Large T1 contrast enhancement using superparamagnetic nanoparticles in ultra-low field MRI

Superparamagnetic iron oxide nanoparticles (SPIONs) are widely investigated and utilized as magnetic resonance imaging (MRI) contrast and therapy agents due to their large magnetic moments. Local field inhomogeneities caused by these high magnetic moments are used to generate T2 contrast in clinical high-field MRI, resulting in signal loss (darker contrast). Here we present strong T1 contrast enhancement (brighter contrast) from SPIONs (diameters from 11 nm to 22 nm) as observed in the ultra-low field (ULF) MRI at 0.13 mT. We have achieved a high longitudinal relaxivity for 18 nm SPION solutions, r1 = 615 s−1 mM−1, which is two orders of magnitude larger than typical commercial Gd-based T1 contrast agents operating at high fields (1.5 T and 3 T). The significantly enhanced r1 value at ultra-low fields is attributed to the coupling of proton spins with SPION magnetic fluctuations (Brownian and Néel) associated with a low frequency peak in the imaginary part of AC susceptibility (χ”). SPION-based T1-weighted ULF MRI has the advantages of enhanced signal, shorter imaging times, and iron-oxide-based nontoxic biocompatible agents. This approach shows promise to become a functional imaging technique, similar to PET, where low spatial resolution is compensated for by important functional information.

hepatocellular carcinoma 9,10 , prostate cancer 11 , sentinel lymph node localization in breast cancer 12 , diagnosis of cardiovascular disease 13 , identifying inflamed carotid plaques 14,15 and cell tracking 16 . Even though T 2 suppression of SPIONs can improve the imaging contrast by reducing the MR signal, T 1 agents are preferred since they enhance signal, preserve the underlying tissue signal, and shorten the imaging times. T 1 contrast enhancement has been reported at high magnetic fields for certain SPIONs [17][18][19][20] including ultra-small-size Fe 3 O 4 nanoparticles (<3 nm) 18 , Gd doped Fe 3 O 4 nanoparticles (4.8 nm) 19 and Fe 3 O 4 nanoplates 20 . In some cases the relaxation is believed to be due to surface-spin canting effects.
Here, we present an alternative approach for achieving T 1 contrast enhancement from SPIONs by going to ultra-low magnetic fields. In ULF MRI systems, the operating magnetic field B 0 is around 0.13 mT, corresponding to a resonance frequency of f 0 = 5.56 kHz, which is four orders of magnitude lower than a typical clinical MRI field of 1.5 T (f 0 = 63.87 MHz) [21][22][23][24][25] . At ultra-low field, the magnetic moment of the SPIONs is not saturated, and they have large spin fluctuations. T 1 (positive) contrast can be greatly enhanced since the proton Larmor precession period is comparable with the relaxation times of magnetic nanoparticles [26][27][28][29] , leading to a large increase in the proton spin longitudinal relaxivity r 1 . The proton spin relaxivities, r 1 and r 2, are the change in relaxation rates R 1 and R 2 per unit Fe concentration, respectively, where R 1 = 1/T 1 , R 2 = 1/T 2 . In addition, the smaller non-saturated magnetic moment of the nanoparticle reduces the transverse relaxivity, r 2 , compared to the high field r 2 , minimizing undesirable negative contrast.

Results
Magnetic resonance measurements. MR measurements were performed in a low-cost ULF MRI system with a Faraday coil detector and a bilateral planar coil array that includes the B 0 coil and the gradient field (G x , G y and G z ) coils. The earth's field was cancelled by two pairs of Helmholtz coils (B x and B y ), as shown in Fig. 1(a). To generate a measurable signal for the ULF MRI scanner, a pre-polarization field pulse (34.5 mT) is applied to enhance the proton magnetization before the MR pulse sequence. For imaging, we used a typical T 1 -weighted ULF MRI gradient echo (GRE) pulse sequence with repetition time (TR) of 400 ms, and gradient echo time (TE) of 27 ms as shown in Fig. 1 The synthesis processes along with transmission electron microscopy images are described in the Methods section and in references [30][31][32] . SPIONs were diluted with deionized (DI) water and transferred to 16 mL sample vials for ULF MRI imaging. The particle concentrations range from 0.25 µg/mL to 40 µg/mL by weight corresponding to Fe concentrations ranging from 0.00325 mM to 0.52 mM. A photograph of a sample vial with SPION solution is shown in Fig. 1(c). T 1 -weighted ULF MRI 2D GRE images are obtained with a pixel size of 2.8 mm × 3.6 mm with imaging duration of 1.6 hours, which is slower than that of superconducting quantum interference device (SQUID) detected ULF MRI (due to our lower signal to noise ratio) but sufficiently fast enough for the study here. The ULF MRI showed significant differences among five types of SPION solutions in Fig. 2(a-e). The ULF MRI signal versus Fe concentration for each SPION solutions are plotted in Fig. 3(a) (with the signal normalized to the DI water signal set to an arbitrary intensity of 1000). The ULF MRI signal initially increases (positive contrast) with increasing Fe concentration, at very low Fe concentration, due to the suppression of T 1 , while at high Fe concentration the signal decreases (negative contrast) due to shorter T 2 times. This leads to a peak in the signal vs. Fe concentration; similar phenomenon has been observed in regular Gd-based T 1 contrast agents 33 . To achieve the same T 1 -weighted ULF MRI signal, the required Fe concentration for the 18 nm Zn 0.3 Fe 2.7 O 4 @SiO 2 nanoparticles is one tenth of that required for the 11 nm Fe 3 O 4 nanoparticles.
Pulse sequence and relaxivities. T 1 of the SPION solutions is measured with a prepolarization evolution pulse sequence 34,35 . In our setup B 0 = 0.13 mT and T 1 = 2.7 s for DI water and T 1 = 0.12 s for the MnCl 2 solution with Mn concentration of 0.167 mM. These results are comparable with SQUID based ULF MRI system of B 0 = 0.128 mT, as reported in ref. 35 . R 1 and R 2 are linearly increasing as a function of the Fe concentration of five SPION solutions as shown in Fig. 3(b,c). The r 1 , r 2 , r 2 /r 1 ratio, and ULF MRI maximum signal for the five SPION solutions are listed in Table 1. The important figures of merit for positive contrast are a large r 1 value and a low r 2 /r 1 ratio. This insures a large signal increase at low Fe concentrations. The 18 nm Zn 0.3 Fe 2.7 O 4 @SiO 2 nanoparticles show a very high r 1 = 615 s −1 mM −1 , with only a modest increase in the r 2 /r 1 = 2.7. In contrast, commercial high-field Gd-based T 1 agents have r 1 = 3 to 7 s −1 mM −1 , with r 2 /r 1 ~ 1.2 to 1.5 1 .
The MR signal is proportional to the transverse proton magnetization, M t , at an echo time TE when the signal is acquired. M t is a function of the pulse sequence parameters TR, TE and Fe concentration-dependent relaxation rates R 1 , R 2 as described by the standard signal equation 36 where C Fe is Fe concentration, and α = π/2 is flip angle of rotation. We use Eq. 1 to simulate how the detected proton spin magnetization varies as a function of Fe concentration in the different SPION solutions, plotted in Fig. 3(d), and obtained good fits to our experimental data using the implemented pulse sequence parameters and measured relaxation rates. The highest value of r 1 (615 s −1 mM −1 ) gives a 6 times enhanced MRI signal at low Fe concentrations (0.01 mM), which implies that ULF MRI may provide a technology for functional imaging using specially designed high-visibility SPIONs giving positive contrast.

The r1 and r2 values at high fields (3 T).
We measured the r 1 and r 2 values at high fields (3 T) using a standard inversion recovery sequence with variable inversion times for T 1 and a spin echo sequence with variable echo time for T 2 . For the 16 nm Fe 3 O 4 solutions we found that r 1 = 11 s −1 mM −1 and r 2 = 137 s −1 mM −1 (r 2 /r 1 ratio of 12.5). After the high field MRI scan, the SPION solutions are sonicated and measured again with ULF MRI, and a reduction of the ULF MRI signal was observed, indicating that agglomeration of the SPIONs took place during the high-field MRI scan. Cluster or chaining of magnetic nanoparticles in large magnetic fields, due to their strong magnetic dipolar attraction, is a major limitation of high-field agents putting limits on the size of magnetic moments that can be utilized and creating stringent requirements on surface ligands to provide adequate separation. In comparison, repeat images in the ULF scanner showed no agglomeration effects, which is a significant advantage of ULF-MRI based nanoparticle contrast imaging.   Magnetic properties of nanoparticles and AC susceptibility. The room temperature magnetization versus applied magnetic field and zero field cooled/field cooled (ZFC/FC) loops of five SPION solutions are shown in Fig. 4(a,b). The closed Langevin-like hysteresis loops and low blocking temperatures (the temperature above which thermal energy overcomes anisotropy energy, causing the net moment of the particle to fluctuate randomly) indicate that all five types of nanoparticles are superparamagnetic at room temperature. The blocking temperatures of Fe 3 O 4 nanoparticles are not linear with particle volume, which is possibly due to different magnetic anisotropy energy densities resulting from different synthetic conditions. To better understand the T 1 contrast enhancement mechanisms in SPION solutions at ultra-low field, we measured the imaginary part of the magnetic AC mass susceptibility (χ") from 10 Hz to 250 kHz for five SPION solutions, shown in Fig. 4(c). The peak in χ" is attributed to either Brownian rotation of the nanoparticle or Néel relaxation of nanoparticle moment and occurs at a frequency that is given by the inverse of the nanoparticle moment relaxation time [37][38][39] . The magnetic fluctuation spectra are proportional to χ" through the fluctuation-dissipation theorem. We therefore expect strong proton spin T 1 relaxation when proton resonant frequency (shown as the dashed vertical line in Fig. 4(c)) is near peaks in χ", where nanoparticle magnetic fluctuation spectral density is large. We found that the χ" spectrum is flat for the 11 nm Fe 3 O 4 solution compared to a higher frequency χ" peak observed at 198 kHz in the 16 nm Fe 3 O 4 solution and at 100 kHz in the 18 nm Fe 3 O 4 @ SiO 2 solution. In comparison, a lower frequency χ" peak is observed at 1000 Hz in the 18 nm Fe 3 O 4 @SiO 2 solution, at 250 Hz in the 22 nm Fe 3 O 4 @SiO 2 solution and at 1000 Hz in the 18 nm Zn 0.3 Fe 2.7 O 4 @SiO 2 solution. We calculated the Néel relaxation frequency of 18 nm Fe 3 O 4 @SiO 2 with the measured magnetic properties and confirmed that the higher frequency of relaxation is Néel relaxation, and the lower frequency relaxation is Brownian relaxation as described in the Materials and Methods section.
We plot the measured r 1 versus χ" at the proton resonance frequency of f 0 = 5.56 kHz for the five different nanoparticles in Fig. 4(d). Also shown is a linear fit of r 1 versus χ"(f 0 ) showing a strong linear correlation. The large and tunable magnetic susceptibility of SPIONS cannot be obtained at high magnetic fields and is a unique attribute of ULF MRI.

Discussion
In SPION solutions, the longitudinal relaxation behavior is due to the relative motion of the water protons near the particle and the magnetic fluctuations of the nanoparticle through physical movement (Brownian translation and rotation) as well as intrinsic fluctuations of magnetic moment (Néel relaxation). According to the fluctuation dissipation theorem, the nanoparticle magnetic fluctuation spectral density should be proportional to the imaginary part of the magnetic susceptibility. These magnetic moment fluctuations give rise to fluctuating magnetic fields experienced by the water protons and lead to strong T 1 relaxation of the proton spins. As seen in Fig. 4(c), it is possible to engineer the magnetic susceptibility and hence the magnetic fluctuations of the nanoparticles, over a wide range of frequencies. While the best sample studied here, the 18 nm Zn 0.3 Fe 2.7 O 4 @SiO 2 , had a peak close to the proton resonance frequency, if we adjust the particle properties or the resonant frequency we could get significant additional enhancement in T 1 relaxivity and agent induced positive contrast. To obtain a high r 1 value, we can tune either the magnetic properties of nanoparticle (moment and magnetic anisotropy energy) or mechanical properties (hydrodynamic shape and volume) to couple the χ" peak with the operating frequency of the ULF MRI. Although our highest r 1 value is due to Brownian relaxation, which may be hindered in more biologically realistic viscous media, Néel relaxation (observed in 16 nm Fe 3 O 4 nanoparticles) still gives appreciable contrast enhancement and can be tuned by engineering the magnetic anisotropy energy. Engineering Néel fluctuations can be done by changing either material properties, size, or shape of the particles, and is advantageous because this contrast will be independent of local tissue environment. Engineering the Brownian fluctuations has the advantage that the agents will give contrast only in special tissue environments (e.g. low viscosity environments) and the contrast may be turned on or off when binding or unbinding from a target.
ULF-MRI with high-visibility positive contrast agents provides a combination of good soft tissue contrast and functional imaging that is more deployable in a surgical suite or a doctor's office. With new high-sensitivity optical magnetometers 40 and advanced machine-learning based image reconstruction techniques 41 coming online, radically new designs are possible with local field probes and coils. Viewing magnetic nanoagents as low frequency stochastic oscillators that strongly couple to nuclear spins, provides a path to greatly increase nanoparticle visibility. These systems may have unique applications for cell tracking, drug delivery monitoring, sentinel lymph node monitoring, and interventional procedures.
In summary, a 6-times T 1 contrast enhancement for the 18 nm Zn 0.3 Fe 2.7 O 4 @SiO 2 solution was achieved using room-temperature ULF MRI system. An extremely high value of r 1 of 615 s −1 mM −1 , over 100 times larger than commercial high field T 1 agents, has been achieved. A linear relationship of r 1 versus χ" at 5.56 kHz, the proton resonance frequency, was observed among all five different SPION solutions, which can be attributed to the magnetic fluctuations of the nanoparticles that relate to either Brownian motion or Néel relaxation. The ability to have large tunable magnetic susceptibility in SPIONS is a unique feature of ULF MRI and cannot be obtained in conventional high field MRI. The ULF MRI platform, combined with the use of designed SPIONs as high visibility positive contrast agents, therefore, creates new possibilities for safe inexpensive functional imaging.

Methods
Magnetic nanoparticle synthesis. The SPIONs have a magnetite structure (Fe 3 O 4 ) as confirmed by X-ray diffraction (XRD) and transmission electron microscope (TEM) 30 . TEM images of the nanoparticles used in this study are shown in Fig. 5. In a typical synthesis of 11 nm Fe 3 O 4 nanoparticles, a mixture of 5.0 mmol of iron acetylacetonate and 15.0 mmol of oleylamine, 3.0 mmol of oleic acid was heated at 200 °C under Ar atmosphere for 4 h. The black colored precipitate was subjected to magnetic separation and washed with a mixture of hexane and acetone several times to remove any uncoordinated amine and acid molecules. The 16 nm Fe 3 O 4 nanoparticles were obtained with the increase of oleic acid concentration to 7.5 mmol. Finally, these amine-capped magnetic nanoparticles were dried at normal ambient conditions. To use these samples as MRI contrast agent, as-prepared Fe 3 O 4 nanoparticles were subjected to surface modification with citric acid. 200 mg of the as-prepared Fe 3 O 4 nanoparticles were dispersed in 20 mL of toluene and mixed with 20 mL of dimethylformamide (DMF) containing 20 mM of citric acid. Under Ar ambient, the mixture was then continuously stirred at 80 °C for 8 h. The final product was subjected to magnetic separation and was washed with ethanol several times to remove uncoordinated citric acid molecules. The SPIONs are diluted with deionized (DI) water and the particle concentrations ranges from 0.5 µg/mL to 200 µg/mL (Fe concentration from 0.0065 mM to 2.6 mM).
Nanoparticles with silica shells and core size of 18 nm Fe 3 O 4 @SiO 2 , 22 nm Fe 3 O 4 @SiO 2 , and 18 Zn 0.3 Fe 2.7 O 4 @ SiO 2 were synthesized by a one-pot solution method through thermal decomposition of a mixture of metal acetylacetonates with surfactants in a high-boiling point organic solvent 31 . Under a gentle flow of Ar, Iron(III) acetylacetonate (2.7 mmol), zinc(II) acetylacetonate (0.3 mmol), sodium oleate (2 mmol) and oleic acid (4 mL) were mixed with benzyl ether (20 mL). The mixture was magnetically stirred under a flow of Ar and then heated to 120 °C for 1 h. Under an Ar blanket, the solution was further heated to reflux (~300 °C) and kept at this temperature for 1 h. The mixture was then cooled down to room temperature by removing the heating mantle. The sizes of nanoparticles were tuned by controlling the heating rate during heating from 120 °C to 300 °C. The silica shells were coated on the hydrophobic nanoparticles via a reverse microemulsion method. The silica coating makes the nanoparticles hydrophilic, leading to aqueous dispersions stable for several years without agglomeration.

Néel/ Brownian relaxation modeling and connection between T 1 and susceptibility.
For application development, the Brownian and Néel relaxation characteristics of magnetic nanoparticles in solution need to be tailored by adjusting particle diameter, composition, and anisotropy (shape and/or crystalline). The dynamic response of a magnetic nanoparticle in solution is described by the complex magnetic susceptibility ω M is the moment per unit volume and → H is the magnetic field in A/m. At low frequencies, the susceptibility has a Debye form characterized by exponential relaxation times τ B,N and corresponding relaxation frequencies π , where the imaginary component of the susceptibility peaks. Néel relaxation dominates when the particle is fixed, and the magnetic moment rotates within the particle. Brownian relaxation dominates when the particle is free to rotate, and the magnetic moment is locked to the internal lattice. For in vivo biomedical applications, magnetic nanoparticles may be lodged in highly viscous tissues that reduce the Brownian rotation. The particle rotation needs to overcome the surface friction in a viscous fluid, which happens at a rotational frequency 38,39 Scientific RepoRTs | (2018) 8:11863 | DOI:10.1038/s41598-018-30264-5 where, k B is Boltzmann's constant, T is the temperature, η is the viscosity of the fluid and R H is the hydrodynamic radius of the nanoparticle. In the case of Néel relaxation the magnetic moment needs to overcome the anisotropy energy barrier and the fluctuation rate is given by where K a is the anisotropy energy density, V is the particle volume, and f 0 is the attempt frequency, here chosen to be 1 GHz. The K a of the particle is typically not that of the bulk material rather an effective K a , influenced by the particle size, surface anisotropy, and magnetite core composition in combination, must be determined. These parameters are controlled using different synthesis processes. The effective anisotropy energy can be estimated from blocking temperature (T B ) determined from the ZFC-FC measurements, where τ is the measurement time.
The extracted K a for 18 nm Fe 3 O 4 @SiO 2 nanoparticle solutions, based on blocking temperature data ( Fig. 4(b)) is K a = 0.15 × 10 5 J/m 3 and K a V/k B T ≈ 3.4 at zero field and room temperature, assuming small variation of K a from the measurement temperature to room temperature. The calculated Néel relaxation frequency (Eq. 1) is f N = 134 kHz. The relaxation spectra (Fig. 4(c)) shows two peaks: one at 100 kHz and the other at 1 kHz corresponding to Néel and Brownian relaxation processes respectively. The Néel relaxation frequency of the 18 nm Fe 3 O 4 @SiO 2 nanoparticle is too high to couple with proton Larmor frequency of current ULF MRI, but it could have potential imaging enhancement at low field MRI operating around 3 mT. Alternatively, one can increase K a by doping 29 or increase the particle volume to slow down the fluctuations. There is an intrinsic connection between magnetic fluctuations and the imaginary part of the magnetic susceptibility given by the fluctuation-dissipation theorem 42 : The magnetic fluctuations give rise to local field fluctuations that cause T 1 relaxation, which in the simplest model 43 is given by: where γ I is the proton gyromagnetic ratio, b is the fluctuating magnetic field amplitude, τ c is the relaxation time of the field fluctuations. To obtain large T 1 relaxation, one needs the relaxation frequency to be close to the proton resonance frequency and have a large fluctuation field amplitude. Magnetic nanoparticles only have large fluctuations at low fields, where the Zeeman coupling between the particles' moment and applied field is sufficiently small relative to thermal energy. At larger fields, Zeeman coupling dominates and the particles become saturated; that is, the net moment of each particle is locked in a unique orientation.
ULF MRI pulse sequences. The imaging MR pulse sequence is illustrated in Fig. 2(b); a pre-polarization field of 34.5 mT pre-aligns the water proton spins to provide a manageable MR signal that can be detected by the Faraday-coil. An AC magnetic field (B 1 ), which drives protons at the Larmor frequency is applied for 1 ms to tip the spins away from B 0 by 90 degrees. An 8 ms "grounding pulse" enables a relay that shorts the detector coil to ground to prevent ring-down noise from saturating the preamplifier. The repetition time (TR) is 400 ms, and the echo time (TE) is 27 ms. For T 1 measurements, a static field B 0 = 0.13 mT is applied continuously. The prepolarizing field B p = 20 mT is turned on for a period t p = 287 ms and switched off adiabatically, allowing the spins to relax for an evolution time t ev in B 0 , before a standard spin-echo imaging sequence is applied. t ev is varied between 0 s and 4 s, depending on the sample. The signal is proportional to the longitudinal magnetization after it has relaxed for a time t ev . The observed signal versus t ev is fit to a simple exponential decay model to obtain T 1 .
For T 2 measurements, a static field B 0 = 0.13 mT is applied continuously, and the prepolarizing field B p = 20 mT is turned on for a period t p = 287 ms. The evolution time t ev = 22 ms is now fixed, after which a 90-degree tipping pulse is applied, followed at a time TE/2 by a 180-degree refocusing pulse. The free induction decay is acquired at a time TE and is proportional to the transverse magnetization after it has decayed for a time TE. TE is varied, and the observed signal versus TE is fit to a simple exponential decay model to obtain T 2 .
The relaxivities are then calculated by plotting R 1 = 1/T 1 , R 2 = 1/T 2 vs Fe concentration, using the slope of a linear fit to the data.

Magnetic properties measurement. Magnetization versus applied magnetic field (M-H) loops and zero
field cooled-field cooled (ZFC-FC) loops are measured using a SQUID magnetometer, as shown in Fig. 3(a,b). The SPION solutions were heat-sealed in polypropylene containers with a particle concentration of 1.0 mg/mL and a mass of about 160 mg. The diamagnetic components are subtracted from the M-H loops using the known mass of the water and sample container. The ZFC-FC loops are measured using the same SPION samples under presence of a 20 mT magnetic field after the initial zero field cooling phase. As shown in Fig. 4(b), the room temperature imaginary part of AC mass susceptibilities of the SPION solutions were measured with a commercial AC susceptometer using SPION solutions with a particle concentration of 1.0 mg/mL, a mass of approximately 180 mg, and an applied AC field of 398 A/m. The ZFC-FC and the AC-susceptibility measurements are done in low magnetic fields, so we expect that the nanoparticles will remain dispersed and in a similar state as probed in the ULF-MRI. When high field measurements are performed, as seen in Fig. 4a, agglomeration and chaining of the nanoparticles may take place altering the magnetic properties.
Data and materials availability. All data needed to evaluate the conclusions in the paper are present in the paper. Additional data related to this paper may be requested from the authors.