SQUID-based ultralow-field MRI of a hyperpolarized material using signal amplification by reversible exchange

The signal amplification by reversible exchange (SABRE) technique is a very promising method for increasing magnetic resonance (MR) signals. SABRE can play a particularly large role in studies with a low or ultralow magnetic field because they suffer from a low signal-to-noise ratio. In this work, we conducted real-time superconducting quantum interference device (SQUID)-based nuclear magnetic resonance (NMR)/magnetic resonance imaging (MRI) studies in a microtesla-range magnetic field using the SABRE technique after designing a bubble-separated phantom. A maximum enhancement of 2658 for 1H was obtained for pyridine in the SABRE-NMR experiment. A clear SABRE-enhanced MR image of the bubble-separated phantom, in which the para-hydrogen gas was bubbling at only the margin, was successfully obtained at 34.3 μT. The results show that SABRE can be successfully incorporated into an ultralow-field MRI system, which enables new SQUID-based MRI applications. SABRE can shorten the MRI operation time by more than 6 orders of magnitude and establish a firm basis for future low-field MRI applications.

Alternatively, one of the most promising methods for enhancing the NMR signal strength was proposed by Bowers and Weitekamp in 1986 9 . They suggested that the addition of H 2 molecules enriched in the para-state can result in the enhancement in NMR signals. This method was initially called para-hydrogen and synthesis allows dramatically enhanced nuclear alignment (PASADENA) and was later named para-hydrogen induced polarization (PHIP) 10 . Another hyperpolarization technique, called signal amplification by reversible exchange (SABRE), which was developed recently, is considered a significant extension of PHIP 11 . The SABRE hyperpolarization technique is relatively fast, cheap, and effective and can be successfully used to achieve a non-Boltzmann state necessary to improve the MR signal strength.
On the other hand, low-field and microtesla MRI are promising tools in several fields that can surpass the merits of high-field MRI. These techniques can avoid the potential risk of high-field MRI, and the greater relaxation contrast in low-field MRI than in high-field MRI has been proposed for the detection of tumour cells [12][13][14][15][16][17] . Furthermore, the simultaneous detection of various MRI-active nuclei 18 is a powerful tool for obtaining important information in a short time. In addition, microtesla MRI can change the magnetic field in a short time and harness magnetic field-dependent MRI information 19 . However, the magnetic signal caused by the Zeeman effect in a microtesla-range magnetic field is extremely low. Therefore, highly sensitive sensors, such as a superconducting quantum interference device (SQUID) 20 or an optical atomic magnetometer 21 , are required to detect the NMR/MRI signals. Because the MR signal obtained by these sensors does not have an inductive detection loss depending on frequency, one may reduce the signal loss at the ultralow field by replacing the conventional inductive coil with these sensors. In addition, to obtain a good MR image in SQUID-based microtesla MRI, a pre-polarization field (B p ) is required before the MR signal from the sample is detected. Therefore, some attempts 17,22,23 were made to obtain a strong B p , including optimizing the B p coils and cooling these coils to reduce noise. Increasing B p using a coil also yielded critical effects: the generation of eddy currents on the walls of a magnetically shielded room (MSR) 24,25 and flux trapping in a pickup coil 26 . These side effects can hamper the development of microtesla MRI in the future.
The hyperpolarization technique can solve this pre-polarization problem to enable the polarization of materials with an extremely high order of magnitude. There have been several recent reports 18,22,[27][28][29][30] on this strategy. The most well-known way of achieved hyperpolarization, DNP, which is used to transfer polarization from electrons to nuclei, was harnessed to obtain a high-contrast microtesla MR image 18,22 . Alternatively, SABRE, which induces hyperpolarization on a variety of substrates via low-cost, high-throughput hyperpolarization without complex devices under normal conditions, utilizes para-hydrogen (p-H 2 ) by forming a metal complex with some target substrates and transfers hyperpolarization from a polarized proton to a weakly bound substrate [31][32][33] . Polarization transfer is carried out by matching the scalar coupling/chemical shift and Zeeman energy of active nuclei in a complex structure. Because the ligands in the complex are weakly bound, the exchange of ligands continuously builds up a pool of hyperpolarization in the solution. Several MR studies with SABRE were performed at low field (47.5 mT) 34,35 and earth's magnetic field 36 . Moreover, Buckenmaier et al. reported the detection of a hyperpolarized NMR signal using SQUID-based NMR with SABRE 29 . However, despite its great potential, technically controlled investigations under low-magnetic-field conditions, especially those under ultralow-magnetic-field conditions, are lacking.
Using the above information, we matched the condition for the highest SABRE polarization and designed a SABRE-based MRI phantom. This phantom can be used to obtain a real-time accumulated MR signal from a hyperpolarized proton signal of a small concentration of pyridine by continuously bubbling (Fig. 1). Therefore, in this study, we report the first SQUID-based microtesla MR image based on the para-hydrogen SABRE system. An enhancement factor of more than 2650 and a clear SABRE-derived 1 H MR image of the phantom were obtained with 8 mT B p at 34.3 µT.

Results
Magnetic field-dependent phase changes for ortho-, meta-, and para-hydrogen on pyridine. SABRE polarization is sensitive to an external field, which induces hyperpolarization in a weak magnetic field. Furthermore, quantum mechanically, p-H 2 gives only the singlet state, which is an anti-aligned hyperpolarized spin state of the proton, when it induces polarization transfer; therefore, its phase in polarization may depend on the external magnetic field due to the polarization transfer mechanism, which is related to the chemical shift and scalar coupling of the active SABRE complex [37][38][39] . The polarized nuclear spin states from p-H 2 are more complex than those from polarization at thermal equilibrium. Both of the above factors present a challenge in obtaining polarization in a low field. The challenge arises from the fact that p-H 2 -derived 1 H NMR signals often have an anti-phase character, which results in significant signal cancellation if resonances are not resolved. Therefore, we observed magnetic field-dependent phase changes for ortho-, meta-, and para-hydrogen on pyridine (Fig. 2) and optimized the external field for the most efficient in-phase polarization transfer on all protons in pyridine. As suggested in other reports 37,40 , polarization of the protons in all positions (ortho-, meta-, and para-) on pyridine was maximized in the range from 5 to 9 mT ( Fig. 2(b,c)). The phase changes are due to the mismatch between the Zeeman effect and the J-coupling when SABRE chemical exchange is conducted in a lower magnetic field, e.g., earth's magnetic field of approximately 50 µT, as proposed in previous studies 37-39 . Dependence of the enhancement factor on the B p strength and the B p time. The microtesla NMR experiment was performed to investigate the effect of the SABRE-enhanced 1 H NMR signal (i.e., the pyridine signal) on the external field and the application time of this field. The dependence of the NMR signal on B p is displayed in Fig. 3(a). Because the chemical shifts of the ortho-, meta-, and para-hydrogen on pyridine, including the methanol peak, cannot be distinguished in the microtesla range, all peaks are mixed in the presence of a p-H 2 gas stream. On the other hand, when the p-H 2 flux is absent, the NMR signal of methanol is the main contributor to the measurable proton signal because the proton population difference after the addition of pyridine is The p-H 2 gas entered the 3D phantom through the p-H 2 inlet and many holes in the bottom and generated bubbles. A mesh was attached to prevent the generation of bubbles in the imaging region. (c) Experimental apparatus for the microtesla MRI experiment. As a B p coil, a silicon-oil-cooled pancake-type coil mounted inside a B p Dewar was used. A double Helmholtz coil was used as the B m coil. Two pairs of square-type Helmholtz coils, aligned mutually orthogonally, and the remaining coils were used as the B 1 coil and three-axis gradient coils, respectively. relatively small. Since the phase of the SABRE-enhanced NMR signal, compared to that of the reference signal, was inverted, as shown in the inset of Fig. 3(a), the SABRE-enhanced NMR signal intensity (closed circles) and the reference signal intensity (open circles) increase in the negative and positive direction, respectively, as B p increases. The inset shows the FFT spectra with (dashed line) and without (solid line) p-H 2 flowing at 8 mT B p . Therefore, the enhancement factor (E SABRE ) can be defined as follows: where S SABRE is the SABRE-enhanced NMR signal, S is the reference NMR signal of all protons when there is no p-H 2 gas flow, and α (=99) and β (=4/5) are the volume and nuclear spin ratios of the pyridine and methanol, respectively. The dependence of E SABRE on B p is shown in Fig. 3(b). E SABRE increases with increasing B p and then reaches a maximum value around a B p of 7 ~ 9 mT. The overall trend in E SABRE is quite consistent with the result for the high-field experiment shown in Fig. 2(c). The E SABRE was also obtained as a function of B p time (t Bp ) at 8 mT B p (Fig. 3(c)). In our experiments, E SABRE began to converge to a maximum value for a B p time above 30 s and increased up to approximately 2658 for t Bp = 60 s.

SABRE-enhanced 1 H MR image.
The SABRE-derived 1 H MR image, obtained with 8 mT B p at 34.3 µT, is displayed in Fig. 4. As more methanol evaporated with increasing total imaging time, the MR experiment was performed while accounting for the total imaging time and image resolution. The following experimental parameters were used: t Bp = 13 s, t pe = 0.12 s, t acq = 0.3 s, = ± = . G G 0 6 x z , max µT/cm, 49 phase encoding steps (ΔG z = 0.025 µT/cm), and 4 iterations. Therefore, the resolution, estimated from the experimental parameters, was approximately 1.6 × 1.6 mm 2 . On the other hand, the enhancement factor in the MR experiment was more than 1800 at a given t Bp (=13 s), which was estimated by interpolation of the data in Fig. 3(c). Because the MR signal obtained by the SQUID sensor is proportional to the magnetic field, including the measurement field (B m ) and B p , it is possible to predict a polarization enhancement of approximately 420,000-fold by comparison with the www.nature.com/scientificreports www.nature.com/scientificreports/ thermal polarization at 34.3 µT B m . In the MR image, the outside regions, denoted as the "mesh" in Fig. 1(b), were only faintly visible because of the bubbles of p-H 2 gas. In contrast, the inside regions were reasonably clearly imaged. Among the nine rectangular blocks located on the right side of the phantom, the three rectangular blocks located in the middle could not be distinguished from each other individually on the MR image because these patterns were smaller than the imaging resolution.

Discussion
We performed high-field NMR and microtesla NMR/MRI experiments. From the high-field NMR results, it was confirmed that all protons of pyridine had in-phase polarization in an external field greater than approximately 3 mT. An enhancement factor exceeding 2650, compared with the NMR intensity of the methanol solution, was obtained with 8 mT B p at 34.3 µT. An enhanced MR image of the bubble-separated phantom was obtained using the SABRE technique in the ultralow field. In that phantom, the bubbles are formed at only the margin, and the hyperpolarized pyridine from the bubbling region is diffused into the imaging region. The advantage of the experimental approach introduced in this study is that the relaxation time is negligible because hyperpolarization generation and MR imaging can be performed at the same place at the same time. One of the breakthroughs that must be made in conventional MRI is a shortening of the operation time, which is usually lengthened by increasing the scan time to obtain a sufficient amount of signal-to-noise [41][42][43] . An enhancement factor on the order of a few thousands in real-time NMR/MRI was achieved in the ultralow field, even though the actual level of polarization was on the order of 0.00007%; this value was estimated from the equation E SABRE × tanh(γħB p /2k B T), in which γ is the gyromagnetic ratio of a proton, T is the temperature, and ℏ and k B are Planck's constant and the Boltzmann constant, respectively. This result means that it is possible to shorten the scan time by more than 6 orders of magnitude. After induction of hyperpolarization on biomaterials near MRI devices under normal conditions, the introduction of a hyperpolarized material into the patient would enable a short-time MRI operation. Until now, however, in vivo imaging by the SABRE-hyperpolarization technique has not been reported. Furthermore, additional sensitivity enhancement would be possible with, for example, more hyperpolarized para-hydrogen, deuterated substances, or polarization transfer to 13 C or 15 N with a long-lived nuclear spin state [44][45][46] . Our result, based on this concept, may open up a new field of application of low-field MRI. This work showed the possibility of obtaining a hyperpolarized SQUID-based microtesla MR image based on the SABRE system, which means that any imaging studies using the SABRE system are extended to further advanced studies, especially low-field hyperpolarized MRI. Furthermore, this method led to polarization enhancement by more than 420,000-fold in hyperpolarized proton MRI. This study has established a firm basis for a future low-field MRI system, which holds many advantages over a high-field MRI system. Furthermore, detailed and large-scale research on the development of new SABRE systems and hyperpolarization in biomaterials, such as SABRE-Relay 47 and SABRE in shield enables alignment transfer to heteronuclei (SABRE-SHEATH) 48 , would reduce the time required for the actual application of SQUID-based microtesla MRI in human and magnetic resonance spectroscopy. These ongoing integrated studies will open up a new area of MRI applications that will allow the detection of many useful materials, tumours, and even pathogens in humans by microtesla MR imaging.

Methods
Chemistry. All chemicals were purchased from Sigma-Aldrich and used without further purification unless otherwise indicated. All SABRE experiments were performed with the Ir catalyst [Ir(COD)(IMes)(Cl)] (2 mM) in methanol (approximately 20 mL for SQUID and 0.7 mL for high-field NMR) without degassing 49 . A pyridine solution of approximately 124 mM (10 µL in 1 mL of methanol) was hyperpolarized by polarization transfer. www.nature.com/scientificreports www.nature.com/scientificreports/ Para-hydrogen. Approximately 50% p-H 2 was generated by the catalytic conversion at liquid-nitrogen temperature using a home-built p-H 2 generator ( Fig. 1(a)). Hydrogen gas (a mixture of spin isomers of hydrogen, i.e., ortho-hydrogen and para-hydrogen) was passed through a heat exchanger filled with a FeO(OH) catalyst in a liquid-nitrogen Dewar.
High-field NMR. A Spinsolve Ultra (Magritek) spectrometer operating at 60 MHz was used to obtain hyperpolarized 1 H NMR spectra of pyridine. The instrument was shimmed on 5% D 2 O using an automated algorithm, and the NMR signals were tested with pyridine in benzene. A mixture of pyridine (10 µL) and methanol (1 mL) was added to a 5 mm NMR tube after lowering the oxygen level in the solution by bubbling pure nitrogen gas for approximately 2 min. A home-built solenoid coil was set up to induce the designed magnetic field. Each designed magnetic field was formed to study the magnetic field-dependent phase change. In each magnetic field, the generated p-H 2 was bubbled into a 5 mm NMR tube for 30 s at 23 °C and 1 atm and transferred into the spectrometer (<1 s). Spectra were obtained at 60 MHz and 23 °C in one scan using a π/2 proton pulse. The NMR data were processed and analysed using Mnova software (Mestrelab Research, S.L.). Figure 1(b) shows the 3D-printed p-H 2 phantom with inner dimensions of 50 (L) × 50 (W) × 21.3 (H) mm 3 . The p-H 2 phantom contained pyridine and an Ir catalyst dissolved in a methanol solution. The thickness of the character "K" was approximately 5 mm. Nine rectangular blocks were placed on the right side of the phantom: (top) three blocks with dimensions of 5 (L) × 3 (W) mm 2 and 2 mm spacing, (middle) three blocks with dimensions of 5 (L) × 1 (W) mm 2 and 1 mm spacing, and (bottom) three blocks with dimensions of 5 (L) × 2 (W) mm 2 and 2 mm spacing. Because the p-H 2 gas entered the 3D phantom through a p-H 2 inlet and many holes in the bottom and generated bubbles, a mesh was attached to prevent the generation of bubbles in the imaging region. Figure 1(c) shows the experimental setup for SQUID-based NMR/MRI experiments installed inside the MSR. A dc-SQUID connected with a second-order gradiometric pickup coil placed inside a liquid-helium SQUID Dewar was used to detect the NMR/MRI signal. B m and B p were generated using a double Helmholtz coil and a pancake-type coil cooled with silicon oil that was mounted inside the B p Dewar, respectively. A circularly polarized nutation pulse (B 1 ) 50 was used to avoid the Bloch-Siegert effect in the microtesla region. Two pairs of square-type Helmholtz coils, aligned orthogonally to each other, were used to create a circularly polarized π/2 B 1 . Three-axis gradient fields (G x , G y , and G z ) were generated using the rest coil sets. Because the p-H 2 generator shown in Fig. 1(a) was placed outside the MSR, the 3D phantom, mounted between the SQUID and B p Dewars, and the p-H 2 generator were connected using an 8.5 m tube. During the NMR and MRI experiments, p-H 2 gas flowed continuously at a rate of 70 mL/min and bubbled into the reaction mixture inside the 3D phantom at room temperature. www.nature.com/scientificreports www.nature.com/scientificreports/ Pulse sequence for microtesla NMR/MRI. Figure 5 shows the NMR and MRI pulse sequences used in this study. Because an external field on the order of millitesla is necessary to create SABRE polarization, B p was applied during the time t Bp . In this study, the variable range of the B p strength, measured at the centre of the top plate of the B p Dewar, was approximately 2.1 to 10 mT. Because the axes of the coils B p and B m are perpendicular to each other, the nuclear spins were initially aligned parallel to the direction of B p . To rotate the spins in the direction of B m , an adiabatic pulse 23 (i.e., an adiabatic change in B p and B m followed by a non-adiabatic change in B m ) was applied for 35 ms. B m strengths of 106.6 and 34.3 µT were used to satisfy the adiabatic condition and detect the NMR/MRI signal, respectively. After the application of circularly polarized π/2 B 1 , the NMR signal or the MRI signal following the additional gradient echo pulse can be measured using the SQUID sensor. During the MRI experiments, G x and G z were used as the frequency-and phase-encoding gradients, respectively.

Data Availability
The relevant data supporting the findings of this study are available within the paper or are available from the corresponding author, K.K., upon reasonable request.