet is limited without the heat of liquid helium reserve vaporization during a power cut.

Conclusion
We designed, manufactured, and tested a 1.5 T cryogenfree magnet with LTS NbTi coils. The magnet features an 850 mm bore diameter, a relatively compact stray field, and field inhomogeneity of <12.1 ppm by active shielding and passive shimming. Good thermal stability was achieved by two-stage cryocooler conduction cooling without any liquid helium. The magnet produced a high image quality compared with the traditional scanner. Overall, our test results demonstrate that our new magnet design is feasible and usable in MRI clinical applications.

Data availability statement
The data generated and/or analyzed during the current study are not publicly available for legal/ethical reasons but are available from the corresponding author on reasonable request.