Principles and applications of magnetic nanomaterials in magnetically guided bioimaging

Abstract

Imaging plays a pivotal role in the precise diagnosis of a variety of diseases. Various imaging modalities have long secured their place in clinics. However, a lot of these techniques come with their own set of secondary complications and side effects, like ionization radiation in X-rays or positron emission tomography (PET) imaging. Unlike those, magnetically guided imaging modalities have an advantage as they do not pose any critical hazard to the body tissues while operating within acceptable thresholds. Magnetic nanoparticles (MNPs) have been thoroughly investigated as a versatile platform as contrast agents for magnetic resonance imaging (MRI) and nanotracers for magnetic particle imaging (MPI), due to their unique physical properties (i.e., super-paramagnetism) at nanoscale dimensions, and ability to function at the molecular level in biological interactions. Based on the underlying mechanism of these techniques, various parameters like shape, size, anisotropy, magnetization, and surface chemistry, become deciding factors in enhancing the performances of MNPs. Focusing on magnetically guided biomedical imaging, this review discusses the fundamental principles of MRI and MPI and their nanoscale imaging agents. This review also summarizes the existing strategies to enhance the performances and significance of MNPs, and new advances in updating current clinical diagnostics and precision nanomedicine.

1. Introduction

A precise diagnosis is the first step toward identifying the disease and formulating a therapeutic strategy for its cure. Various imaging modalities like X-ray scans, ultrasonography, computed tomography (CT) scans, single-photon emission computed tomography (SPECT), positron emission tomography (PET), and magnetic resonance imaging (MRI) have long been established as reliable imaging techniques and have been helping clinicians in diagnosing a wide variety of diseases. The approach to further improve their performances is a constantly evolving field and has shown constant breakthroughs in innovation. A lot of these techniques (like X-ray, and PET/SPECT) involve the use of harmful ionizing radiation and their long-term exposures may lead to secondary complications [1,2]. However, MRI and a newly emerging magnetic modality, known as magnetic particle imaging (MPI) [3–5], circumvent these unwanted side effects as they utilize the external magnetic fields generated within the machine [6–9]. The usefulness of MPI and its strengths in active clinical settings is yet to be evaluated fully but its “laboratory performance” has been quite encouraging for applications like tumor detection and cell tracking in small animal models.

In the art of nanotechnology, both MRI and MPI utilize nanoparticle-based platforms and have demonstrated effective results in the accuracy of early diagnosis and treatment. These multifunctional imaging nanoprobes play an important part in enhancing their performances of the respective modalities. Apart from being an imaging agent, these nanoprobes can also be functionalized and simultaneously utilized in therapeutics (termed theranostics) for diseases like cancer, diabetes, neuropathic disorders, arthritis, tissue engineering, etc. [10–13] Reduction in size, down to nanoscale, imparts a variety of unique physicochemical properties to the materials which have been known to enhance their practical uses when compared to their bulk counterparts. This change in property affects their behavior in the biological microenvironment and facilitates their interactions at the molecular and cellular levels. This assertively argues that in order to operate these imaging modalities to their maximum potential, the nanoprobes require tailorable properties and should be engineered accordingly. Parameters like mode of synthesis and surface chemistry, shape, size, anisotropy, and magnetization, are significantly responsible for boosting the quality of the nanoprobes [14–18].

Being in use for human patients since the early 1980s (when it was first approved for humans), MRI is based on the relaxation of the magnetic moment of a water proton and how it is affected by the application of an external magnetic field [19,20]. Its relaxation is further influenced by the use of MNPs in the immediate microenvironment of the target tissue/organ. The nanoparticles are spatially distributed and utilized as contrast agents as they produce a local magnetic field gradient [21–25]. This influences the spin-spin and spin-lattice relaxation of protons in tissues and allows for a darker or brighter signal contrast. Thus, MRI does not directly identify the MNPs but its effect on water protons is what contributes to image generation. This imaging technique can successfully provide anatomical and functional information over the targeted area with exceptional image quality. With that said, MNPs are being continuously examined to be implemented in a clinical setting to overcome these challenges and improve the time needed for imaging.

On the other hand, MPI is an emerging, non-invasive technique that is based on the application of alternating external magnetic fields as well. Unlike MRI, MPI allows direct mapping of magnetic nanoparticles with enhanced sensitivity and short image acquisition times [26,27]. This technique exploits the superparamagnetic property of the nanoprobes which means that the nanoparticles exhibit a sigmoidal, non-linear magnetization response in alternating external magnetic fields. One of the technique’s shortcomings is that it cannot provide any kind of anatomical information. This necessitates it to be combined with other imaging techniques, such as computer tomography (CT) to offer anatomical mapping and sharp contrast. However, it has an edge over other modalities for having high sensitivity and high resolution with shortened acquisition times. When comparing MPI with MRI, the MPI modality also displays an advantage as the signal intensity is seen to be proportional to the amount of nanotracers present which in turn can be used for their quantitative analyses in vivo. At this time, MPI is only accessible for pre-clinical applications and substantial improvements on the scanners and tracers are being made to advance the process of this modality to be used in a clinical setting [28–30].

Moving forward, the following sections elaborately discuss these modalities for a better understanding of their underlying mechanisms. This would also necessitate a detailed discussion regarding their respective nanoprobes/tracers, their design and engineering, and their applications in imaging and theranostics.

2. Magnetic nanoparticles as contrast agents in MRI

2.1. Fundamental principle of MRI

MRI is a non-invasive diagnosis technique that utilizes the magnetic relaxation of bodily water protons to produce three-dimensional (3D) tomographic images. This modality facilitates the process of differentiating between healthy tissue and diseased tissue. In a static magnetic field (B₀), the magnetic moments of water protons align parallel and antiparallel to the field B₀ with a net parallel spin population of 0.001% [31]. This net proton along B₀ starts precessing around B₀ with a net magnetic moment of M_Z and at a precession frequency of ω₀ = γB₀ (γ = 2.67 × 10⁸ rad/s·T for ¹H) [32]. Furthermore, the irradiation of the resonant radio frequency (RF) pulse (perpendicular to B₀) led to resonant excitation resulting in an increase in the transverse magnetization. The transverse magnetization (M_xy) increases and starts precession in the transverse plane. When the RF pulse is withdrawn, the M_xy gradually returns to the initial state of equilibrium by realigning to B₀. Such spin relaxation process is involved in two different pathways: (a) M_z recovery, which occurs when energy is lost from the excited state to its surroundings and is known as spin-lattice relaxation, and (b) M_xy decay, which occurs when the precessing proton spins on the xy plane lose phase coherence owing to spin-spin interaction (Fig. 1) [19,33]. These processes can be described as the following equations [34].

$$M_z=M_0\left(1-e^{-t/T_1}\right)\left(\text{Longitudinal}\right)$$ (1)

$$M_{xy}=M_0\left(e^{-t/T_2}\right)\left(\text{Transverse}\right)$$ (2)

where T₁ is the time taken by M_z to recover to 63% of its equilibrium value, while T₂ is the time taken by M_xy to decline to 37% of its starting magnetization value in these relaxation processes. Considering these spin relaxations carry anatomical information, they have been recorded by generating 3-D gray-scale MR images. In practical MR imaging, the proton relaxation is progressive and is unable to offer detailed information about the diseases at the molecular and cellular levels. In order to acquire anatomical and functional information, the region to be imaged is supplied with a probe that is capable of affecting proton relaxation and is known as a contrast agent. For example, to accelerate the relaxation processes, a material with high paramagnetic properties (Gd-based) has been utilized, which generates a sufficient signal to produce a vivid MR image. However, superparamagnetic nanoparticles follow an alternate path to achieve the desired MR signal contrast. The magnetic moment of a single superparamagnetic nanoparticle can be thousands of times higher in magnitude than that of a paramagnetic material, resulting in a substantial difference in the magnetic field gradients. This reduces the T₂ relaxation time by this perturbation field, resulting in an increase in the reciprocal of the relaxation time (R₂ = 1/T₂), or the ‘R₂ relaxivity’ of the surrounding proton nuclei. As a result, a high-contrast MR image with details regarding molecular and biological information of the relevant region is constructed.

Fig. 1.

Schematic representation of longitudinal and transverse relaxation of the net proton magnetization [35]. At equilibrium, the net moment of water proton is aligned in the direction of the static magnetic field and precess around B₀ with Larmor frequency ω = γB₀. When a 90° radio frequency (RF) excitation is applied, the net moment switches into the xy plane. The proton spins then precess about the static field, but gradually dephase (causing T₂ relaxation) and simultaneously lose energy to the environment (T₁ relaxation). The spiral indicates the trajectory of the magnetization vector as it relaxes back to its equilibrium value, and the equations express the values of the longitudinal and transverse components of the magnetization in the rotating frame as a function of time [33].

2.2. Mechanisms of MNPs’ contrast enhancement in MRI

Considering magnetic nanoparticles as dipoles of radius ‘R’, the induced local magnetic field is given by equation (3) [34].

$$\Delta B_{NP}=\frac{{\mu }_{0}M}{3}{\left(\frac{R}{r}\right)}^3\left(3{cos}^2\theta -1\right)$$ (3)

where M is the nanoparticle’s net magnetization, θ is the angle with respect to the z-direction, and r is the distance from the nanoparticle center. As a result of the ensuing local magnetic field gradient, the magnetic moments of nearby protons precess at various frequencies with a spectrum width of Δω = γΔB_NP. This leads to a quicker dephasing of protons’ moments in comparison to the dephasing in the absence of MNPs resulting in signal loss (Fig. 2a and b). This subsequently improves the dark contrast of the MR image. The MRI contrast is dependent upon both the image acquisition parameters (such as echo time) and the magnetic properties of the nanoparticles. Since the magnetic properties of MNPs are vastly dependent on their size, composition, shape, and surface features, these parameters can be controlled and engineered to improve MRI contrast. For instance, for the same volumetric measurement, a nanorod would have a higher surface-to-volume ratio when compared to their spherical counterparts. Due to the increased surface area, significant magnetic field inhomogeneities can be generated over a wider outer-sphere volume, allowing for more water protons to be present near the nanorods (Fig. 2c). Therefore, under the local magnetic field of nanorods, more proton nuclei are disturbed/affected, resulting in a faster relaxation due to nanorods than a sphere of the same volume of nanomaterial.

Fig. 2.

The role of MNPs for MR imaging. R₂ relaxation is shown schematically in the following conditions: (a) without a contrast agent; (b) with magnetic nanoparticles (NPs); and (c) with magnetic nanorods (NRs). The water protons de-phase more quickly, enhancing R₂ relaxation, because of the magnetic interaction between the induced local magnetic field of MNPs and the proton spins. Because nanorods have a larger surface area than spheres with the same material volume, more water protons are disturbed in their magnetic field, leading to faster relaxation in the case of nanorods.

The contrast of T₂-weighted MR images depends on the properties of the MNPs, namely, their size, shape, and magnetic moment. Theoretically, the dephasing (which enhances the R₂ value) of water protons in a weak magnetic field gradient generated by the MNPs is categorized into three different regimes [36]. In the case of relatively small-sized MNPs, the proton diffuses significantly faster between MNPs than the resonance frequency shift (Δω). The rapid diffusion process results in R₂ independent of echo time, whereas the R₂ value increases quadratically with M_S and effective diameter of the outer sphere. This process is known as ‘motional averaging regime’ (MAR) introduced by Brooks [37]. The R₂ value in MAR is given by [38]

$${\text{R}}_2=\frac{1}{{\text{T}}_2}=\frac{16}{45}\text{f}{\tau }_{\text{D}}{\left(\Delta \omega \right)}^2=\frac{4}{405\text{D}}\text{f}{\mu }_0^2{\gamma }^2{\left(\text{d}*\text{M}\right)}^2$$ (4)

where ω = γB₀ is the Larmor frequency, Δω = γμ₀M/3 is the frequency experienced by protons at the equator of MNPs, f is the volume fraction of the MNPs in the solution, and d is the outer sphere diameter. The τ_D is a typical time a water molecule can take to diffuse across the outer-sphere diameter and can be defined as, τ_D = d²/4D (D is the diffusion coefficient of water). According to equation (4), the R₂ relaxation (for d < 20 nm) can be tuned by controlling the nanoparticles size and their M_S value.

Since the magnetic field gradient created by larger MNPs is stronger and distributed over a larger volume, proton diffusion is not a prominent component of signal attenuation. Therefore, the R₂ relaxivity becomes independent of diffusion and nanoparticle size. However, it is dependent on M_S and increases when the M_S value rises (directly proportional). This theory was introduced by Yablonskiy and Haacke and is referred to as ‘static dephasing regime’ (SDR) [39]. The strong field inhomogeneities caused by the nanoparticle moment increase the R₂ relaxivity of water protons, which is given by equation (5) [38]:

$${\text{R}}_2=\left(\frac{2\pi }{\sqrt{27}}\right)\text{f}\Delta \omega =0.4\text{f}\gamma {\mu }_{\text{0}}\text{M}$$ (5)

Therefore, by developing large-sized magnetic nanoparticles that lie within the SDR with a higher magnetization value, the R₂ relaxivity can be maximized.

In the case of sufficiently large-sized magnetic nanoparticles (τ_D > 2 τ_TE), the refocusing of the transverse magnetization M_xy is ineffective and results in a lower R₂ value. This process is known as ‘echo limited regime (ELR)’ and was first introduced by Gillis and coworkers [40]. The relaxivity R₂ of water protons in ELR is given by equation (6)

$${\text{R}}_2=\frac{7.2{\text{fDx}}^{1/3}{\left(1.52+\text{fx}\right)}^{5/3}}{{\text{d}}^2}$$ (6)

where x = Δωτ_TE. Finally, we can say that for a certain volume percentage of MNPs, the R₂ value is expected to increase with the increase of MNP size (i.e., MAR) and peak at a specific size ‘d_sd’, which corresponds to the start of SDR. The R₂ value will remain constant until a larger critical size ‘d_el’ is reached, at which point τ_D > 2 τ_TE and the R₂ value decreases as the size is increased (ELR). Here d_sd and d_el are critical diameters of the beginning of SDR and ELR, and are defined as ${\text{d}}_{\text{sd}}={\left(\frac{5\pi \text{D}\sqrt{3}}{2\Delta \omega }\right)}^{1/2}$ and ${\text{d}}_{\text{el}}=2{\left(\frac{1.49{\text{Dx}}^{1/3}{\left(1.52+\text{fx}\right)}^{5/3}}{\Delta \omega }\right)}^{1/2}$, respectively [38]. Thus, if the particle size increased, R₂ values of MNPs are seen to first plateau and then decrease as shown in Fig. 3a [41]. Multicore MnFe₂O₄ nanoparticles with average size 30–130 nm were synthesized to understand the relationship between R₂ and size (see Fig. 3b). An optimum R₂ relaxivity value of 498 mM⁻¹s⁻¹ is achieved for 80 nm-sized MnFe₂O₄ MNPs. In accordance with the outer sphere relaxation theory, the R₂ value in MnFe₂O₄ NPs strongly depends on nanoassembly’s size and reveals all three regimes: MAR, SDR, and ELR. A maximum R₂ (i. e. darker contrast) corresponds to the SDR is observed for 60–90 nm MnFe₂O₄ nanoparticles.

Fig. 3.

MRI contrast properties of MNPs of different sizes. (a) Size dependence of R₂ relaxivity of single core- and multicore-magnetic nanoparticles [41] and (b) Phantom MR images of Fe₃O₄ multicore nanoparticles with magnetic field 3.0 T.

2.3. Development of T₂-weighted contrast agents

2.3.1. Magnetic ferrite-based T₂-weighted contrast agents

According to the discussions above, in MAR, the transverse relaxivity ‘R₂’ is directly proportional to the M_S and effective outer-sphere diameter (d) of MNPs. It has also been well established that the value of M_S increases with an increase in the size of the MNPs. The MNPs can be viewed as core-shell structures with the core magnetically ordered, while the shell appears as a disordered structure known as spin canting/spin disorder layer (see Fig. 4a) [42–45]. Fig. 4b shows the size dependence of M_S values for different magnetic ferrites [46]. As the surface spin canting effects decrease with an increase of MNPs size, the M_S value increases sharply to the corresponding bulk magnetization values. This proportional relationship between M_S and particle size can also result in an enhancement in MRI signal intensity with increasing the size of MNPs [47]. For instance, Cheon et al. [48] observed that the R₂ value gradually increases from 208, 265, and to 358 mM⁻¹ s⁻¹ at 1.5 T as the MnFe₂O₄ nanoparticles’ size increases from 6, 9 and to 12 nm, respectively (Fig. 4c). A similar trend of R₂ values with the size of Fe₃O₄ nanoparticles has also been observed [49–51]. Furthermore, MnFe₂O₄ exhibits a higher MR contrast effect when compared to Fe₃O₄, due to its high magnetization values [52]. The magnetization values of contrast agents can be further controlled by introducing different transition metals (M = Ni, Co, Fe, and Mn) having different magnetic moments into the host Fe₃O₄ nanoparticles. Fig. 4d correlates the effect of M_S values and MNPs on the R₂ relaxivity in a series of different magnetic ferrites and their multicore structures [53]. We can clearly notice that when average MNPs size is in the range of 6 nm–16 nm, the R₂ relaxivity values are strongly related to M_S value. Recently, a pronounced MRI contrast effect has been reported in Zn²⁺ doped MnFe₂O₄ nanoparticles [54]. By replacing Zn²⁺ (with magnetic moment of 0 μB) for Fe³⁺ in the tetrahedral (T_d) sites, the antiferromagnetic interactions between the magnetic ions in the octahedral (O_h) sites and tetrahedral (T_d) sites are reduced, resulting in an increase in magnetization value. The M_S value increases initially as the concentration of Zn²⁺ rises and reaches its maximum when Zn²⁺ is at x = 0.4 in (Zn_xM_1–x)Fe₂O₄ (M=Mn²⁺, Fe²⁺), and then decreases with further substitution of Zn²⁺. The modulation of the M_s values significantly improves the R₂ values from less than 100 to almost 700 mM⁻¹s⁻¹.

Fig. 4.

The effect of the size of MNPs on T₂ MR contrast properties. (a) Schematic representation of the spin-canting effect on the variation of saturation magnetization, and blocking temperature with the MNPs size. (b) Saturation magnetization vs. MNPs size plot for different magnetic ferrite NPs [46]. (c) Size-dependent R₂ relaxivity of MnFe₂O₄ nanoparticles in an aqueous solution at 1.5 T [48]. (d) The variation of R₂ values as a function of the MPNs size and the magnetization measured at 0.47 T for different ferrites. Using the outer-sphere theory, solid lines are computed [53].

Theoretically, the R₂ relaxivity can be further enhanced by controlling the nanoparticle size within the SDR regime. However, iron oxide nanoparticles larger than 20 nm are ferrimagnetic, and the strong dipolar interaction between them causes agglomeration with poor colloidal stability, a short circulation period, and an increase in toxicity. Alternatively, the R₂ relaxivity is improved by synthesizing magnetic nanoassemblies that comprise smaller magnetic nanoparticles, which results in a large magnetic size [55–57]. Recently Barick et al. synthesized Fe₃O₄ nano-assemblies of a size of 40 nm, which comprised ~6 nm nanoparticles by thermal decomposition of Fe-chloride, ethylenediamine and ethylene glycol [55]. The prepared Fe₃O₄ nanoassemblies not only exhibited high M_S when compared to 6 nm isolated counterparts (Fe₃O₄ nanoparticles), but also demonstrated a strong enhancement in the T₂ MRI contrast property (Fig. 5a). The considerable enhancement in the T₂ MRI signal is ascribed to the cooperative magnetism of individual Fe₃O₄ nanoparticles when aggregated into nanoassemblies and the larger effective diameter. Furthermore, the diffusion and dephasing of water protons around nanoassemblies is highly influenced by the assembly size. Therefore, multiple distinct diffusion regimes are expected around the nanoassemblies depending on the diffusion time. Pöselt et al. prepared nanoassemblies of Fe₃O₄ nanoparticles with diameters of 30–200 nm using a PEG-based ligand [56]. Fig. 5b and c shows experimental as well as theoretical R₂ relaxivity values of 30–200 nm-sized Fe₃O₄ nano-assemblies. Motional averaging regime (MAR), static dephasing regime (SDR), and echo-limiting regime (ELR) are seen with increasing MNP size. The maximum R₂ relaxivity is observed in the SDR, which is consistent with the theoretical discussion as mentioned above (see section 2.2). The highest relaxivity for nanoassemblies was observed for the ones that comprised of 13.1 nm sized Fe₃O₄ nanoparticles.

Fig. 5.

The synergistic magnetism and greater outer-sphere diameter of Fe₃O₄ nanoparticle nano-assemblies significantly improved the T₂ MR signal. (a) Relaxation rates T₂ of the Fe₃O₄ nano-assemblies, Fe₃O₄ nanoparticles and ferumoxytol as a function of the Fe concentration. (b) The variation of R₂ values of Fe₃O₄ nanoassembly with the hydrodynamic diameter (d_hyd) [56]. The relaxivity measurement was taken in 1.41 T with pulse spacing of 1 ms (black Square), 0.3 ms (red circle), 0.1 ms (green triangle), and 0.05 ms (blue diamond). The Fe₃O₄ nanoassembly is made by using 13.1 nm-sized nanoparticles. The black lines are analytical curves using the formula given by Gillis et al. [40] (c) A schematic illustration of outer-sphere–inner-sphere model of a Fe₃O₄ nano-assembly based on H₂O in the vicinity and H₂O diffused into the porous structure of the assembly.

Although optimization of the M_S value has been successfully explored by controlling size, composition, and effective diameter of Fe₃O₄ nanoparticles, a much higher R₂ relaxivity can be attained by designing anisotropic nanoparticles. From previous discussions [42,58], we have observed that surface spin disorder changes with shape anisotropy; hence, altering the morphology of nanoparticles can improve M_S and, as a result, R₂ values. Additionally, the effective diameter of nanoparticles is highly dependent on its morphology and can also be controlled by making differently shaped (cube, rod, octapod, plate, disc and etc.) Fe₃O₄ nanoparticles. Recently, Zhao et al. successfully synthesized octapod (Fig. 6a–c) and spherical iron oxide nanoparticles with a similar M_S value and material volume (a M_S and solid volume of 30 and 20 nm length octapod is equivalent to the 16 and 10 nm sphere, respectively) [59]. The R₂ relaxivity values of 30, 20 nm octapod and 16 and 10 nm sphere are found to be 679, 209, 125, and 59 mM⁻¹s⁻¹, respectively. This demonstrates that the property of shape anisotropy can be successfully translated into efficient MRI contrast agents with stronger T₂ contrast effects. For example, in their work, the R₂ value of a 30 nm-sized octapod is seen to be approximately 5.4 times higher than that of a 16 nm sphere. The higher R₂ value for octapod Fe₃O₄ is attributed to a larger effective diameter. Under an external magnetic field, B₀, a strong inhomogeneity in the local magnetic fields is generated over a larger outer-sphere volume for the octapod shape than for a spherical shape. This induces rapid proton dephasing and enhances T₂ shortening. Shape properties of MR contrast enhancement are also explored by several groups [59–62]. A high-temperature solution phase approach is utilized to synthesize spherical and faceted irregular (FI) CoFe₂O₄ nanoparticles of various sizes (Fig. 6d–g) [60]. With increasing size, the R₂ value of these CoFe₂O₄ nanostructures’ is seen to rise, from 110 to 301 mM⁻¹s⁻¹ for 6and 15 nm spherical CoFe₂O₄, and from 155 to 345 mM⁻¹s⁻¹ for 12 and 25 nm FI CoFe₂O₄. Additionally, the relaxivity of FI CoFe₂O₄ is smaller than that of its spherical counterparts. However, in terms of M_S value, the FI CoFe₂O₄ exhibits improved R₂ value than the spherical CoFe₂O₄, which could be attributed to a higher outer-sphere volume (because of larger surface to volume ratio of NPs) of FI CoFe₂O₄. These findings show that the R₂ value of CoFe₂O₄ is controlled by its shape as well as its M_S value. Despite all these advancements in MR contrast properties of ferrite-based nanoparticles, it is also feasible to improve the magnetic characteristics and R₂ relaxivity by developing new shapes, synthesizing superparamagnetic nanoparticles in SDR, and controlling the surface properties of the resulting nanostructures.

Fig. 6.

The effect of the shape of nanoparticles on T₂ MR contrast properties. (a) TEM images of 30 nm (edge length) Fe₃O₄ octapod particles. (b) R₂ relaxivity of spherical and octapod-shaped Fe₃O₄ nanoparticles with the same geometric core volume. [59] (c) T₂-weigthed MR images of octapodshaped MNPs(20 nm and30 nm) and spherical-shaped MNPs (10 nm and 16 nm) in aqueous dispersions with different mM of Fe, using a 7T MRI scanner. The schematic illustration in the (c) panel shows the analogous outer-sphere models of both the MNPs of the same volume and Ms. The octapod-shaped MNPs have a relatively larger effective radius (radius, R) than the spherical-shaped one (radius, r) with R~2.4r. (d) TEM image of CoFe₂O₄ nanoparticles synthesized without (sphere) and with (faceted irregular) magnetic field. [60] (e) Plot of R₂ value versus saturation magnetization of spherical and FI nanoparticles. (f) Plot of R₂ value versus magnetic nanoparticle sizes between different spherical and FI nanoparticles.

Controlling the shape of Fe₃O₄ NPs to a high-aspect-ratio nanorods with lengths of 30–70 nm and widths of 4–12 nm has resulted in an efficient MRI contrast agent with a high R₂ relaxivity values [63]. (Fig. 7a and b). Fe₃O₄ nanorods encapsulated in polyethyleneimine with a length of 70 nm exhibit an extremely high R₂ relaxivity value of 608 mM⁻¹s⁻¹. This study acknowledges the roles of larger surface area and anisotropic morphology in increasing the MR contrast of the nanorods. The larger surface area of nanorods induces a greater magnetic field perturbation over a larger outer sphere. COMSOL Multiphysics is used to calculate and compare the local magnetic fields generated by both spherical-shaped and rod-shaped nanoparticles. According to the simulation findings, the rods produce significantly more local field inhomogeneity than the sphere (see Fig. 7c and d). A greater local magnetic field is thus produced over a bigger volume than the sphere due to the rod shape structure and larger outer-sphere diameter of the nanorods. Nanorods have a higher R₂ value than spheres of similar material volume because a larger volume of water protons are exposed to the induced magnetic field over a larger volume and quickly dephase as a result.

Fig. 7.

Effects of shape anisotropy on MR contrast properties. The fluctuation of R₂ value with (a) nanorod length and M_S value, and (b) nanoparticle diameter and M_S value. The R₂ values rise linearly with nanorod size, from 312 to 608 mM⁻¹ s⁻¹. This linear increase is attributable to an increase in both the M_S value and the surface area. The local magnetic field created by equivalent material volume rod and sphere Fe₃O₄ NPs under a 3 T applied magnetic field. (c) and (d) Illustrate the induced magnetic field distribution of nanorod and nanosphere, respectively [63].

2.3.2. Magnetic metal nanoparticles-based T₂-weighted contrast agents

Besides the aforementioned magnetic ferrite nanoparticles, the magnetic metallic nanoparticles have also been examined as the MRI contrast agents. Metallic nanoparticles have a high M_S value, which results in a high MRI signal intensity. Pharmaceutical grade Fe nanoparticles dispersion has been produced by laser-induced pyrolysis of iron pentacarbonyl (Fe(CO)₅) and flowed by coating the nanoparticles with dextran [64]. From in vivo MRI experiments performed on rats [64], it was demonstrated that the inclusion of a Fe-based metallic core in the nanoparticles enhances the contrast of the T₂ MR images by 60% in comparison to Feridex. High saturation magnetization values have been observed in Fe/Fe₃O₄ or Fe/MFe₂O₄ (M = Fe, Co, and Mn) core/shell nanostructures [65–68]. The magnetic oxide shell stabilizes iron core under ambient conditions and also in physiological solutions. The surface functionalized Fe/Fe₃O₄ NPs displayed a R₂ value of ~220 mM⁻¹ s⁻¹at 1.5 T, which is nearly twofold larger than that of the Feridex [67]. High-performance T₂ contrast properties with an R₂ ~ 464 mM⁻¹ s⁻¹ (in 7.0 T magnetic field) have been observed in iron carbide (FeC_x) nanoparticles due to the high Ms value of 125 emu/g [69–71]. Roosbroeck et al. [72] used colloidal lithography to create phospholipid-functionlized Au/Ni₈₀Fe₂₀multi-layered structures with a nanodisc shape morphology and diameters ranging from 90 nm to 525 nm (Fig. 8a and b). Their magnetic examination showed an astonishingly low remanence, which is required to prevent particle agglomeration. The T₂ contrast properties of the nanostructures shown in Fig. 8 demonstrated higher R₂ values in a 9.4 T magnetic field if compared to superparamagnetic iron oxide nanoparticles.

Fig. 8.

MRI contrast properties of high-aspect ratio Ni and Fe nanowires [73,74]. (a) A schematic depicting the synthesis of metallic nanowires using the anode aluminum oxide (AAO) template-based pulsed electrodeposition method. (b) Top view SEM image of an AAO porous membrane (~40 nm pore diameter) utilized in the electrochemical fabrication of Ni/Au (or Fe/Au) nanowires; SEM cross-section of an AAO template filled with Ni/Au multilayer nanowires; and SEM image of Ni/Au nanowires after separated from AAO membrane. (c) Plot demonstrating how R₂ values for Fe nanowires with a diameter of 35 nm drop linearly with increasing length at 1.5 T [74]and 25 °C. The inset shows a T₂ color gradient map of Ni nanowires acquired in 3 T.

The use of nanowires made of 3d transition metals and their alloys with high aspect ratios have also been investigated as MRI contrast agents. Baobre-López et al. [73], for example, investigated the relaxivity features of a water stable colloidal dispersion of poly-acrylic acid (PAA)-covered Ni nanowires with prominet shape anisotropy. Pulsed electroplating multilayered Ni/Au nanowires within a nanoporous aluminum oxide template, followed by template removal and acidic etching of the Au layer, was used to create the ferromagnetic Ni nanowires. The relaxation times of Ni nanowires with mean diameters and lengths of 36 nm and 600 nm, respectively, were then measured at 37 °C using a relaxometer at a frequency of 60 MHz in two distinct magnetic fields, 1.41 T and 3.0 T. Fig. 8c shows an MRI scan of a phantom that demonstrates nanowires as efficient T₂ contrast agents. Fe and Fe–Au nanowires were also produced via template-assisted electrodeposition; these nanowires were then coated with dopamide-PEG (Dop-PEG) to bind other biological molecules and direct the nanowires to particular cell types [74]. Comparable to commercial Fe-oxide nanoparticles, the nanowires showed the highest R₂ = 77.1 mM⁻¹ s⁻¹. This study also revealed that the R₂ values in pristine Fe nanowires increased linearly as the length of the nanowire decreased. This trend contradicts the hypothesis that increasing shape anisotropy would also enhance R₂. Since longer iron nanowires have more mass, they sediment faster. This could explain why, despite having less surface area per particle, the shorter 0.5 μm nanowires had the highest R₂ = 14.7 mM⁻¹ s⁻¹. When the R₂ values for metallic nanowires were compared to the Fe₃O₄ nanorods discussed in the previous section, the latter showed a higher R₂ value despite having low saturation magnetization values. This difference is primarily due to the effective diameter, which is extremely large for nanowires synthesized via electrochemical processes. As a result, the metallic nanowires are expected to be in the echo limited regime, with a low R₂ value. To examine the real influence of shape anisotropy and M_S on MRI contrast properties, as well as create a high-performance MRI contrast agent based on metallic nanowires, the precise control of length and diameter in the sub-100 nm range is necessary.

2.4. Development of T₁-weighted MRI contrast agents

The T₁ contrast enhancement stems from the fluctuating unpaired electron spins of the nanoparticles coupling with the spins of water protons, resulting in energy transfer. Because of this rapid energy transfer, a shortened T₁ occurs in the vicinity of the MNPs, leading the proton spins to relax quickly towards their equilibrium magnetization after excitation. This in turn results in an enhanced signal around the nanoparticle. When compared to conventional Gd-based paramagnetic T₁ contrast agents, the superparamagnetic MNPs have a higher magnetic moment due to exchange coupling among the large number of electron spins per particle, which theoretically should result in a larger local T₁ effect. However, because most clinical MRI instruments use high magnetic fields (1.5 T or 3 T), the high external field suppresses local magnetic fluctuations, which results in a lower T₁ effect for MNPs than the T₂. Even though T₂ suppression of MNPs results in a significantly higher signal reduction, as discussed earlier, T₁ agents are preferred because they improve signal while also shortening image acquisition times. T₁ contrast enhancement has been reported for superparamagnetic MNPs with prominent surface-spin canting effect, including ultra-small-size Fe₃O₄ nanoparticles (<3 nm) [75,76], ultrathin Fe₃O₄ nanoplates [77], and Gd doped Fe₃O₄ nanoparticles (4.8 nm) [78,79]. The effects of particle size on T₁ contrast have been studied in 1.5–12 nm iron oxide NPs(see Fig. 9) [75]. The R₁ of 3 nm iron oxide NPs is 4.77 mM⁻¹ s⁻¹, which was found to be higher than the R₁ value of 12 nm NPs (2.37 mM⁻¹ s⁻¹). Furthermore, the 3 nm NPs have a relatively lower R₂/R₁ ratio of ~6 than the 12 nm NPs (R₂/R₁ ~ 24), demonstrating that the ultra-small sized NPs are superior for T₁-weighted MRI. Recently, amorphous-like ferric oxide nanoparticles synthesized at ambient temperature in aqueous solution and without the use of surfactants [80]. These particles showed superior T₁ MRI contrast effects and a higher spatial resolution than those obtained in 3 nm iron oxide nanoparticles.

Fig. 9.

T₁ contrast effect in ultra-small sized iron oxide nanoparticles. (a) T₁-weighted MR image of 3 nm iron oxide NPs. (b) 1/T₁ vs. Fe concentration plots of different size iron oxide NPs. T₁-weighted MR image of MCF-7 cell pellets incubated with (c) 3 nm and (d) 12 nm iron oxide nanoparticles, respectively [75].

Alternatively, MNPs can be used in ultra-low field (ULF) MRI to develop strong T₁-contrast effects [81–83]. The ULF has attracted considerable attention because of several potential benefits: (i) narrower instrumentation line widths, (ii) improved T₁ contrast, (iii) reduced susceptibility artifacts caused by metallic implants, and (iv) artifacts caused by the presence of air [53,84–86]. Since the magnetic moment is not saturated at ultra-low fields, we are operating in a regime where large low-frequency fluctuations of the MNPs can be tuned to the protons’ Larmor frequency, resulting in a large increase in R₁ values, even when the nanoparticle size is greater than 10 nm. Fig. 10 shows that the T₁ and T₂ relaxivities of various sized MNP solutions increase linearly as a function of Fe content [84]. A high R₁ value and a low R₂/R₁ ratio are key figures of merit for a positive contrast. A considerable signal increase at low Fe concentrations is overserved for 16 nm Fe₃O₄ and 18 nm Zn_0·3Fe_2·7O₄ nanoparticles. The R₁ = 615 mM⁻¹ s⁻¹ of the Zn_0·3Fe_2·7O₄ nanoparticles is especially high, with just a minor increase in the R₂/R₁ ratio of 2.7. In contrast, the commercial Gd-based T₁ agents show R₁ value up to 7 mM⁻¹ s⁻¹, with R₂/R₁ ~ 1.5. When the 16 nm Fe₃O₄ MNPs were scanned using a high field (3 T) with a standard inversion recovery sequence, the sample shows R₁ ~11 mM⁻¹ s⁻¹ and R₂ ~ 137 mM⁻¹ s⁻¹ (R₂/R₁ ratio of 12.5). This finding implies that MNPs are T₂-weighted contrast agents in high-field MRI, whereas they have a strong T₁ effect in ULF MRI. The highest R₁ value observed for 18 nm Zn_0·3Fe_2·7O₄ nanoparticles is due to the Brownian and Néel relaxations. As a result, the ULF MRI platform, together with tailored superparamagnetic iron oxide nanoparticles (SPIONs) as high visibility T₁ MRI contrast agents, offers new promises for safe, low-cost functional imaging. The effect of nanoparticles’ shape on T₁ has not yet been studied. However, Néel fluctuations can be engineered by changing the shape of the particles.

Fig. 10.

Size-dependent T₁ MR contrast properties. (a) The 0.13 mT T₁-MRI signals versus Fe concentration for different sizes of Fe₃O₄ and 18 nm ZnFe₂O₄ nanoparticles and (b) the corresponding T₁ relaxation rate versus Fe concentration. (c) The T₂ relaxation rate versus Fe concentration for different sizes of Fe₃O₄ and 18 nm ZnFe₂O₄ nanoparticles. (d) Simulated MRI signal intensity versus Fe concentration with corresponding R₁ and R₂ relaxivity of each nanoparticles with TR = 400 ms, TE = 27 ms [84].

2.5. Contrast agents for clinical applications

MRI is a powerful technique contributes immensely to clinical diagnosis and therapeutic monitoring. It can provide invaluable and intricate details of anatomical structures which arise from interactions between water protons and biological macromolecules present in the tissues. Nevertheless, relying solely on water protons, MRI images often encounter poor spatial resolution in identifying diseased regions of interest due to minimal differences in the dynamic constituents between healthy and unhealthy tissues. This issue was addressed by using exogenous materials which enhance the differential contrast of these regions earning the name “contrast agents”. Although the first whole-body MRI machine came into existence in 1977 [87], the use of contrast agents for image enhancement was not employed until the late 1980s. The development of paramagnetic gadolinium-based (Gd³⁺) chelates as MRI contrast agents was predominant in the initial years, as the presence of seven unpaired electrons generates a magnetic moment high enough to influence the relaxation times of neighboring water protons. As Gd³⁺ shows some level of toxicity to the body and poses health risks, it is complexed with organic ligands to improve its compatibility [88]. Since then, improving the biocompatibility of these compounds has also been a major focus of researchers [89–91]. It was later demonstrated that transition elements like copper (Cu²⁺), chromium (Cr³⁺), manganese (Mn²⁺) and iron (Fe³⁺) were also potential candidates for generating stable compounds as contrast agents for clinical use [92,93]. Thus, later works showed a paradigm shift from Gd³⁺ towards the exploration of superparamagnetic materials to serve as contrast enhancing agents [94–96]. To add to the advantages of these nanomaterials, they were found to be more biocompatible with minimal toxicity when compared to their Gd-based counterparts [97–100]. With two types of contrast mechanisms in play, conventionally, MRI contrast agents are divided into two categories. The contrast generated by the shortening of longitudinal relaxation (T₁) results in positive signal enhancement while shortening of transverse relaxation (T₂) results in negative enhancement, thus named as T₁-weighted and T₂-weighted MRI respectively.

Current clinical settings employ T₁-weighted paramagnetic Gd³⁺ chelates for positive enhancement while superparamagnetic ferrites mainly for T₂-weighted negative enhancement. At this point, attributed to a plethora of research advances, there exist a wide variety of nanomaterial-based contrast agents in clinics for uses in brain imaging [101,102], neurodegenerative disorders [103,104], cardiac and vascular imaging [105–107], cell mechanisms [108–110], cancer [111–114] and many other [115].

3. Magnetic nanoparticles as MPI tracers

3.1. Fundamental principle of MPI

Magnetic particle imaging (MPI) is a non-invasive imaging technique that has made its scientific breakthrough in recent years [116–118]. By performing direct imaging and the use of only magnetic fields, the efficiency of MPI is dependent on the tracers used, the most common being superparamagnetic nanoparticles. Under current developments, magnetic nanoparticles provide a three-dimensional visualization of their distribution in space. The images that are produced through MPI have outstanding contrast and sensitivity, when compared to other imaging techniques including MRI and optical imaging [119]. The spatial distribution of the magnetic tracer material, which results in high contrast and resolution, is measured by collecting data of the magnetization change of the tracers in time-varying external magnetic fields. The magnetization of MNPs can be described by the Langevin magnetization which states that they are constantly in thermal equilibrium (net zero magnetization state) in the absence of an external magnetic field, and the magnetic moment always tends to align itself with the externally applied magnetic fields. Superparamagnetic materials are best suited for being MPI tracers due to their non-linear magnetization behavior under externally applied fields [120]. When the alternating external magnetic field is applied, the magnetization of MNPs grows rapidly in response to the increasing magnetic field, then plateaus after saturation is reached (Fig. 11a) [121].

Fig. 11.

A representation of MPI signal generation through two techniques: X-space and harmonic-space. X-space and harmonic space graphics provide tracer selection, characterization, and transmit/receive coils. The magnetics/field free point (FFP) are where spatial localization/excitation and signal generation occurs, from that location image is reconstructed. [121].

After being exposed to an oscillating magnetic field, the magnetic moments of the MNPs that are not magnetically saturated will change with the direction of the field while those of saturated particles will not. The MNPs will relax as a result of Brownian (physical) and Néel (internal) rotations (Fig. 11b) [119,122]. The superparamagnetic relaxation is appropriately described by the Néel-Brown theory according to equation (7) [123,124]:

$${\tau }_N={\tau }_{0}exp\left[\frac{E_a}{k_{B}T}\right]$$ (7)

where the flipping angle of the nanoparticles is determined by its anisotropy energy (E_a = KV), τ₀ is an attempt frequency factor and its value ranges from 10⁻⁹ to 10⁻¹³ s, and k_B is Boltzmann constant. The relaxation time exponentially increases with the increase of MNPs size. The Brownian relaxation time τ_B is represented by equation (8) [125].

$${\tau }_B=\frac{3\eta V_H}{k_{B}T}$$ (8)

where T, V_H, and η represent, temperature, hydrodynamic volume of the MNPs and viscosity of the medium, respectively. Rosensweig proposed [123,124] that the Néel and Brownian relaxations take place in parallel, with effective relaxation angular frequency being proportional to effective relaxation duration, given by 1/τ_eff = 1/τ_N + 1/τ_B. Because the relaxation of MNPs is directly related to their size and shape, changing their size and shape can be an effective way to control the MPI sensitivity and resolution.

To utilize this superparamagnetic property towards MPI, the machine is designed with three main parts: (a) selection field, (b) drive field, and (c) receiving coil. The selection field is a static gradient field of sufficient magnitude generated across coils of the same polarity facing each other. Within the Maxwell configuration, this selection field generates a symmetrical point at which net magnetic field is zero and is identified as the field free point (FFP) (Fig. 11a). In the FFP, the magnetization of the particles is free to follow the excitation field, but the particles outside of the FFP are not able to move with the excitation field because they are saturated, which does not produce signals for MPI. The drive field is the alternating magnetic field which directly affects the magnetic moments of the tracer MNPs in question. When a MNP having a non-linear magnetization (like ferrimagnetic or ferromagnetic particles) is placed within this FFP in conjunction with an externally applied oscillating magnetic field, a potential difference is generated. This potential difference is translated by the receiving coils into an observable MPI signal.

To establish image reconstruction, MPI signal detection must occur which directly depends on the relation between magnetization and external magnetic fields being nonlinear. In a linear relation, the signal cannot be produced because the excitation field overlaps the signals produced by the MNPs. In MPI image reconstruction, harmonic-space and X-space are the techniques being used to convert signals into images. Harmonics is a quantitative 3D scan of the tracer distribution that is found for each FFP recorded, where the particle concentration and system performance both affect the observed signal [126]. Harmonics uses two sub-techniques: measurement-based and model-based techniques. The former technique requires multiple measurements of the nanoparticle positions to achieve high spatial resolution, which necessitates lengthy measurement durations and big data sets. Model-based system function uses computations in mathematics to simulate a variety of factors, including the magnetic field, particle magnetization, and the voltage induced in the receiving coil. In X-space, the point spread function (PSF) of the system and the MNPs spatial distribution are convoluted to create the image. The MPI signals that emerged are temporal scans as the receiving coils can only detect changes in magnetization. The image is then recreated in two steps: velocity adjustment and then immediate gridding of the signal to the location of the FFP [127]. Nonetheless, both work in similar ways, by adding spatial encoding which helps to detect the origin of signals, this overall field is the selection field. An amplitude generated from the oscillating drive field leads to a magnetization response that has higher harmonics than the drive fields. The generated voltage then reaches the receiving coils (Fig. 11a) [128], creating an MPI signal. Furthermore, the data obtained from the harmonics is applied for MPI usage.

For 1D spatial encoding, the signals from harmonics must be encoded by using a selection field, having a non-zero gradient in the x-direction, over the entirety of the field of view. It can only cross zero at one signal point which we refer to as the FFP, all other areas away from the FFP are saturated by the applied selection field. The FFP travels over each spatial position at different times, and each of these positions can be addressed by the characteristics of spectral response. For particles that are under the influence of both harmonics and a selection field, the reconstruction process is done by calculating the sum over the weighted harmonics with Chebyshev polynomials [128]. However, for 2D/3D spatial encoding, it is possible for it to be done by using a 3D selection field with a 1D drive field, to look at 2D/3D imaging. By using a harmonic drive field and a maxwell coil setup, the total field can be approximated. Only near the FFP, there can be field change that is rapid enough for a stimulation of a particle response that generates a frequency high enough to be detected (Fig. 12).

Fig. 12.

Spectral response within the “system matrix” which provide tracers at every location in the FOV (Field of View)- Harmonic-space MPI [121].

3.2. Development of MPI nanotracers

Unlike MRI, the MPI detects the tracer directly and provides no information of the surrounding environment, hence, the quality and performance of the tracer is critical in generating quality images. This necessitates research on suitable MPI tracers tailored towards MPI physics and their thorough evaluation. The differences in basic imaging principles between these two techniques lead to the difference in approach while developing optimal tracers. The performance of MPI tracers heavily relies on the instruments’ sensitivity and spatial resolution but their engineering can help enhance these parameters to some extent. MPI performance of tracers’ work based on their field-dependent first-order derivative of magnetization (dM/dH) [37]. By maximizing the tracer’s magnitude for spatial resolution (dM/dH), its signal sensitivity can be improved. Tracers with high signal response can be functional in visualizing morphological details of tissues and diagnosis of diseases. By altering the size, shape, composition, surface, and aggregation state in MNPs, one can enhance the sensitivity and the signal levels [129].

For magnetic nanomaterials, magnetization is strongly dependent on its size, which is critical for the development of high-performance MPI tracers. Since larger MNPs have greater Ms, precise control over its size can aid to improve the MPI signal and resolution. In a study, iron oxide nanoparticles were synthesized into various sizes by varying the concentration of reactants and reaction temperature [130]. The measured Ms of their synthesized nanoparticles increased from ~36 emu/g to 60 emu/g when their size was increased from 8 nm to 18 nm. At the same concentration, the observed MPI signal intensity was found to be increased with increasing iron oxide nanoparticles within a certain range. When compared with commercially available VivoTrax^®, the 8 nm and 18 nm iron oxide nanoparticles showed an MPI signal of 94.73 and 134.2 respectively, which is 1.89- and 2.68-times larger. However, the shift from superparamagnetic to ferromagnetic behavior is a limiting factor in increasing the MNPs size over the threshold. Although the larger-sized MNPs and ferrite nanoparticles exhibit high M_S values, the developed ferromagnetic properties (mostly with onset of the remanence magnetization) always lower MPI signals.

Thus, synthesizing particles within the size range exhibiting superparamagnetic behavior is advised. To this end, another study attempted to determine the optimum magnetic material core size by magnetic particle spectroscopy (MPS) measurements and simulations based on Langevin equations [131]. The obtained results revealed that the MPI performance can be increased by increasing the core size up to MNPs size of 28 nm. Above 28 nm, the interparticle interactions make the relaxation effects difficult because of particle agglomeration and resulted in decreased signal quality. Several experimental and theoretical investigations have been performed to further evaluate size-dependence of MPI performances [129,132–134]. These studies revealed that the MPI signal is very sensitive to both NP size and environment. The dM/dH analyses revealed that Néel relaxation is the dominant mechanism determining MPI response if MNPs size is below 20 nm [135]. Whereas the larger-sized MNPs show hysteretic reversal if the applied field amplitude overcomes the coercivity. Concluding from these studies on the subject, it can be stated that the core size of the magnetic nanoparticles has a relational role in determining MPI performances, in sizes ranging from 10 to 30 nm. Readers interested in learning more about the requirements for designing tracers for MPI and MPI clinical applications should read the referred articles [7,136–139].

Alternatively, the MPI signal intensity and resolution can also be modulated by considering MNPs’ shape into account as magnetic properties are strongly influenced by the shape anisotropy [140,141]. When compared with spherical morphology, the anisotropic MNPs generally have lower surface anisotropies and less spin canting effects, which are favorable to obtain high M_S value. Despite there being enough available evidence that shape anisotropy can directly affect the magnetic properties of MNPs, there are not many studies which have utilized this aspect for preliminary or pre-clinical MPI. To investigate this direction, Wang et al. synthesized cubic-shaped iron oxide nanoparticles of sizes 17–46 nm via a thermal decomposition route [140]. Due to the defined surface facets, the cubic NPs possess less disordered surface spins, resulting in significantly larger M_S than that of the spherical counterparts. The enhanced M_S and lower effective magnetic anisotropy led to high sensitivity and resolution of MPI. A significantly stronger MPI signal is observed in 22 nm cube (Fig. 13). When compared with the commercial VivoTrax^®, the 22 nm-sized cubes generated ~4.1-fold higher MPI signal intensity and exhibited efficient cellular absorption. Whereas in the larger-sized cubes, magnetic relaxation was low due to the ferromagnetic properties which develop due to large effective magnetic anisotropy and in turn led to poor MPI resolutions.

Fig. 13.

Size and shape effects on MPI performance of MNPs. (a–c) TEM micrographs of iron oxide nanocubes (CIONs) of sizes 22, 26, and 46 nm and (d) iron oxide sphere (SIONs) of 17 nm (scale bar = 20 nm). (e) The corresponding MNPs size distribution in aqueous dispersion, and (f) MPI performance with Fe concentration of 1.7 mM. (g) MPI linear scanning spectra of 22 nm cubes and VivoTrax with Fe concentration of 5 mM [140].

Transition metal ferrites are favored by the workers for biomedical applications for their long-established biocompatible properties and tailorable physico-chemical properties. Propelled by the requirement of enhanced magnetic properties for MPI, other superparamagnetic nanomaterials have also been explored like FeCo nanoalloys. FeCo is known to have higher saturation magnetization among the binary metal alloys which can be tailored by varying its elemental composition [142]. In transition metals (like Fe, Co, and Ni) the exchange coupling is positive, favoring the ferromagnetic ordering of spins [46]. However, a “magnetic dead layer” is also generated as discussed in section 2.3.1. If the particle size decreases, a majority of spins are populated at the surface of nanoparticle. Furthermore, if alloyed with another similar metal, the lattice parameter of the alloy can decreases/increase depending on the atomic radius of metal, resulting in overlap of atomic orbitals, which in turn can reduces the atomic dipole moment and exchange coupling. These effects synchronize together and disorder the surface spins to generate the magnetic dead layer. As the name suggests, the presence of this layer reduces the effective magnetic response of the alloy which can be limited by increasing the nanoparticle size. This further solidifies the previous findings of size threshold requirements for optimum magnetic properties towards MPI tracer materials. These binary nanoalloys have also been known to show toxicity to a certain degree, which can be overcome by the use of carbon-based shell in the form of graphene or carbides. On one hand, the presence of carbon helps in improving their biocompatibility but in another way, it also provides a platform for anchoring various surface modifications. It has already been explored as a viable contrast agent for MRI [143]. Encouraged by these results, Song et al. [144] explored this FeCo nanoparticles as a multimodal “smart” platform for therapeutics and imaging with MPI being one of the imaging modalities. MPI signal increased with an increase in core size up to 10 nm but decreased with further increase in size. The study also revealed that the ratio of the transition metals in their nanoparticles also plays a role in its MPI performance. The FeCo nanoparticles with metal ratio of Fe/Co (1:0.87) and average particle size around 10 nm showed high Ms of 192 emu/g. More importantly, the MPI signal intensity of was found to be nearly 15-times and 6-times higher than that of Feraheme and VivoTrax, respectively (see Fig. 14a–d). Furthermore, the observed low detection limit of FeCo nanoparticles up to 5 ng and low signal to noise ratio (SNR) of ~1.3 determines its efficacy for ultra-sensitive MPI tracers.

Fig. 14.

The effects of saturation magnetization and magnetic alignment on the MPI performance. 2D projection MPI images of: (a) FeCo nanoparticles, Feraheme and VivoTrax with 800 ng of core materials in PCR tubes and (b) FeCo nanoparticles (5 ng of core) in PCR tube after background subtraction 2D average MPI images (scanned 25 times) [144]. (c, d) The corresponding MPI point-of-spread function of the images. (e) Schematic representation of MNP-chain prepared in magnetic field and random sample. (f) Improved point-spread-function for x-space reconstruction. Comparison of MNP chains to VivoTrax shows a ~40-fold improvement in signal intensity and ~10-fold improvement in spatial resolution [141].

As MPI is related to the nature of magnetization loop, particularly the steeper slope on the magnetization-applied field curve, an alternative approach to improve the sensitivity, spatial resolution, and SNR can be the alignment of MNPs into ordered patterns prior to the scanning. It has been found that the interparticle interactions (mostly dipolar and exchange interactions) among the MNPs within the assembly gives rise to local magnetic anisotropy which can improve magnetic remanence and coercivity. Moreover, just like how ferromagnetic order and properties depend on the crystal structure of a material, the dipolar-ferromagnetic order and collective properties can be emerged from the geometry of an assembly of interacting nanoparticles [145]. Further tuning of the dipolar interaction and magnetic ordering in the superlattice sites can result from the alignment of nanoparticles along the so-called ‘easy axis’ by using an external magnetic field [146,147]. In recent work, we have synthesized face-centered cubic (fcc) superlattice structure of FeCo/-CoFe₂O₄ core/shell magnetic nanoparticles in an applied magnetic field. The assemblies exhibit a strong enhancement of magnetic susceptibility and remanent magnetization compared to the as-prepared nanoparticles in random and assemblies prepared without application of magnetic field. The enhanced collective magnetic properties are apparently associated with the alignment of magnetic moments of the nanoparticles in the superlattice sites and the strong intra-superlattice interactions. However, there has not been much research performed in this direction for designing high-performance MPI tracer. Nevertheless, Conolly et al. [141], studied the MPI characteristics of oriented MNP assemblies (i.e. MNP-chain) and compared it with the MNP in random. The experimental data shows an intense inductive signal in MNP-chain as shown in Fig. 14e and f. In addition, the point-spread-function measurements also revealed that the MNP-chains have around 40-fold higher peak intensity and a 10-fold narrower peak compared to VivoTrax.

Although several MNPs formulations have been reported in showing promising MPI signal intensity and resolution, there are still several major challenges that need to be overcome if clinical MPI is to become a reality. Firstly, there lacks a general design principle to guide experimentalists to develop MNPs with the maximized sensitivity and resolution relevant to clinical applications, and secondly, recent experiments indicate that there are significant effects of MNP surface, environment, and dipolar coupling, seemingly inconsistent with classical magnetization reversal model [129]. Addressing these major roadblocks will accelerate the adoption of MPI for clinical applications, especially for applications of theranostic purpose, which requires more surface modification for targeting or therapeutic moieties.

3.3. MPI tracers for clinical applications

In recent years, MPI has been garnering significant attention. However, being introduced for just short of two decades [148], the technique is still infantile, and researchers are working tirelessly to translate this modality to clinical settings. Once scaled for human imaging, it has the ability to have a significant role in diagnostic imaging [29,149]. With its unique characteristics from high temporal resolution and sensitivity, it can be used in many areas such as neurological, vascular, cardiovascular, and cancer diagnosis [150]. The MPI tracers also give good insight into the future, because SPIONs were approved by FDA for liver imaging, there seems to be no real concern for superparamagnetic nanotracer element which can be administrated to a patient within certain dosages over the course of treatment.

When compared to the imaging techniques that are readily available now such as CT, MRI, and ultrasound, MPI offers the lack of ionizing radiation and with better tracers the possibility of less repetitive and greater follow-up measurements. This allows the source to be more biocompatible, because MPI has other capabilities, that can allow for rapid 3D imaging via special rotated instruments. MPI tracers can also be used to visualize vessels in 3D with the help of intravenous injection and electrophysiologic interventions [151,152]. The tracers are also suitable for long-term cell tracking (months) which can be beneficial in cases involving cancer and regeneration therapy. This is a significant difference in comparison to PET and SPECT tracers which have half-lives ranging from minutes to days [153]. Additionally, this diagnostic imaging has no depth restriction and no tissue auto-luminescence interference. Most importantly, it has both characteristic advantages found in MRI and PET techniques such as high spatial resolution and high detection sensitivity. MPI is very promising and can be a root to how medicine and imaging can evolve, from cell labelling to being radiation free, MPI can have immediate and considerable effects in various areas as schematically illustrated in Fig. 15 [154].

Fig. 15.

Biomedical use with MPI Tracers; Stem, Cancer, Islet Cells can be tracked through cellular tracking, In vivo imaging/multi-color imaging- (applications for cancer therapy and addition of other imaging sources such as MRI, PA, Optical imaging.

MPI-based cell tracking was among the earliest applications of the technique. Cell tracking is a valuable tool in cell therapy for various diseases including type 1 diabetes, neurodegenerative, and cancer. Despite the enormous potential of cell therapies, their clinical results have been inconsistent because of differences in cell source, preparation, and route of administration/implantation methodology. This is the point where image-guided therapy has an edge and MPI has been reported to play a major part in. [155–158]. One of the cell therapy and transplant approaches is utilized in a challenging niche of neural cells. The nervous system is a closed-loop and unforgiving network of cells. It does not accept foreign grafts as easily as other locations in the body. Until now, successful graft and integration of neural cells has been confirmed through invasive techniques like histology. In order to track the fate of these grafts, MPI has been shown as a reliable non-invasive modality [159,160]. Fig. 16 depicts one such example of tagging macrophages with nanoparticles and imaging using MPI for their retention in the brain after the mouse had a stroke. This pilot study aimed to understand their retention times in the brain and saw a significant difference between 48 h, 72 h, and 96 h images. Successful use of MPI-based image monitoring would further aid in overcoming the limitations posed in the treatment and monitoring of neurological ailments.

Fig. 16.

Retention times of SPION-labeled macrophages in the brain of a stroke mouse. 1–2 × 10⁶ nanoparticle (VivoTrax)-labeled mouse macrophages (Raw 264.7) were administered to BALB/c mice through tail veins 24 h after stroke. 2D-MPI was performed using a MOMENTUM MPI scanner at 48, 72, and 96 h poststroke with scanning parameters: FOV = 4 cm, 55 projections, best image quality, and default scan mode. Anatomic images were collected on the eXplore CT-120 microCT. For in vivo iron oxide quantification, a fiducial marker containing a known concentration of tracer was placed beside the animal. [160].

One of the diseases which utilizes the promise of cell therapy is type 1 diabetes. It is an autoimmune disorder in which the immune system tends to react against the insulin-producing beta cells of the pancreas leading to their destruction. This in turn incapacitates the blood glucose regulation. To combat this issue, isolated islets and islet organoids generated from the stem cells have been successfully transplanted to restore the body’s glucose metabolism capability. Monitoring these transplanted islets thus plays a critical role in minimizing loss of function. MPI has been an effective modality for monitoring as well as quantification towards a practical therapeutic solution (Fig. 17) [161].

Fig. 17.

3D MPI imaging confirms the transplant location for both liver as well as the kidney capsule. Longitudinal quantification of iron content from intensity profile can also be performed in coronal and sagittal views (left - coronal; right - sagittal). (a) MPI signals from the islets transplanted under the kidney capsule (green arrows), (b) MPI signals from the islets transplanted in the liver (red arrows), and (c) Control. No signal was seen in the control. Scale Bar = μgFe/mm² [161].

As the technique is still in its nascent stage, researchers are also looking into the improvement of the digital image processing associated with MPI. Taking advantage of the linear relationship between the MPI signal and its corresponding nanoparticle content, various approaches are being explored for image-based quantification. These approaches would also be helpful to differentiate between signal of interest and the background when working with low amount of nanotracers. Various algorithms are currently under development which involve various levels of artificial intelligence and unsupervised machine learning [157, 162].

Another widespread disease which utilizes MPI in its diagnosis and imaging is cancer. Tumor detection, therapeutic monitoring, and tumor regression is attempted via MPI. The treatment strategy of cancer hugely relies on the stage of the diagnosed cancer. Thus, techniques promising high sensitivity and resolution become critical to an early detection of the disease [116,163–166]. Not only early detection, MPI has been reported for successful monitoring of the delivery of anti-cancer drugs to the target site [167]. This is valuable in combined monitoring, quantification and tumor regression in mouse models. MPI has also been reported in visualizing the step-by-step treatment of the pro-apoptotic drug delivery approaches (Fig. 18) [168].

Fig. 18.

(a) Near-infrared fluorescence (NIRF) images of the mice with tumor xenografts. The untreated group was administered with PBS (top) while etoposide and CTX was used for the treated group (bottom). The groups were injected with AF647-AnxV-SPIO. (b) Intensity profile of the untreated and treated group. A significant difference was observed as early as 4 h post-injection (n = 3, P < 0.05). (c), (d) In vivo (top) and ex vivo (bottom) assessment of the uptake of MPI nanotracer by tumor xenografts. No MPI signal was detected in vivo from the tumor, while an obvious difference was detected in the tumor xenografts excised from mice (n = 3, P < 0.05) which may be due to the shadowing of fainter signal from tumor when compared to signal from liver [168].

Related to cancer therapeutics, magnetic hyperthermia has also found its application. As the MNPs can generate both MPI signals and heat simultaneously when exposed to the AC magnetic field, MPI can be used as a precise modality for imaging-guided therapy on a localized tumor area [130,152,169–171]. MPI is also expanding its reach into the application of image-guided cancer immunotherapy. Cancer being the prime example, a non-invasive imaging modality recognizing immune checkpoint biomarkers is advantageous in detection and evaluation of therapeutic efficacy. Its ability to track and image cells of immune system has facilitated the guidance of immunotherapy throughout its course [172–174]. The prediction of the body’s response to these therapies and their underlying mechanisms remains a challenge. As the MPI is capable of imaging the nanotracers independent of the tissue depth gives it an upper hand when compared to other imaging techniques. Thus, MPI-guided monitoring has found use in predicting pre-clinical and clinical responses to immunotherapy to overcome this limitation (Fig. 19) [175,176]. Most of the recent works involve the immune checkpoint molecules as the target of interest, which are defined as ligand-receptor pairs that exert inhibitory or stimulatory effects on immune responses. One example is that T lymphocyte activation can be suppressed via inhibitory signaling pathways governed by checkpoint proteins including programmed cell death protein 1 (PD-1) and its cognate ligands, programmed cell death ligand 1 (PD-L1) [177]. When bound to these ligands, PD-1 works to regulate the adaptive immune response by initiating immunosuppressive signals leading to the induction of apoptosis and reduced cell proliferation [178,179]. In recent years, immune checkpoint therapy has revolutionized the field of oncology. Many cancer cells possess genetic and epigenetic irregularities allowing them to utilize immune checkpoints to promote survival from immune surveillance. Combining antibodies of such immune checkpoint targets with MPI-active magnetic nanoparticles is seen to aid in visualizing PD-L1 expression in vivo CT26 tumor model. The levels of PD-L1 are crucial in designing a precision therapeutic approach and the upregulated PD-L1 expression can also be associated with MPI signal enhancement which in turn is quantifiable. These results further add to the evidence suggesting great futuristic clinical potential for MPI [175, 176].

Fig. 19.

Magnetic particle imaging (MPI) study of CT26 tumors using anti-PD-L1 antibody (aPDL1)-conjugated magnetic fluorescent hybrid nanoparticles (MFNPs-aPDL1). (a) Time-dependent in vivo MPI of CT26 tumors treated with MFNPs-IgG and MFNPs-aPDL1, respectively. (b) Quantification of MPI signal generated by the treated groups at pre-determined time intervals. (c) Ex vivo MPI of harvested tumors with different treatments. (d) Prussian blue staining of tumor tissues treated with different imaging probes depicting the presence of iron in the tissue. The protocol stains the iron containing nanoparticles as dark blue. Scale bar = 100 μm. (e) MPI and FMI of CT26-tumor-bearing mice at 24 h post-injection with MFNPs-aPDL1 on both right and left sides. (f) Comparison of MPI signal generated from panel (e). (g) Comparison of fluorescence intensity calculated from panel (e) (*P < 0.05, ***P < 0.001, ns P > 0.05) [175].

Another promising application of MPI is seen in cardiovascular imaging. The MNPs presence in the blood stream, are detected without any surrounding information making the identification/detection of cardiac ailments relatively easy [180–185]. With the development of human sized MPI scanners, the application of MPI can be further expanded into a variety of clinical applications and needs more investigation [186]. As the research towards MPI progresses, for scanners and tracers, it can be integrated in clinics equally well as MRI is. New technologies including artificial intelligence will also play a great role for imaging quantification and analysis for clinical applications [162,187,188].

4. Concluding remarks

Being similar in underlying physics and mechanisms, the performances of both MRI and MPI revolve around the magnetic quality of the nanomaterial used. MRI has been in active clinical use for more than 50 years now and workers are still looking for ways to enhance its performance. Primarily based on iron oxide nanomaterials, a variety of contrast agents have been studied and commercialized.

Although still in pre-clinical stages, MPI shows encouraging results that back up its ability to be a biomedical imaging modality that can aid in the needs of treatments and diagnosis of diseases. MPI being a non-invasive technique adds to the benefits of how it can be used, from cellular tracking to imaging-guided cancer therapy via hyperthermia/chemotherapy. It is also shown that a similar concentration of tracer element can have higher resolution than current leading commercial contrast tracers, which reduces the need for additional contrast elements to be applied in a diagnosis or treatment setting. Exploring various parameters in the development of tracer platforms can help us utilize MPI to its full potential and provide great insight into the future of this imaging technique. As for future directions, the sensitivity of MPI would be a challenging parameter which requires optimization for in vivo applications. MPI-based diagnoses can also help to streamline the treatment process, in combination with other therapeutic modalities, thereby solidifying its place in precision medicine.

Data availability

Data will be made available on request.