Recent advances in engineering iron oxide nanoparticles for effective magnetic resonance imaging.

Iron oxide nanoparticle (IONP) with unique magnetic property and high biocompatibility have been widely used as magnetic resonance imaging (MRI) contrast agent (CA) for long time. However, a review which comprehensively summarizes the recent development of IONP as traditional T 2 CA and its new application for different modality of MRI, such as T 1 imaging, simultaneous T 2/T 1 or MRI/other imaging modality, and as environment responsive CA is rare. This review starts with an investigation of direction on the development of high-performance MRI CA in both T 2 and T 1 modal based on quantum mechanical outer sphere and Solomon-Bloembergen-Morgan (SBM) theory. Recent rational attempts to increase the MRI contrast of IONP by adjusting the key parameters, including magnetization, size, effective radius, inhomogeneity of surrounding generated magnetic field, crystal phase, coordination number of water, electronic relaxation time, and surface modification are summarized. Besides the strategies to improve r 2 or r 1 values, strategies to increase the in vivo contrast efficiency of IONP have been reviewed from three different aspects, those are introducing second imaging modality to increase the imaging accuracy, endowing IONP with environment response capacity to elevate the signal difference between lesion and normal tissue, and optimizing the interface structure to improve the accumulation amount of IONP in lesion. This detailed review provides a deep understanding of recent researches on the development of high-performance IONP based MRI CAs. It is hoped to trigger deep thinking for design of next generation MRI CAs for early and accurate diagnosis.

Introduction

Various biomedical nanomaterials have been developed to act as agents to fulfill cell labeling [[1], [2], [3], [4], [5]] and separation [[6], [7], [8], [9]], biological sensing [[10], [11], [12], [13], [14]], and disease diagnosis/therapy [[15], [16], [17], [18]]. Among all biomedical nanomaterials, iron oxide nanoparticle (IONP) has been intensively investigated due to their unique magnetic property and high biocompatibility [[19], [20], [21], [22], [23]]. IONP based commercial MRI contrast agent (CA) (e.g. Resovist, Feridex), which shorten the transverse relaxation (T2) of proton, have been approved by the U.S. Food and Drug Administration (FDA) for clinical diagnosis. However, commercial available IONP based MRI CA show low MRI contrast ability due to low crystallinity or poor surface modification, which lead to reduced sensitivity and accuracy in MRI contrast imaging and limited application in early and accurate diagnosis [[24], [25], [26]]. Over the past years, tremendous efforts have been made to improve the relaxivity of IONP by engineering crystal and surface structure [[27], [28], [29]]. On the basis of relaxivity-structure assessment, IONP with high T2 relaxivity were discovered to show several critical characteristics, including high saturated magnetization, large effective size, high inhomogeneous magnetic field, and suitable surface modification [[30], [31], [32]]. According to the classical outer-sphere theory, saturated magnetization and large effective size are responsible for the field perturbation area for the outersphere protons and proportional to the T2 relaxivity of MRI CA [[33], [34], [35]]. Thus, synthesis of IONP with high saturated magnetization and large size have firstly been investigated to increase the T2 relaxivity of IONP. Additionally, high inhomogeneous magnetic field was founded to accelerate the dephasing process of protons around IONP, which significantly increased its T2 relaxivity. With the development of nanoparticle synthesis, IONP with unique morphology has been developed through morphology controllable synthesis [[36], [37], [38]]. These IONP show large effective size and high inhomogeneous of surrounding magnetic field, causing the elevation of T2 relaxivity. In addition, surface coating layer was discovered to affect the magnetic property of IONP and diffusion of surrounding proton [[39], [40], [41]]. Therefore, optimizing surface ligand of IONP have been conducted to form proper surface coating layer and increase T2 relaxivity of IONP.

IONP based traditional T2 CAs exert darken contrast in the region where they reside, which may result in false positive detection caused by the confusion between lesion and bleeding, calcification, or metal deposition. This limitation encourages researchers to develop IONP based longitudinal relaxation (T1) CA, which exerts bright contrast to distinguish tissue [[42], [43], [44]]. However, traditional IONP exhibits high r2/r1 ratio, hindering its application as a T1 CA. Based on the Solomon-Bloembergen-Morgan (SBM) theory, T1 contrast ability is highly depend on the chemical exchange efficiency between magnetic ions and protons. Currently, three key parameters have been unearthed to alter the chemical exchange efficiency and determine T1 contrast capacity of IONP. First, high saturated magnetization causes high r2/r1 ratio and impairs T1 contrast imaging. Decreasing the size and crystallinity have been proven to successfully reduce saturated magnetization and r2/r1 ratio of IONP [[45], [46], [47]]. Second, coordination number of water (q) on the surface of IONP is another important parameter to tune. Since coordination number of iron ion is constant, the method to optimize the q is directly increasing the number of iron ions on the surface or introducing other magnetic ions with large amounts of unpaired electrons. Thanks to the controllable synthesis and development of new generation of surface ligands, various IONP with high surface-to-volume ratio, specific composition, and optimal surface coating have been developed to show high T1 contrast and exhibit potential to achieve accurate tumor contrast imaging [[48], [49], [50]]. Recently, an undesirable structure, that is the presence of undesirable Fe (II) ion on the surface of IONP, have been noticed. Due to its short electron relaxation time (τs), Fe (II) ion is unfavorable to achieve efficient relaxation enhancement and severely reduces the T1 contrast. Strategy to replace the Fe (II) on IONP surface with other magnetic ions with long τs have been proved to significantly increase T1 contrast for high-performance T1 contrast imaging [51].

Apart from fabricating IONP with high T2 or T1 contrast ability, high-performance IONP based MRI CA could also be fulfilled by increasing its contrast efficiency. Compared to the IONP in the form of single T2 negative or T1 positive, introducing T1 moiety or other imaging contrast moiety into IONP could yield complementary diagnostic information to increase the contrast efficiency [[52], [53], [54]]. Recently, integration of IONP and gadolinium species have been developed as dual-modal CA (DMCA). A facile strategy is coating IONP with gadolinium based nanostructure or gadolinium complex [[55], [56], [57]]. However, the spin alignment on T1 contrast shell is opposite to the local magnetic field generated by IONP core, which reduces its T1 contrast and limits its application in accurate T1/T2 dual modal contrast imaging. Interestingly, this phenomenon is highly dependent on the distance between IONP and T1 contrast moiety [58,59]. Introducing SiO2 isolating layer to separate T1 domain and IONP or forming Janus structure have been applied to increase the distance between IONP and T1 moiety to develop high-performance IONP based T1/T2 DMCA [60]. Lately, doping manganese or gadolinium ions into IONP have been found to be another effective method to increase T1 contrast of DMCA [[61], [62], [63]]. Apart from development of IONP with dual-modal MRI contrast, direct conjugation of secondary even tertiary imaging moiety with IONP is another effective method to improve its contrast efficiency. The main strategy to fabricate IONP based multi-modal imaging CA is surface conjugating IONP with secondary imaging components, such as single-photon emission computed tomography (SPECT) [64,65], positron emission tomography (PET) [[66], [67], [68]], computed tomography (CT) [[69], [70], [71]], fluorescence (FL) imaging [[72], [73], [74]], ultrasound (US) imaging [[75], [76], [77]], and photoacoustic (PA) imaging [[78], [79], [80], [81]]. Based on the complementary signals, the regions of interest, including tumor, cardiovascular disease, and cell tracking could be rapidly and accurately detected.

For disease diagnosis, the contrast efficiency of MRI CA depends on the signal difference between lesion and adjacent healthy tissue. Therefore, there is a growth of interest on developing internal (lesion microenvironment) and external (specific physical condition) responsive MRI CAs [[82], [83], [84], [85], [86], [87]]. Environment responsive MRI CA could respond to specific characters and exhibit triggered or switchable signals for highly specific and precise MRI images [[88], [89], [90], [91], [92], [93], [94]]. IONP based responsive MRI CA could be divided into three different manners, active, recovery, and switchable. Compared to the traditional MRI CA with fixed contrast capacity, IONP based responsive MRI CA with signal active or recovery capacity could enhance T1 or T2 signal in response to a specific stimulus of lesion. Recently, IONP based environment responsive MRI CA with modal switchable capacity have been developed to further increase the signal difference between lesion and normal tissue. Similar to the traditional MRI contrast imaging, there are two modals of switchable tumor microenvironment (TEM)-responsive MRI contrast imaging: T2 contrast switch to T1 contrast (mode I) and T1 contrast switch to T2 contrast (mode II). The strategy to develop modal switchable MRI CA is based on the fact that formation of magnetic nanocluster or nanoaggregation could improve T2 relaxivity of ultrasmall IONP. Normally, ultrasmall IONP are considered as typical T1 CA when it is monodisperse. With the specific surface modification, ultrasmall IONP could achieve the status change between cluster and monodisperse. These structural changes result in the decrease or increase of T2/T1 contrast, which endow CA to achieve modal switchable contrast imaging.

The signal difference between normal tissue and lesion is also directly influenced by the amount of IONP in lesion, which is mainly determined by its in vivo behavior. When IONP enter the physiological environment, serum proteins rapidly adsorb on its surface and form protein corona, resulting in the decrease of blood circulation time and accumulation in lesion [[95], [96], [97], [98]]. IONP Surface modification of nanoparticles with specific ligands, such as polyethylene glycol (PEG) or zwitterionic small molecules, can effectively reduce the non-specific protein adsorption and elevate the accumulation amount of IONP in lesion [[99], [100], [101], [102], [103], [104], [105], [106]]. In addition, the accumulation amount of IONP in lesion could be further increased by introducing targeting motif,e.g. antibodies, proteins, peptides, and aptamers to achieve targeted MR contrast imaging with enhanced contrast efficiency [27,[107], [108], [109], [110], [111]].

During the past decades, a number of reviews have been published to summarize the development of IONP as MRI CA. Yet a review comprehensively discussing high-performance IONP based MRI CAs with high T2 relaxivity, high T1 relaxivity, environment responsive contrast ability, multi-modal contrast capacity, and optimized in vivo behavior in recent progress is needed. This review discuss the parameters to affect IONP based MRI CA from contrast ability of IONP and contrast efficiency of IONP in vivo (Fig. 1). Detailed mechanism on development of IONP based MRI CA with high T2 or T1 relaxivity was firstly discussed according to classical quantum mechanical outer sphere and SBM theory. More importantly, some instances to increase r2 and r1 values of IONP through adjusting the theoretical related key parameters, including magnetization, size, effective radius, and inhomogeneity of surrounding generated magnetic field, crystal phase, coordination number of water, electronic relaxation time, and surface modification are presented to point out the direction of development of IONP with high contrast ability. After detailed discussion on contrast improvement, representative rational designs and advances of IONP to improve its contrast efficiency are reviewed. The enhancement of contrast efficiency could be fulfilled by introducing other imaging modals, endowing IONP with environment response capacity, and designing the surface ligand to optimize in vivo behavior. We hope this detailed review could build a bridge between the proof–of-concept to translate high-performance MRI CAs, further promoting rational design of magnetic nanoparticles for early and accurate lesion detection.

Fig. 1

Schematic diagram of key parameters for the design of high-performance IONP based MRI CA. T2 and T1 relaxivity of IONP directly determine the contrast ability of IONP. Theoretically, contrast ability of IONP could be altered by magnetization, size, effective radii, inhomogeneity of surrounding generated magnetic field, crystal phase, coordination number of water, electronic relaxation time, and surface modification. Additionally, improving in vivo contrast efficiency of IONP through introducing second imaging modal, endowing IONP with environment response capacity, and optimizing the in vivo behavior is another strategy to construct high-performance IONP based MRI CA.

Key parameters enhancing T2 relaxivities of IONPs

Based on its specific magnetic property, especially high saturated (Ms) value, IONP could accelerate the transverse relaxation of proton in surrounding tissue and decrease signal intensity in T2-weighted MRI images. Thus, IONP has been used as T2 CA for the past decades. Several commercial IONP based T2 CAs have been approved by FDA in clinical application. However, some defects, such as low crystalline and poor surface modification, result in the low contrast and reduce the quality of MRI contrast images. Over the past decades, many attempts and theoretical investigations have been conducted to improve the T2 contrast of IONP.

Outer-sphere theory

Theoretically, T2 relaxation of proton could be interpreted by three mechanisms, dipole-dipole coupling [112], curie spin relaxation [113,114], and scalar or contact relaxation [115,116]. An accelerated relaxation phenomenon is observed in the existence of magnetized particles with weak magnetization [117], while the relaxation enhancement is limited. With the development of nanotechnology, IONP have been discovered to enhance the T2 relaxation of proton [25,[118], [119], [120], [121]]. With high magnetic moment, IONP could generate a local magnetic field under the external magnetic field which can effectively shorten the relaxation of surrounding proton [122]. According to the outer sphere theory, the proton dephasing process can be divided into three stages, motional average regime (MAR), static dephasing regime (SDR), and echo-limiting regime (ELR) based on the value of τD [118,123]. τD is the diffusion time and could be calculated by the following equation: τD = d2/4D, where d is efficient radius of particle and D is the water diffusion coefficient. MAR condition is fulfilled when τD < 1/(γBeq), where γ is the proton gyromagnetic ratio and B is the equatorial magnetic field. On the basis of that protons rapidly diffuse around IONP, protons experience quick magnetic field changing in MAR [124], which is effectively time-averaged. According to the quantum-mechanical outer-sphere theory, T2 relaxaivity of IONP is given byWhere Ms, L, and r are the saturation magnetization, thickness of an impermeable surface coating, and efficient radius of magnetic nanostructure. These key parameters are determined by the crystal and surface structure of IONP. Based on this equation, T2 contrast of IONP are proportional to Ms and r. Thus, one can develop IONP with high T2 contrast ability by adjusting these key parameters. T2 contrast of IONP can not uncontrolled increase with the increase of particles size. When the size of particle or cluster reaches a certain limit, the T2 contrast IONP reaches the maximum value and fulfill the SDR [118,125]. In SDR, the T2 relaxivity of IONP are given bywhere A is the lattice parameter, N0 is the Avogadro constant, Z is number of formula units per unit cell, γ is the proton gyromagnetic ratio. Based on this equation, the r2 value of IONP in SDR is highly dependent on its saturated magnetization. Therefore, one can further increase the T2 contrast ability of IONP in SDR through elevating the saturated magnetization.

Along with the further increase of size, T2 relaxivity of IONP reach to ELR [126]. In this regime, proton shows limited diffusion in a time interval. This undesired property leads to T2 contrast reduction with their sizes increase, meaning that construction of high-performance T2 MRI CAs cannot be achieved by endless increasing the radius of IONP.

Magnetic behavior

Based on the quantum-mechanical outer-sphere theory, improvement of Ms can significantly elevate the T2 contrast of IONP. Since magnetic properties of magnetic materials are highly dependent on their crystal structure, such as crystallinity, crystal composition, and crystal size, engineering the structure of IONP is an effective method to ameliorate magnetic behavior and T2 MRI contrast. In the following sections, we will discuss the strategies to optimize crystal structure of IONP to improve Ms value for enhanced T2 contrast.

Crystallinity

The simplest attempt to improve magnetic moment of IONP is improving crystallinity. Structure order, determined by the crystallinity, affects the state of magnetic spin in IONP and determines its magnetic property. IONP could be mainly divided into three phases, those are magnetite, Wüstite, and Maghemite phase. Due to the growth kinetics of Wüstite phase, IONP exhibits typical mixed crystal phase and result in low crystallinity [36,127]. The existing undesired mixed phase in IONP would destroy the long-range-order of magnetic spin and reduce magnetic moment. With the development of nanosynthsis, IONP with high crystallinity could be obtained by tunning the synthesis parameter. For example, Hyeon and colleagues reported that synthesis of IONP via thermal decomposition of precursors in high boiling points solvent provided IONP with high crystallinity and magnetic moment [128,129], paving the way to fabricate high-performance T2 MRI CA. In addition, IONP with high crystallinity could be obtained by other methods, such as polyol and microwave synthesis. Hachani et al. adapted polyol synthesis to get IONP at high temperature and pressure conditions. The as-prepared IONP show high crystallinity and Ms, resulting in high T2 relaxivity [130].

Crystal composition

As an effective method, doping have been widely used to engineer the crystal composition of nanomaterial, which determines the magnetic property of magnetic material [[131], [132], [133]]. Magnetite phase show spinel or inverse spinel structure, which corresponds to the oxygen-packed face-centered cubic lattices with octahedral (Oh) and tetrahedral (Td) site, respectively. Typically, Fe3+ ions occupies the Oh and Td site, while Fe2+ occupies the Oh site. Under an external magnetic field, the magnetic spin in Oh site is aligned in parallel to the external magnetic field, while the magnetic spin in Td site alignes antiparallel to the external magnetic field. Therefore, increasing the magnetic spin in Oh site or reducing the magnetic spin in Td site theoretically increases the magnetization of IONP. Cheon and co-author engineered the structure of Fe3O4 nanocrystal by introducing different divalent magnetic ions with different magnetic spin magnitudes, including Mn2+ (5 μb), Fe2+ (4 μb), Co2+ (3 μb), and Ni2+ (2 μb), to investigate the effect of dopant on its magnetic property and T2 contrast (Fig. 2a and b) [27]. Interestingly, the magnetization and T2 contrast gradually decreases with the order of Mn2+, Fe2+, Co2+, and Ni2+, which is highly consistent with the change of magnetic spin magnitude. It should note that Mn2+ doped Fe3O4 (MnxFe3-xO4) exhibits the highest saturation magnetization (110 emu/g = 110 A·m2/kg) and T2 relaxivity (358 mM−1s−1 at 1.5 T). MnxFe3-xO4 NP shows mixed spinel structure with Mn2+ occupied either Td or Oh site. The rise of Mn2+ doping level may lead to two opposite effects: one is increasing magnetic spin in Fe3O4 nanocrystal and elevating the magnetic moment, the other one is disturbing the long-range order of magnetic spins and resulting in the decrease of magnetic moment. Recent research indicates that the saturation magnetization and T2 contrast of MnxFe3-xO4 nanocrystal gradually increases with the Mn2+ doping level rising from x = 0 to 0.43, while decreases with the doping level further rise to x = 1.06 [134]. When x = 0.43, MnxFe3-xO4 nanocrystal shows highest saturation magnetization (89.5 emu/g = 89.5 A·m2/kg) and T2 relaxivity (904.4 ± 11.1 mM−1s−1 at 7 T). Apart from the magnetic ions, dopant non-magnetic ions into Fe3O4 nanocrystal have been proved to be an effective method to increase its magnetization and MRI contrast ability. Jang et al., constructed Zn2+ doped Fe3O4 nanocrystal [(ZnxFe1-x)Fe2O4] with spinel structure (Fig. 2c–e) [135]. Extended X-ray absorption fine structure analysis indicates that Zn2+ mainly occupies the Td sites, which could effectively reduce the antiferromagnetic coupling interactions between Td and Oh sites and increase the magnetic moment. The magnetic property investigation reveals that the saturation magnetization could be enhanced to 161 emu/g (161 A·m2/kg) when x = 0.4. This high saturation magnetization generate superior T2 contrast with the r2 value of 687 mM−1s−1 at 4.5 T, which is approximately 2.5 and 1.6 times higher than that of Fe3O4 (276 mM−1s−1) and MnFe2O4 (422 mM−1s−1) nanocrystal.

Fig. 2

Magnetic behavior effect on T2 relaxivity of IONP. (a) Mass magnetization values and schematics of spin alignments and (b) T2 relaxivities of MnFe2O4 (MnMEIO), Fe3O4 (MEIO), CoFe2O4 (CoMEIO) and NiFe2O4 (NiMEIO). Reproduced with permission [27]. Copyright 2007, Nature Publishing Group. (c) Magnetic spin alignment diagram of (ZnxFe1-x)Fe2O4 nanoparticles with x = 0, 0.2, and 0.4 under applied magnetic field. (d) Ms and (e) r2 values of (ZnxMn1-x)Fe2O4 and (ZnxFe1-x)Fe2O4 nanoparticles with different Zn doping level. Reproduced with permission [135]. Copyright 2009, John Wiley & Sons, Inc. (f) TEM image and (g) r2 value of iron/iron oxide core/shell nanoparticle. Reproduced with permission [139]. Copyright 2011, John Wiley & Sons, Inc. (h) r2 values of iron nanoparticle coated by various magnetic shell at the fixed larmor frequency. Reproduced with permission [141]. Copyright 2011, John Wiley & Sons, Inc. (i) Diagram of spin canting effect in various sized IONPs. (j) M − H curve of IONPs with sizes of 1.5, 2.2, and 3 nm at 300 K. Reproduced with permission [46]. Copyright 2011, American Chemical Society. (k–l) TEM images of cube and sphere nanoparticles. (m–n) Simulated magnetic spin state of cube and sphere, indicating the degree of spin canting against external magnetic field. Reproduced with permission [149]. Copyright 2012, American Chemical Society.

Core-shell structure

Compared to the IONP, metallic iron nanoparticle exhibit significantly higher saturation magnetization at room temperature. Introducing metallic iron moiety into IONP could effectively increase its saturated magnetization and T2 contrast [[136], [137], [138]]. To improve magnetic moment of IONP, Tilley and co-authors synthesized high crystalline iron/iron oxide core/shell nanocrystal by naturally oxidation iron nanoparticle (Fig. 2f and g) [139]. Introducing iron core endows this core/shell nanostructure with high magnetization with the value of 150 emu/g (150 A·m2/kg) (Fe), which is significantly higher than that of normal Fe3O4 nanocrystal. This far elevation of magnetization results in the obviously rising of T2 relaxivity from 145 mM−1s−1 to 324 mM−1s−1 at 9.4 T. It should note that the crystallinity of core/shell nanostructure is an important parameter to its T2 relaxivity [140]. The amor-Fe/Fe3O4 exhibits remarkably lower T2 relaxivity (67 mM−1s−1) than bcc-Fe/Fe3O4 (220 mM−1s−1) with high crystallinity, even much lower than Ferridex (110 mM−1s−1). The crystal structure of iron oxide shell is another parameter to affect its magnetization and T2 relaxivity. Yoon et al. coated monometallic iron with MnFe2O4, Fe3O4, CoFe2O4, and FeO (Fig. 2h) [141]. These iron/iron oxide core/shell nanostructures exhibite increased magnetization and T2 relaxivity with shell structures altered from FeO to CoFe2O4, Fe3O4, and MnFe2O4, consistent with the magnetization change of the shell structure. These results mean that coating monometallic iron with a magnetic shell with high magnetization could further improve its magnetization and T2 relaxivity.

Disordered spin

Previous researches indicate that IONP could be considered as a core/shell structure with the magnetic core and magnetically dead layer, where the magnetic spin is lack of full alignment. In the magnetically dead layer, the spin canting and other effects destroy the long-range-order of magnetic spins and result in the disorder of magnetic dipoles, which lead to the decrease of magnetic moment [[142], [143], [144], [145], [146], [147]]. The thickness of spin-canting layer has been determined as about 0.9 nm. Thus, along with the increase of size, the ratio of spin-canting layer to whole magnetic nanoparticle decreases (Fig. 2i and j). It has been reported that the magnetization moment and r2 value of IONP increases with the decrease of magnetically dead layer ratio. This phenomenon has been found in MnxFe3-xO4 nanoparticle as well, which implies that one can develop IONP with high T2 relaxivity through decreasing the magnetic dead layer ratio [132]. However, recent researches indicate that this elevation effect is connected to the size of IONP. Rinaldi and co-author synthesized IONP at the present of molecular oxygen to increase the magnetic diameter [148]. Interestingly, increasing the magnetic diameter for the IONP with physical diameter of 21 nm leads to slight improvement in the T2 relaxivity. This result could be attributed to relatively low ratio of magnetically dead layer in IONP with large physical size, limiting the improvement on T2 relaxivity by further decreasing the magnetically dead layer.

Morphology as a key factor to determine the surface structure of nanomaterials, has been proven to affect the canting spins on the surface of IONP as well. Cheon and co-authors compared the saturated magnetization of cubic and spherical (ZnxFe1-x)Fe2O4 nanoparticle with the same magnetic volume (Fig. 2k-n) [149]. The orientations of the overall magnetic spin structure shows that the disordered spins are homogeneously distributed on the surface of spherical IONP with certain thickness. While, the disordered spins are mainly distributed at the corner of the cubic IONP. Compared to the spherical IONP without a certain exposed facet, cubic IONP shows the single crystal facet. This unique feature results in the similar spin state in the core and surface, which reduce surface anisotropy. The calculated disordered spins of cube are about 4%, which is significantly lower than that of sphere with the value of 8%. Therefore, the saturated magnetization of cube (165 emu/g = 165 A·m2/kg) is remarkably higher than that of sphere (145 emu/g = 145 A·m2/kg).

Effective radius

Apart from Ms, size also determines the region, in which proton relaxation could be accelerated by IONP. Normally, the region could be simulated to a sphere covering the full IONP. The diameter of this region could be defined as the effective radius. Effective radii determined by the size is responsible for the field perturbation area for the outersphere protons and proportional to the T2 relavixity in MAR, which means improvement of T2 relaxivity of IONP could be achieved by increasing the effective radius with maintained saturated magnetization. However, T2 contrast of IONP are not uncontrolled increase with the increase of efficient radius. When the efficient radius of particle or cluster reach to a certain limit, the T2 contrast of IONP reach to its maximum value and fulfill the SDR. In this section, we will focus on the strategy to develop high-performance T2 CA by increasing effective radius of IONP through controllable synthesis.

Individual IONP

In MAR T2 relaxivity of IONP is proportional to the square of effective radius according to outer sphere theory [124]. For spherical IONP, its effective radius is directly determined by its diameter, thus, increasing its diameter have been proven to be the simplest method to improve its T2 relaxivity. Cheon and co-authors find that IONP shows a typical diameter-dependent T2 relaxivity, that is, r2 value increase with the rising of diameter [150]. This diameter-dependent T2 relaxivity has been observed in other researches, which is highly consistent with the theoretical analysis [33,[151], [152], [153]]. Besides, sophisticated morphology of magnetic nanostructure can increase the effective radii of particle cores and improve intensity, direction, and gradient of the magnetic stray field, accelerating surrounding proton diffusion and dephasing. Zhao et al. developed a novel IONP with octapod morphology with the assistance of chloride ion (Fig. 3a–e) [154]. Due to its unique morphology, the calculated effective radius of IONP with octapod morphology is about 2.4 times higher than that of spherical IONP with the same solid volume and similar magnetization. Further MRI investigation indicates that the octapod IONP with edge length of 30 nm show an ultrahigh T2 relaxivity with the value of 679.3 ± 30 mM−1 s−1 at 7 T, which is about 5.4 times larger than that of spherical IONP with the same geometric volume. Inspired by this research, a large number of IONPs with anisotropic morphologies and improved effective radii, such as plate [155,156], cube [157,158], and tripod [159], have been reported to show enhanced T2 relaxivity. Recently, Yang et al., systematically investigate the morphology effect on the effective radius and T2 relaxivity of IONP (Fig. 3f) [160]. They synthesized MnxFe3-xO4 nanoparticles with the morphology of sphere, cube, plate, tetrahedron, rhombohedra, and octapod with the same solid volume and similar magnetization. The calculated effective radius shows a gradually decrease with the order of octapod, rhombohedra, tetrahedron, plate, cube, and sphere. The MRI analyses indicate that the T2 relaxivities of all samples show descending tendency from octapod to sphere, which is the same to the tendency change of effective radius. The increased T2 relaxivity along with the rising of effective radius demonstrate that effective radius can eventually determine T2 relaxivity in MAR.

Fig. 3

Effective radius effect of single IONP on its T2 relaxivity. (a) TEM and HRTEM images of octapod IONP with four-armed star-like particles. (b) Schematic cartoon shows the ball models of octapod and spherical IONP with the same geometric volume. With the same geometric core volume, octapod IONP shows larger effective volume than spherical IONP. (c) M − H curves of Octapod-30, Octapod-20, Spherical-16, and Spherical-10 measured at 300 K. (d) T2-weighted MRI images and (e) r2 values of Octapod-30, Octapod-20, Spherical-16, and Spherical-10, respectively. Reproduced with permission [154]. Copyright 2013, Nature Publishing Group. (f) r2 values of IONPs with different morphologies, including sphere, cubes, plates, tetrahedrons, rhombohedra, and octapod. Reproduced with permission [160]. Copyright 2018, American Chemical Society. (g) Color-coded T2-weighted MRI images and (h) comparison of r2 values of ferrimagnetic cubic IONPs. Reproduced with permission [158]. Copyright 2012, American Chemical Society. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

In contrast to the MAR, the large magnetic nanoparticles generate strong dipolar field around it in SDR, which significantly reduce the influence of proton diffusion on signal decay. Thus, the T2 relaxivity of IONP can not increase with the rise of particle size and reach to a plateau [123,124]. Recently, Hyeon and co-workers investigate the T2 relaxivity of cubic IONP with the size of 22, 28, 32, 42, and 49 nm (Fig. 3g and h) [158]. Due to the proper hydrodynamic diameter, cubic IONP with the size of 22 and 28 nm are in the SDR and show extremely high T2 contrast with the similar r2 values of approximately 800 mM−1s−1. While, cubic IONP show reduced r2 value when the size exceed 30 nm. These observations could be attributed to the large size of cubic IONP lead to them out of SDR and reach ELR. These results are highly consistent to the theoretical studies and point out that accurately controlling the particle size of IONP in SDR could obtain high-performance T2 MRI CA.

Assembled IONP

Clustering of IONP is another effective method to increase their effective radius with increased T2 relaxivity [126,[161], [162], [163]]. To generate assembled IONP, polymers have been chosen as candidates due to the finely controlled size and colloidal stability. Ai et al. encapsulate hydrophobic IONP inside the core of polymeric micelle to form IO nanocluster (Fig. 4a–d) [164]. The T2 relaxivity of IO nanocluster is approximately 6 times higher than individual IONP coated by PEG. Additionally, they observed that T2 relaxivity increased with the increasing of the size and loading density of IONP. Compared to polymer, silica with high surface area and high-biocompatibility has been used as the matrix to construct IO nanocluster. For example, modifying dye doped silica with multiple IONP to improve the T2 relaxivity of IONP has been reported (Fig. 4e–g) [165]. MRI contrast ability investigation indicates that r2 value of silica based IO nanocluster is about 2.8 times higher than that of dispersed IONP. This clustering effect based T2 relaxivity enhancement has also been observed in other researches, indicating its universality [21,[166], [167], [168]]. Besides, encapsulated IONP within silica matrix have been proved to develop high-performance IO nanocluster [162]. Weissleder and co-authors develop a strategy to encapsulate multicore IO nanocluster within the thin silica shell. With the increasing of size, IO nanocluster locates in MAR, SDR, and ELR, respectively. The IO nanocluster shows the highest T2 relaxivity in SDR with the value of 695 mM−1s−1. These size-dependent T2 relaxivity changing tendency of IONP has also been observed by Weller and co-authors. Moreover, Weller and co-authors further investigate the effect of size of clustered single IONP on the T2 relaxivity of IO nanocluster [163]. They found that the T2 relaxivity is regardless of the size of clustered single IONP in MAR. In contrast, T2 relaxivity of IO nanocluster in the SDR seems to increase with size rising of the clustered single IONP. Since the size of IONP is proportional to its magnetic moment, these results may indicate that the T2 relaxivity in SDR could be further increased by optimizing the magnetic property.

Fig. 4

Assembled effect on T2 relaxivity of IONP. (a) TEM images of IONP-loaded micelles after negative staining. (b) T2 relaxivity and (c) T2-weighted MRI images of 4 nm IONP-loaded PCL5k-b-PEG5k micelles and DSPE-PEG5k micelles at 1.5 T. (d) Table of T1 relaxivities, T2 relaxivites, and MRI sensitivity of different IONP-loaded micelles. Reproduced with permission [164]. Copyright 2005, John Wiley & Sons, Inc. (e) TEM images of IONP-MSN. (f) Plot of T2 relaxation and (g) T2-weighted MRI images of IONP-MSN and free IONP with the same Fe concentration. Reproduced with permission [165]. Copyright 2009, American Chemical Society.

Synergistic effect of magnetic behavior and effective radius

Since individually increase the Ms and effective radii of IONP could improve its T2 relaxivity, one can hypothesize that simultaneous increase Ms and effective radius of IONP could further increase its T2 relaxivity. Unfortunately, the traditional synthetic method limits the simultaneous controlling the crystal structure and morphology. With the development of synthetic method, more and more researches on simultaneous improving Ms and effective radius of IONP have been reported. Zhao et al. developed a novel strategy to improve Ms and effective radii of IONP through cation exchange reaction (Fig. 5a–e) [169]. They successfully synthesized cubic and octapod IONPs with high effective radii and magnetization moment. These engineered IONPs exhibit outstanding capacity to achieve T2 contrast enhancement, especially zinc engineered octapod IONP (r2 = 754.2 mM−1s−1 at 7 T). More recently, Gao and co-workers further elevate the dopant level of zinc in octapod IONP to further increase Ms of IONP with high effective radius (Fig. 5f–h) [170]. The saturated magnetizations show the anti-V shape trend with the rising of zinc doping level. With the doping level of x = 0.44, the Ms value reaches to the peak with the value of 88.9 emu/g (88.9 A·m2/kg). On the basis of the high effective radius and Ms value, octapod ZnxFe3-xO4 nanoparticle (x = 0.4) exhibit the notably high T2 relaxivity with the value of 989.1 mM−1s−1 at 7 T, which endow it as a sensitive T2 CA to fulfill the sensitive detection on orthotopic and metastatic tumor on mice model.

Fig. 5

Simultaneous optimization of magnetic behavior and effective radii to improving T2 relaxivity. (a) Scheme of cation exchange in IONP. (b) TEM images of octapod IONP treated by Mn and Zn cations. (c) Comparison of Ms and (d) r2 values and T2-weighted MR images of octapod IONP treated by Mn and Zn cations. (e) In vivo MRI images of metastatic hepatic carcinomas before and 1 h after intravenous injection of octapod IONP treated by Zn cations. Reproduced with permission [169]. Copyright 2016, American Chemical Society. (f) Comparison of Ms and (g) r2 values of Zn doped IONP with different ratio. (h) T2-weighted MR images of Zn doped IONP with different ratio at 7 T. Reproduced with permission [170]. Copyright 2019, American Chemical Society.

Inhomogeneous magnetic field

Theoretically, the diffusion of protons in the inhomogeneous magnetic field will lead to the dephasing of proton, which is the basis of the T2 relaxation generation. The existence of T2 MRI CA could generate an inhomogeneous magnetic field around it and accelerate the dephasing process of around protons, which is the basic goal of T2 CA [119,121,171,172]. The improvement of magnetic moment and size of IONP can enhance the local gradients of magnetic field around it and shorten the spin-spin relaxation time. Besides, IONP with anisotropic morphology, especially with sharp corner, could enhance the inhomogeneity of surrounding magnetic field. For instance, Gao and co-authors investigate the effect of the stray field gradient generated by MnxFe3-xO4 nanoparticles with different anisotropic morphologies on their T2 relaxivities (Fig. 6a–f) [160]. The Landau-Lifshitz-Gilbert (LLG) equation simulations indicate that different morphologies could effectively affect the shape, intensity, and gradient of the stray field. Moreover, the stray field generated by octapod morphology with eight sharp corners shows the highest inhomogeneity, resulting in the highest T2 contrast among all samples. Recently, Zhou et al. investigate the inhomogeneity of stray magnetic field generated by IO based nano-cluster (Fig. 6g–s) [173]. They prepared a series of IO nano-clusters, including C1 (5 nm IONP only), C2 (15 nm IONP only), C3 (mixed 5 and 15 nm IONP), C6 (cubic IONP), C7 (plate IONP). The LLG simulation results indicate that magnetic field generated by C3 shows stronger inhomogeneity compared to C1 and C2. Therefore, C3 present significantly higher T2 relaxivity (533.4 mM−1s−1) than C1 (231.6 mM−1s−1) and C2 (358.3 mM−1s−1). This magnetic field inhomogeneity based T2 relaxivity enhancement is also observed in C6 and C7. These IO nano-clusters constructed by IONPs with anisotropic morphologies show enhanced magnetic field inhomogeneities and improved T2 relaxivities compared to the traditional IO nano-clusters. Since electrons of atom surrounding the magnetic nanoparticles undergo circulation under external magnetic field, which could generate a small opposite magnetic field to the external magnetic field and further increase the inhomogeneity of local magnetic field, local magnetic field inhomogeneity could also be increased by the surface ligand of IONP as well. Gao and co-author investigated the influence of anchoring group on the local magnetic field [174]. They found that IONP modified by catechol and hydroxamate group generated stronger inhomogeneity than that modified by diphosphate group, due to the greatly contribution of π electron circulation on increasing the inhomogeneity of local magnetic field.

Fig. 6

Inhomogeneous magnetic field effect on T2 relaxivity of IONP. (a) Scheme of water molecular diffusion and relaxation process around spherical IONP. The color indicates the intensity of local field induced by IONP under an external magnetic field. (b) The spatial distribution of stray fields caused by IONP with different morphologies. (c) Values of magnetic susceptibility and (d) stray field gradient vs distance from the surface of IONP with different morphologies. (e) r2 values and (f) T2-weighted MR images of IONP with different morphologies at 7 T. Reproduced with permission [160]. Copyright 2018, American Chemical Society. (g) Schematic cartoon illustrates the simulation of two adjacent IONP of different size. (h–l) Landau-Lifshitz-Gilbert simulation results of the stray field for model C1–C5, respectively. (m–o) Simulation models and the calculated stray fields for the models C6 and C7, respectively. (p) Comparison of r2 values and (q) T2-weighted MR images of IO-5, IO-15, C1, C2, and C3, respectively. (r) Comparison of r2 values and (s) T2-weighted MR images of cubic IONP, C6, and C7, respectively. Reproduced with permission [173]. Copyright 2017, Nature Publishing Group. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Surface coating structure

To act as a MRI CA for diagnosis, IONP need to be coated with hydrophilic layer, including small molecule, polymer, and protein, to disperse in aqueous solution [5,[175], [176], [177], [178]]. There are two main effects of surface coating on the T2 relaxivity of IONP. Firstly, the coating layer require an anchoring moiety to chelate with Fe2+ or Fe3+ ions on surface, which may affect its magnetic property. Secondly, the coating layer can limit the diffusion of proton and hinder the interaction between proton and magnetic field induced by IONP.

Effect on magnetization moment

Due to the chelating with Fe2+ and Fe3+ ions, the coating layer may affect the arrangement of surface atoms and magnetization moment. Serna and co-authors report that the coordination of oleic acid to the surface of IONP can effectively reduce the spin canting effect and increase the magnetic moment [179]. They observed that saturated magnetization of IONP coated with oleic acid (about 78 emu/g = 78 A·m2/kg) was higher than that without oleic acid. However, recent study indicate that chelating IONP with high affinity group may introduce structure defect and disturb the long-range-order of magnetic spin. Zeng et al. investigate the different surface coating ligand on the saturated magnetization of IONP (Fig. 7a–c) [174]. They found Ms value and T2 relaxivity of IONP decreased with the order of surface agent as hydroxamate > catechol > diphosphate, which was inversely correlated to their binding affinity to Fe3+ ions. This study suggests that coating IONP with moderate affinity group could optimize its magnetic property and improve its T2 contrast.

Fig. 7

Effect of surface coating structure on T2 relaxivity of IONP. (a) Chemical structure of PEG used for exchanging and TEM images of PEGylated IONP with the sizes of 3.6 and 10.9 nm, respectively. (b) Ms values and (c) plot of T2 relaxation rates of PEGylated IONPs with the sizes of 3.6 and 10.9 nm. Reproduced with permission [174]. Copyright 2014, John Wiley & Sons, Inc. (d) TEM images and (e) r2 values of silica coated IONPs with different silica shell thickness. Reproduced with permission [180]. Copyright 2013, Springer Nature. Normalized magnetic field of IONP with the sizes of (f) 5 and (g) 14 nm. The color bar represent the magnitude of the magnetic field strength. Starting from center, the dash lines indicate PEG550, PEG750, PEG1000, PEG2000, and PEG5000. T2 relaxivities of different PEG coated IONPs with constant (h) iron concentration or (i) particle concentrations. Reproduced with permission [181]. Copyright 2010, American Chemical Society. (j) TEM images of casein coated IONP. (k) Pictures of gel analyses of casein and oligosaccharide or casein coated IONP. (l) T2-weighted MR images and (m) T2 relaxation rates of casein and polymer coated IONP. Reproduced with permission [182]. Copyright 2013, American Chemical Society. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Proton diffusion

T2 relaxation of proton is highly determined by its diffusion behavior in the magnetic field gradients. The coating layer could affect the diffusion of proton by reducing the diffusion magnetic field strength, limiting proton diffusion, and forming hydrogen bond to proton. Previous research reported that the magnetic field generated by IONP decreased with the increase of the distance between IONP surface and proton. When the coating thickness increases to 10.8 nm, the magnetic field strength reduces to 2.3% of that on the surface [183]. This undesirable decrease may lead to the drop of T2 relaxivity of IONP. Dravid and co-authors found that with the increasing of silica shell thickness from 1 to 14 nm, the T2 relaxivity of IONP decreased from 94 to 32 mM−1s−1 (Fig. 7d and e) [180]. Meanwhile, the coating layer reduces the proton diffusion efficiency within the layer, which benefits to its T2 relaxivity. Thus, one can optimize the coating layer structure to enhance the T2 relaxivity of IONP. Tong et al. investigated the effect of molecular weight of PEG on the T2 relaxivity of IONP (Fig. 7f–i) [181]. They modified 5 and 14 nm IONP by PEG chain with molecular weight of 550, 750, 1000, 2000, and 5000 Da, respectively. Regardless of the core size, T2 relaxivity of PEG modified IONP is highly related to the molecular weight of PEG and show specific plateau with a certain value. Additionally, bio-macromolecule, especially protein, have been proved to be a new candidate to optimize the coating layer. Mao and co-authors developed a high-performance T2 CA by coating IONP with casein (Fig. 7j-m) [182]. The T2 relaxivity of casein coated IONP is about 2.5 times higher than that coated by amphiphilic polymer. The high T2 relaxivity could be mainly attributed to the high permeability, high affinity to proton, and abundant hydrate group of casein. Based on these unique features, the casein coating layer benefits the diffuse of surrounding proton and increases the exchange efficiency between hydrate water and bulky one. This T2 relaxivity enhancement effect could be expanded to other proteins, such as human serum albumin. The detailed theoretical and experimental discussion on T2 contrast of IONP clearly indicate the limitation parameters, such as Ms, effective radius, dephasing process and proton diffusion, on the r2 value of IONP (see Table 1). We highlight those strategies to optimize these parameters to improve T2 contrast of IONP, which encourage further investigations on improving T2 contrast of IONP.

Table 1

The limitation parameters on the r2 value of IONP and current strategies to optimize these parameters in different regions.

regime	Optimized Parameter	Strategy	r2 value (mM−1s−1)
MAR	Ms	Dopant with Mn ions	358 (1.5 T) [27]
		Dopant with Mn ions	687 (4.5 T) [135]
		IONP coated iron	324 (9.4 T) [139]
		Coated monometallic iron with Mn doped IONP	430 (0.47 T) [141]
		Reducing magnetically dead layer	175 (1.7 T) [148]
	Effective radii	IONP with octapod morphology	679 (7 T) [154]
		Clustering of IONP through polymer	471 (1.5 T) [164]
	Effective radii/Ms	Zn doped octapod IONP	989 (7 T) [170]
	Dephasing process	IO cluster with muti-single core	533 (7 T) [173]
		IONP modified by hydroxamate group	92 (3 T) [174]
	Proton diffusion	Adjusting thickness of shell	385 (7 T) [181]
		Coating IONP with casein	273 (3 T) [182]
SDR	Ms	Cubic IONP	800 (3 T) [158]
		Increasing the size of clustered single IONP	∼330 (1.41 T) [163]

IONP as T1 CA

IONP shows high biocompatibility, which is highly desirable for biomedicine application. However, traditional IONP exhibits relatively low T1 relaxivity and high r2/r1 ratio, hindering its application as the T1 CA [46,155]. Theoretically, T1 relaxivity of IONP is determined by the chemical exchange efficiency between magnetic ions and proton, which is highly dependent on its crystal structure. Recently, a growing attention have focused on engineering crystal structure of IONP to improve its T1 relaxivity and construction of IONP based T1 CA.

SBM theory

T1 CAs lead to the energy loss of spin through dipole-dipole interactions between water protons and magnetic ions, which causes the T1 relaxation occur [184,185]. There are three regions for interactions between water proton and magnetic ions, inner-sphere, secondary intermediate sphere, and outer sphere [117]. The inner-sphere describes direct interaction between water protons and magnetic ions and dominates the T1 relaxation enhancement for the T1 CAs. In secondary intermediate sphere and outer sphere, magnetic ions interact with the diffusing protons and exchangeable protons, which are not directly bind to the magnetic ions and accompany the exchange through hydrogen bond [186,187]. However, the secondary intermediate sphere and outer sphere mediated T1 relaxation enhancement are negligible compared to that caused by the inner sphere. Therefore, secondary intermediate sphere and outer sphere are often ignored due to the present of water proton in inner sphere in a real system. The r1 value in inner sphere is given by [188,189].Where Pm is the mole fraction of water coordinating to the metal center, T1m is the applicable dipole-dipole relaxation, q is the coordination number of water, τm is proton residence lifetime, τci [i = 1,2] is the correlation time, τr is the molecular tumbling time, r is the distance between magnetic ions and proton, and τιs is the electronic relaxation time. Ideally, a given T1 CA with large q, short τm, long τr, and long τs exhibits high T1 relaxivity. Since IONP with large size show long τr compared to metal complex, improvement on q, τm, and τs is the main method to increase its T1 relaxivity. However, IONP with high T1 relaxivity does not mean that it is suitable to be used as the T1 CA. The key parameter to determine whether a specific IONP could be considered as T1 CA is its r2/r1 ratio. For example, 12 nm MnxFe3-xO4 nanoparticle show high T1 relaxivity with the value of 38.2 mM−1s−1 while high r2/r1 ratio with the value of 7.4. The relative high r2/r1 ratio suggests that 12 nm MnxFe3-xO4 nanoparticles should be considered as the T2 CA [153]. In this section, we will discuss the effective strategies to construct IONP based T1 CA with the low r2/r1 ratio.

Reduction of magnetic moment

IONP exhibits high magnetization, meaning high T2 contrast ability and high r2/r1 ratio. This defect highly hinders its application as a T1 CA and should be overcome to develop IONP based T1 CA. The magnetic property of IONP is highly determined by its crystal structure. One can optimize some key parameters, including size, crystallinity, and surface modification, to adjust the magnetic moment of IONP and develop T1 CA. Recently, many strategies have been reported to reduce the magnetic moments of IONPs and optimize their r2/r1 ratio, which have been proved to be effective method to develop IONP based T1 CA.

Spin disorder surface

Magnetic moment of IONP is highly dependent on its size. Normally, the magnetic moment decreases with the drop of its size, which could be ascribed to the increase of the spin disorders on the surface of IONP. This spin disorder surface of IONP are lack of full alignment and may destroy the long-range-order of magnetic spins, which could trigger the decrease of magnetic moment [144,146,190,191]. Since the thickness of this spin disorder surface is almost constant, the ratio of spin disorder surface to whole particles increase with the size decrease. For example, the proportion of spin disorder surface to entire particle is about 35% for 12 nm IONP, while increases to 48% when the size decreases to 5 nm [192]. Therefore, reducing the core size of IONP is the most straightforward method to reduce the magnetic moment of IONP and develop T1 CA. Hyeon and co-authors synthesized IONP with size of 3 nm and investigated its T1 contrast (Fig. 8a–e) [46]. Due to the remarkable spin canting effect, IONP with the size of 3 nm shows lower magnetic moment, higher T1 relaxivity, and lower r2/r1 ratio compare to 12 nm IONP. However, the T1 relaxivity of IONP does not always increase with the decrease of their sizes. Shen et al., reported that with the size increase, the r1 value of IONP shows an anti-V shape curve and reach the plateau with the size of 3.6 nm (Fig. 8f and g) [193]. Interestingly, the r2/r1 change exhibits a contrary tendency to r1 change and reach the minimum at the size of 3.6 nm.

Fig. 8

Effect of spin disordered surface and crystal phase on T1 relaxivity of IONP. TEM images of IONP with the sizes of (a) 1.5, (b) 2.2, (c) 3, and (d) 3.7 nm, respectively. (e) Plot of T1 relaxation rate of IONP with different sizes. Reproduced with permission [46]. Copyright 2010, American Chemical Society. (f) r1 value and r2/r1 ratio of exceedingly small IONP as a function of sizes. (g) Relative intensity of MR images for exceedingly small IONP with the sizes of 3.3, 3.6, and 4.2 nm, respectively. Reproduced with permission [193]. Copyright 2017, American Chemical Society.

Crystal phase

Magnetite phase of IONP show the highest magnetic moment and is considered as the best candidate for T2 contrast imaging among all phases. However, the high magnetic moment result in the high r2/r1 ratio and hinder its application as T1 CA. Wüstite and Maghemite phase have been reported to show relative lower magnetic moment than magnetite phase, endowing them with the suitable magnetic property to reveal low r2/r1 ratio and be used as the T1 CA [194,195]. Compared to Wüstite phase composed with Fe2+ cations (4 unpaired electrons), maghemite phase composed with Fe3+ cations (5 unpaired electrons) presented higher q values and T1 relaxivity. Bawendi and co-authors fully oxidized the synthesized IONP from magnetite phase to maghemite by trimethylamine N-oxide [196]. The resultant product shows high T1 relaxivity with the value of 5.2 mM−1s−1 and relatively low r2/r1 ratio with the value of 2.0 at 1.5 T, ensuring it to fulfill MRA and conventional T1 MRI contrast in vivo.

Coordination number of water (q)

Theoretically, q is proportional to the T1 relaxivity of T1 CA. For IONP, its T1 shorten effect could be attributed to existed iron ions on its surface. Since q of iron ion is constant, the straightforward method to optimize the q of IONP is increase the number of iron ions on surface [155,197,198]. The number of iron ions exposed on the surface of IONP is determined by two key factors: One is the surface to volume ratio, the other is the exposed crystal facet. Ordinarily, high surface to volume ratio and iron rich facet exposure are desirable to increase the number of iron ions on the surface of IONP. Thanks to the development of synthetic method, more and more IONPs with high surface to volume ratio and iron rich exposed surface have been reported to pursue IONP based high-performance T1 CA [44,[199], [200], [201], [202]]. In addition, q of T1 CA is highly dependent on the number of unpaired electrons. Recently, introducing other magnetic ions with large amounts of unpaired electrons into crystal structure of IONP have been discovered to increase the q of IONP and elevate its T1 relaxivity [53,61,[203], [204], [205], [206], [207]]. Here, strategies to increase q of IONP to raise its T1 relaxivity have been reviewed in the following section.

Hollow structure

Surface iron ions of IONP, conducting chemical exchange with protons and accelerate their longitudinal relaxations, could be considered as the effective iron ions. Unfortunately, the chemical exchange between inner iron ions and protons have been blocked by the outershell, resulting in a reduction of the number of effective iron ions in IONP. Hollow structures with two interfaces between nanocrystal and surrounding environment can exceedingly rise the number of exposed magnetic ions, which is beneficial to elevate T1 contrast of IONP [[208], [209], [210], [211]]. Additionally, the hollow structure may disturb the long-range-order of magnetic spin and reduce magnetic moment, lowering the r2/r1 ratio. Inspired by these results, IONP with hollow structure have been synthesized to develop IONP based T1 CA. Wei et al. synthesized hollow porous IO nanobox via the template-based method (Fig. 9a–e) [210]. The r1 value of hollow IONP with the size of 14 nm is about 27.2 mM−1s−1, which is about 2.5 times higher than that of solid spherical IONP with similar size. More importantly, hollow IONP exhibits lower r2/r1 ratio than corresponding solid IONP with the value of 2.0 vs 10.4 due to its low magnetic moment. These results clearly demonstrate the promising potential of hollow IONP as a new T1 CA with improved T1 contrast. However, there are two key factors that should be paid attention to fabricating hollow IONP based T1 CA, those are density and thickness of the shell. Since the elevation of T1 relaxivity is mainly attributed to the addition of a new interface between IONP and proton environment, the efficiency of chemical exchange between iron ions in inner interface and proton determine the T1 relaxivity of hollow IONP. The density and thickness determine the efficiency of water proton enter the cavity and achieve the chemical exchange. Thus, the desired hollow IONP based T1 CAs should exhibit porous and thin shell.

Fig. 9

Effect of surface-to-volume ration on T1 relaxivity of IONP. TEM images of hollow porous IONP with the sizes of (a) 21, (b) 14, and (c) 9 nm, respectively. (d) Relaxation measurements and (e) T1-weighted phantom images of hollow porous IONPs with different sizes at 0.5 T. Reproduced with permission [210]. Copyright 2018, American Chemical Society. (f) TEM images of ultrathin IO nanowhisker. (g) T1 relaxation curve and (h) T1-weighted images of water and IO nanowhisker. Reproduced with permission [201]. Copyright 2015, John Wiley & Sons, Inc.

Morphology

The T1 relaxivity of spherical IONP have been comprehensively investigated. Changing the morphology of IONP from spherical to some specific shape can increase its surface-to-volume ratio and elevate the number of exposing magnetic ions. Consequently, the efficiency of chemical exchange reaction to surrounding proton and T1 relaxivity of IONP improved. For example, the surface-to-volume ratio of cubic morphology is remarkably higher than that of spherical morphology. Demir and co-authors synthesized IO nanocubes with the size of ∼9.7 nm [212]. The T1 relaxivity assessment indicates that the r1 value of this cubic IONP is about 8.3 mM−1s−1, which is higher than that of 3 nm spherical IONP. Author contribute this elevation effect on T1 relaxivity to the increased surface-to-volume ratio from spherical to cube. Besides, T1 contrast ability of IO nanowhisker with small diameter have been evaluated by Macher et al. (Fig. 9f–h) [201]. IO nanowhisker shows large aspect ratio with the length of 20 nm and diameter of 2 nm, which endows it with large surface-to-volume ratio to achieve efficient chemical exchange reaction to water. With high surface-to-volume ratio, this IO nanowhisker show high T1 relaxivity with the r1 value of 6.13 mM−1s−1, which is significantly higher than that of Magnevist with the r1 value of 3.3 mM−1s−1. It should note that this IO nanowhisker shows paramagnetic behavior due to the high surface-to-volume ratio, which result in forming iron-ligand complexes layer and reducing T2 relaxivity. This effect ensure this IO nanowhisker with low r2/r1 ratio (1.83) to achieve sensitive T1 contrast imaging.

Additionally, nanoparticles with anisotropic morphology can exhibit unique atomic package and show distinct interface to surrounding environment, which can affect the occupancy rate of magnetic ions on the surface and T1 relaxivity of IONP. Zhou et al. obtained IO nanoplate with different thickness by controllable synthesis (Fig. 10a–h) [155]. This IO nanoplate exposes two Fe3O4 (111) facets, which are Feoct2-ter1 terminated. The exposed crystal facet could increase the interaction between surface iron ions and surrounding water proton. The r1 value of IO nanoplate with the thickness of 2.8, 4.8, and 8.8 nm are 14.36 ± 1.24, 43.18 ± 3.33, and 38.11 ± 1.04 mM−1s−1, respectively, which are significantly higher than that of spherical IONP. It seems that T1 relaxivities of IO nanoplates are highly determined by the ratio of (111) area to volume except for the 2.8 nm nanoplate, which could be attributed to the existence of spin-disorder at its corner. Further investigation on the effect of different exposed facet on T1 relaxivity of IONP have been performed by Gao and co-authors (Fig. 10i–k) [160]. They studied T1 relaxivity of MnxFe3-xO4 nanoparticles with the morphologies of sphere, cube, plate, tetrahedron, and octapod with the exposed facet of (100), (111), (110), and (311), respectively. They found that different exposed facets of MnxFe3-xO4 nanoparticles could provide different amounts of effective magnetic ions and result in the different increase degree on their T1 relaxivities. The order of effective metals per a2 on each facet is (110) > (111) ≈ (311) > (100). Unfortunately, the r1 values of plate, tetrahedron, and octapod are almost the same. These results could be ascribed to the order of surface-to-volume ratios of these morphologiesare octapod, plate, and tetrahedron, which partly offset the effect caused by the facet.

Fig. 10

Effect of anisotropic morphology on T1 relaxivity of IONP. TEM images of IO nanoplate with the thickness of (a) 8.8, (b) 4.8, and (c) 2.8 nm, respectively. (d) HRTEM image of IO nanoplate, indicating the (220) planes. (e) Perspective and (f) top views of Feoct2-ter1-terminated (111) planes of Fe3O4 structure, showing the iron iron-rich characteristics. (g) Relationships of T1 relaxivity with the (111) surface of IO nanoplate compare to spherical IONP with equivalent whole surface areas. (h) T1 NMRD profiles of IO nanoplate with different thickness as the function of applied magnetic field. Reproduced with permission [155]. Copyright 2014, American Chemical Society. (i) The exposed faces of (100), (110), (111), and (311) of Mn doped IONP. (j) The relationship of surface-to-volume ratio and T1 relaxivity. (k) The relationship of r1 value and the number of effective magnetic metal ions on exposed facets. Reproduced with permission [160]. Copyright 2018, American Chemical Society.

Unpaired electron

Some magnetic ions, such as manganese and lanthanide ions own large amounts of unpaired electrons, great potential to increase q of IONP and regulate its T1 relaxivity. Theoretically, the more unpaired electrons, the higher q could be achieved. Compared to iron ions, those ions could be divided into two types, one exhibits longer τ and more electrons (including Mn, Gd, and Cu), the other only shows more electrons (such as Eu). In this section, we will discuss effect of single improvement of unpaired electrons on the T1 relaxivity of IONP. Europium (III) ions with 6 unpaired electrons and τ with the value of ∼10−14 s have been to be reported to improve the T1 relaxivity of IONP. Yang et al. synthesized Eu doped iron oxide (EuIO) nanocube with the size of 14 nm [213]. T1 relaxivity investigation indicates that r1 value and r2/r1 ratio of this EuIO nanocube are 36.79 ± 1.16 mM−1s−1 and 2.65, which is more suitable to act as a T1 CA compare to IONP. The elevation effect could be ascribed to higher chemical exchange of surface Eu (III) ions with nearby protons compare to iron ions.

Electronic relaxation time (τ)

According to SBM theory, magnetic nanoparticle with long τ exhibits high T1 relaxivity. Since Fe (II) ion exhibits fewer unpaired electrons and significantly shorter τ than Fe (III) ion (10−9–10−11 s and 10−12–10−13 s for Fe (III) and Fe (II) ion) [[214], [215], [216]], this structure deficiency limits efficiency of relaxation enhancement of IONP in form of magnetite and limits its application as T1 CA. Therefore, replacement of Fe (II) ions in magnetite by other magnetic ions with relative long τ may overcome the structure deficiency and increase the T1 relaxivity of magnetite. In the following section, we will introduce some recent attempts to replace Fe (II) ions in magnetite and discuss the effect of different replaced magnetic ions on the T1 relaxivity of magnetite.

Manganese

Manganese ions exhibit diverse valence state, such as +2, +3, +4, and +7. Among them, Mn (II) ion based probe has widely be used as a T1 CA in MRI contrast imaging due to its paramagnetic behavior. Compared to Fe (II) ion, Mn (II) ion owns higher unpaired electron (5 unpaired electrons for Mn (II) ion and 4 unpaired electrons for Fe (II) ion) and exhibit longer τs (10−8 s for Mn (II) ion and 10−12–10−13 s for Fe (II) ion) [217,218]. Therefore replacement of Fe (II) ion in magnetite by Mn (II) ion may efficient improve the coordination number and τs, optimizing its T1 relaxivity. Huang et al. synthesized manganese engineered IONP (MnIONP) to investigate the effect of introducing Mn (II) ion on its T1 contrast [153]. The T1 relaxivity assessment indicates that engineered IONP with Mn (II) ion could efficiently improve its T1 relaxivity. Gao and co-authors further engineered IONP with anisotropic morphologies by Mn (II) ion and investigated the influence of Mn (II) ion on their τs and T1 relaxivity (Fig. 11a–g) [51]. The electron paramagnetic resonance spectrum analyses indicate that introducing of Mn (II) ion can remarkably increase the τs of pure IONP regardless of the morphologies. τs of MnIONP with the shape of cube, octapod, and plate are about 0.26, 0.19, and 0.23 ns, which are approximately 2.17, 2.38, and 1.44 times higher than those of corresponding pure IONP, respectively. Coupled with the abundantly exposed magnetic ions, this τs elevation effect remarkably increase the T1 relaxivities of IO nanocube, octapod, and plate to 57.8, 62.1, and 22.4 mM−1s−1 at 0.5 T.

Fig. 11

Effect of τ on T1 relaxivity of IONP. (a) Schematic illustration of water chemical exchange and proton relaxation enhancement phenomena in magnetic systems with surface magnetic ions in IONP and MnIONP. TEM and EDX mapping images of MnIONP with the morphologies of (b) cubes, (c) octapods, and (d) plates, respectively. r1 values, r2/r1 analyses, and T1-weighted images of MnIONP with morphologies of (e) cubes, (f) octapods, and (g) plates, respectively. Reproduced with permission [51]. Copyright 2018, The Royal Society of Chemistry. (h) TEM and HRTEM images of Gd engineered IO nanoplate. (i) EDX line-scanning and (j) mapping of Gd engineered IO nanoplate. (k) T1-and T2-weighted phantom images of Gd engineered IO nanoplate at 3.0 and 9.4 T. (l) Comparison of r1 values of Gd engineered IO nanoplate, IO cubes, and Gd engineered IONP. Reproduced with permission [221]. Copyright 2015, American Chemical Society. Selected STEM-HAADF images of CuIONP with the Cu dopant ratio of (m) 1.7%, (n) 4%, and (o) 28%, respectively. (p) r1 values and (q) r2/r1 ratio analyses of CuIONP with different dopant ratio. Reproduced with permission [222]. Copyright 2019, American Chemical Society.

Gadolinium

Due to its large amounts of unpaired electrons and long τs, paramagnetic gadolinium ions (Gd3+) are widely used to construct T1 CAs and exhibit high T1 contrast through effective interacting with adjacent water protons. Compared to IONP, ultrasmall Gd2O3 nanoparticles and Gd2O3 nanoplate have been proved to exhibit more effectively T1 contrast with lower r2/r1 ratio due to the low magnetic susceptibility. These results inspire researchers to embed Gd2O3 nanocluster into IONP and investigate their T1 contrast [219]. Small sized gadolinium-embedded IONP (GdIONP) have been reported to show high T1 relaxivity, due to the spin-canting effects and introduction of Gd species [220]. GdIONP with the size of 4.8 nm showes T1 relaxivity with the value of 7.85 mM−1s−1 and r2/r1 ratio of 5.24 at 7 T. Gao and co-authors further engineered IO nanoplate with metal-rich exposed crystal surface by Gd ions to optimize its T1 relaxivity (Fig. 11h-l) [221]. Based on the synergistic effects of exposed metal-rich Fe3O4 (100) facet and embedded Gd2O3 clusters, this engineered IO nanoplate show an ultrahigh T1 relaxivity with the value of 61.5 mM−1s−1 and r2/r1 ratio of 2.4 at 0.5 T. The enhanced T1 relaxivity could be ascribed to two main reasons, those are exposed Fe and Gd ions and terminated Fe3O4 (100) basal plane. The exposed Fe and Gd ions could provide sufficient paramagnetic island and result in synergistic enhancement. In addition, the flatten surface is beneficial to the hoping of water proton on the surface of Gd engineered IO nanoplate and further improve the chemical exchange efficiency. It should be noted that its T1 relaxivity is highly dependent on the amount of Gd ions. r1 values increased from 46.7 ± 2.0 to 66.3 ± 3.1 mM−1s−1 with the rising of Gd percentage from 11% to 26%. We speculate that this result may be a consequence of increased τs caused by the increase amount of Gd ions.

Copper

Copper ions with the τs value of ∼10−9 have been reported to be a candidate T1 MRI CA in the form of CuO and CuS nanoparticles [[223], [224], [225], [226]]. In addition, the magnetization moment of copper based nanoparticles are significantly lower than that of IONP. Thus, engineering IONP by copper ions may improve its τs and reduce its magnetization moment, which may improve the T1 relaxivity of IONP. One attempt have been carried out to improve the T1 relaxivity of IONP by copper dopant (Fig. 11m-q) [222]. Herranz and co-author developed a copper doped extremely small IONP (CuIONP). When the dopant percentage of copper is 4%, CuIONP exhibits high T1 relaxivity with the r1 value and r2/r1 ratio of 15.7 mM−1s−1 and 2.1. The relative high r1 value and low r2/r1 ratio endow these engineered IONP with good T1 contrast to achieve MRI based angiography and tumor imaging.

Chemical exchange

T1 relaxivity of MRI CA is dependent on the chemical exchange efficiency between CAs and water protons at the interface. Surface ligand of CA could affect water diffusion, retention, and interaction with the magnetic centers, which determine τe, τR, τm, and the chemical exchange efficiency between IONP and water proton. A deep investigation of the effect of surface ligand on the T1 relaxivity of IONP can provide guidance in improving its T1 contrast.

Molecular weight

r1 value of IONP is mainly determined by the exchange rate of water proton in its inner sphere. Theoretically, IONP stabilized by low molecular weight (Mw) ligand would reduce the hydrodynamic volume, resulting in the accessibility reduction of water proton to the surface of IONP. Therefore, altering the Mw or chain length of the surface coating ligand seem to be the simplest way to adjust T1 relaxivity of IONP. In previous study, the relationship between Mw of polyethylene glycol (PEG) and T1 relaxivity have been investigated (Fig. 12a) [41]. It seems that r1 value increases with the rising of Mw of PEG. IONP coated by PEG with the Mw of 1000, 2000, 5000 Da show higher r1 values than that with Mw of 550 and 750 Da. This Mw related T1 relaxivity also been reported in other research. Tromsdorf et al. found that r1 value of PEG coated IONP gradually increases with rising of Mw from 350 to 1100 Da, but decreases when Mw increases to 2000 Da [199]. The r1 value decreasing at 2000 Da may be attributed to the aggregation of IONP.

Fig. 12

Effect of chemical exchange efficiency on T1 relaxivity of IONP. (a) r1 values of IONP coated by DSPE-PEG with different molecular weights. Reproduced with permission [41]. Copyright 2007, John Wiley & Sons, Inc. TEM images and schematic cartoon of EuIONP with surface ligand of (b) citrate (Cit), (c) alendronate (Ale), and (d) PMAO-PEG (PP). (e) r1 relaxivities of EuIONPs coated with different surface ligands as a function of contact angle. (f) T1-weighted images of EuIONPs coated with Cit, Ale, and PP with different magnetic ions concentration. Reproduced with permission [228]. Copyright 2018, American Chemical Society. (g) TEM images of IONP coated by GO. (h) Linear fitting of 1/T2 and (i) 1/T1 of the reference IONP, IOPNs coated by GO, and IONP coated by NH2-cis-aconitic acid-DOX and GO. Reproduced with permission [232]. Copyright 2019, The Royal Society of Chemistry.

Hydrophilicity

Hydrophilicity increasing of surface area benefits the water access to the surface of IONP and elevates the chemical exchange efficiency. Bao et al. crosslinked tannic-acid coated IONP onto bovine serum albumin to form nanocluster [227]. The relaxivity measurement indicates that r1 value of nanocluster is about 2 times higher than that of free nanoparticles. This r1 value elevation could be ascribed to the high hydrophilicity around IONP and reduced mobility within the nanoclusters. In another research, Seo and co-authors modified Eu doped IONP (EuIONP) with citrate (Cit), alendronate (Ale), and PMAO/PEG (PP), which showed different hydrophilic (Fig. 12b–f) [228]. The hydrophobicity of all samples were measured by the contact angle. The contact angles of Cit, Ale, and PP are immeasurable, 9.8°, and 112.4°, which means the hydrophobicity order is Cit < Ale « PP. This result is highly consistent with the changing of T1 relaxivity with the order of EuIONP-Cit > EuIONP-Ale » EuIONP-PP.

Electronic and magnetic interactions

It is well known that graphene oxide (GO) could interact with semiconducting oxide nanoparticles through excited-state electron transfer. The charge-transfer between electronic and magnetic interactions have been reported to existence between IONP and GO [[229], [230], [231]]. This unique phenomenon may affect the chemical exchange interaction between GO coated IONP and surrounding water proton. Recently, the effect of GO on the T1 relaxivity of IONP have been investigated (Fig. 12g–i) [232]. T1 relaxivity assessment indicate that GO coated IONP show T1 relaxivity with the value of 2.82 mM−1s−1, which is about 6 times higher than that of free IONP with the r1 value of 0.46 mM−1s−1. The enhancement on T1 relaxivity of IONP is believed to contribute to two mainly reasons. One is the energy exchange and charge transfer between GO and IONP, the other is the homogeneous dispersion of IONP. Notably, this research indicates that T2 relaxivity of IONP could be suppressed by coating of GO, which could be ascribed to the limitation effect of GO on the spin of proton and local magnetic field. The decreasing effect on T2 relaxivity could reduce r2/r1 ratio, further improving the availability of GO coated IONP as T1 CA.

IONP based dual-modal imaging

Traditional strategies to achieve sensitive MRI contrast imaging by IONP is increasing its T1 or T2 relaxivity, which could improve the signal difference between normal tissue and lesion. Although T1 image shows high tissue resolution and T2 image exhibits high feasibility of detection under the assistant of single modal CA, such single mode CA are not yet perfect and facing huge challenges in accurate imaging of tiny lesion. To improve the sensitivity and accuracy, simultaneous acquisitions of T1 and T2 CA have attract considerable interest. Combination of T1 and T2 modal can provide complementary information for doctor to differentiate lesion from normal tissue with improved sensitivity and accuracy. In recent ten years, a number of IONP based T1/T2 dual-modal MRI CAs have been developed [52,57,[233], [234], [235], [236], [237], [238]]. Besides to develop IONP with MRI based dual-modal contrast, direct conjugation secondary even tertiary imaging moiety with IONP is another effective method to improve its contrast efficiency. Currently, there are several representative imaging modalities have been used in clinical or preclinical research, including MRI, SPECT, PET, CT, FL imaging, US imaging, and PA imaging. Introducing of these imaging modality to IONP can effectively overcome the instrumental limitation of MRI and increase its sensitivity to satisfy the requirement of accurate imaging. In the following section, we will focus on introducing recent method to fabricate IONP with dual-modality contrast to improve its diagnosis efficiency.

T1-T2 dual-modality

For the purpose of dual-modal enhancement, the straightforward method to construct IONP based T1/T2 dual-modal CA is introducing T1 contrast domain into IONP. Since the direct chemical exchange to water proton is the basis of T1 contrast imaging, this strategy often adopts as the T1 contrast domain exposed on the outer shell [56,[239], [240], [241]]. Im et al. synthesized MnO coated IONP with core-shell, dumbbell-like, and flower structure and assessed their capacity on simultaneous enhancement on both T1 and T2 modal (Fig. 13a–d) [242]. The r1 values and r2/r1 ratios of Fe3O4@MnO, Fe3O4/MnO dumbbell, and Fe3O4/MnO flower are 1.3, 1.4, and 0.6 mM−1s−1 and 28, 56, and 235, which are not suit o applied as the T1/T2 dual-modal CAs to early and accurate diagnosis. The relatively high r2/r1 ratio of this kind of core-shell structure have been found in IONP coated by Gd2O3 shell as well (Fig. 13e–h) [243]. Sun et al. constructed a yolk-like nanostructure using IONP as core and mesoporous Gd2O3 layer as shell. The r2/r1 ratio of this nanostructure is about 48.9, which is significantly higher than that of typical T1/T2 dual-modal CA. These results could be ascribed to the fact that IONP could slow the T1 spin fluctuation, which is ineffective for water proton relaxation and leads to a low T1 MRI signal. Cheon and co-worker discovered that the speed of electron spin fluctuation of T1 contrast domain is determined by the distance between T1 and T2 contrast domains (Fig. 13i-n) [58]. When the distance of the T1 and T2 domains is above a certain value, the electron spin fluctuation of T1 domain accelerate water proton relaxation and result in a stronger T1 MRI signal. Thus, tuning the distance between T1 domain and IONP have been adopted to construct effective T1/T2 dual-modal CAs [244]. Cheon and co-workers utilized SiO2 as the isolation layer to separate T2 domain (MnFe2O4 nanoparticles) and T1 domain (Gd2O(CO3)2 shell) (Fig. 14a–c) [245]. Along with the thickness increase of SiO2 layer in the order 4, 8, 12, 16, and 20 nm, the r1 value changes in the order 2.0, 4.0, 25.1, 33.1, and 32.5 mM−1s−1. On contrary, T2 relaxivity decreases with the increasing of separating layer due to reduction effect on local magnetic field caused by SiO2 layer. These results endow this core-shell nanostructure as a potential T1/T2 dual-modal CA. The generality of introduction of SiO2 isolating layer to separate T1 domain and T2 domain have been demonstrated by the same group [60]. Various T2 domains, such as Fe3O4 and CoFe2O4, and T1 domains, including Eu2O(CO3)2, Dy2O(CO3)2, and [ImH][Mn(BTC)(H2O)] have been combined to construct artifact filtering imaging agent (AFIA). All of the AFIAs exhibit simultaneous remarkable T1 and T2 signal enhancement. Except to SiO2, polydopamine have been discovered to be another isolation shell to construct IONP based T1/T2 dual-modal CAs [233]. Another structure to regulate the distance between T1 domain and IONP is Janus structure. Cheng et al. synthesized dumbbell hybrid nanotrimers and utilized Pt nanocube to isolate Gd chelators attached Au nanoparticles and IONP (Fig. 14d–f) [236]. The linking distance between Au and IONP have been accurately regulated with the value of 5.4, 9.6, 10.7, and 12.9 nm. Consistent with the previous study, the dumbbell nanotrimer shows increased T1 relaxivity with the rising of distance between Au and IONP due to the reduced magnetic coupling between Gd and IONP. Based on their proper distance, dumbbell nanotrimers with the linking distance of 10.7 and 12.9 show obvious simultaneous T1 and T2 signal enhancement effect. Altering the structure of IONP have been proved to be effective strategy to adjust the contrast of IONP in T1 and T2 modal. The critical parameter, which determine a given IONP is T1 or T2 dominated MRI CA, is its r2/r1 ratio. Previous study indicate that engineering the structure of IONP, especially morphology, could increase its T1 relaxivity and reduce its T2 relaxivity, which endow it with favorable r2/r1 ratio to exhibit T1/T2 dual-modal contrast.

Fig. 13

Effect of direct coating IONP with T1 contrast moiety on its T2/T1 dual-modality contrast capacity. TEM images of (a) Fe3O4, (b) Fe3O4@MnO, (c) Fe3O4/MnO dumbbell nanoparticles, respectively. (d) Comparison of r1 values, r2 values, and r2/r1 ratios of different nanoparticles. Reproduced with permission [242]. Copyright 2013, Elsevier Ltd. (e) TEM and EDX mapping images of Gd2O3 coated IONP. (f) r1 and (g) r2 value of Gd2O3 coated IONP. (h) T1-and T2-weighted MR images obtained from Gd2O3 coated IONP at varied concentrations. Reproduced with permission [243]. Copyright 2017, American Chemical Society. (i) Schematic cartoon illustrates the distance dependent magnetic resonance tuning. (j) A schematic representation of the nanoscale ruler based on the MRET effect. (k) T1-weighted and color mapped MR images of solution containing enhancer and quencher with varied distances. (l) A plot of r1 values versus separation distance. (m) The r1 values of the enhancer with a separation distance of 2, 7, and 12 nm at various larmor frequencies. (n) A plot of T1e versus the separation distance at 15 K. Reproduced with permission [58]. Copyright 2017, Nature Publishing Group. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 14

IONP based core-shell T1/T2 dual-modal CA. (a) Schematic and TEM image of core-shell type dual-modal CA. (b) TEM images of dual-modal CAs with different separating layer thickness. (c) r1 and r2 values of dual-modal CAs, MnFe2O4, Gd-DTPA, and Feridex. Reproduced with permission [245]. Copyright 2010, American Chemical Society. (d) Schematic illustrate the magnetic coupling of T1 and T2 CAs in core-shell and dumbbell structures. (e) Illustration of constructions of four different types of dumbbell-like or dumbbell heterostructures. (f) T1-and T2-weighted MRI images of dumbbell hybrid heterostructures. Reproduced with permission [236]. Copyright 2014, American Chemical Society.

Recently, The T1 and T2 contrast ability of IO nanoplate with the thickness of 4.8 nm have been investigated [155]. Its exposed Fe3O4(111) facet provided sufficient iron ions to achieve chemical exchange to water proton, which increased its T1 contrast ability. Besides, the thin thickness improved the spin canting effect and reduce its magnetization and T2 contrast ability. Due to the lowering T2 contribution and enhancing T1 contribution, IO nanoplate with the thickness of 4.8 nm show significant signal enhancement in both T1 and T2 contrast imaging. This morphology dependent typical T1 and T2 dual-modal contrast have also been found in IONP with other morphology, such as nanocube [212].

In addition, dopant of paramagnetic ions, especially gadolinium ions into the crystal structure of IONP have been proved to be another effective method to construct IONP based T1/T2 dual-modal CA. Gao and co-authors embedded gadolinium cluster into IONP, which lead to a synergistic enhancement on T1 and T2 relaxivity of IONP (Fig. 15) [246]. The r1 value of gadolinium doped IONP (GdIONP) is ∼69.5 mM−1s−1 in terms of Gd; and the r2 value of GdIONP is about 146.5 mM−1s−1 in terms of Fe. Due to the dopant of gadolinium ions, the r1 and r2 values of GdIONP are both higher than Gd2O3 and IONP and show obvious signal enhancement in both T1 and T2 modal. Dopant other paramagnetic ions into IONP, such as manganese and europium, have been proved to achieve enhancement of T1 and T2 signal as well [213,234]. Xiao et al., have synthesized DSPE-PEG coated MnIONP as T1/T2 dual-modal CA [247]. They found that MnIONP coated with DSPE-PEG with the mass ratio of 1:20 showed harmonious T1 and T2 relaxivity and could be considered as an excellent candidate as T1/T2 dual-modal CA.

Fig. 15

IONP based T1/T2 dual-modal CA. (a) TEM and (b) EDX mapping images of GdIONP. T1-and T2-weighted MR images of (c) GdIONP, (d) IONP, and (e) Gd2O3 nanoparticles, respectively. The analyses of (f) T2 and (g) T1 relaxation rate of GdIONP, IONP, and Gd2O3 nanoparticles. Reproduced with permission [246]. Copyright 2012, John Wiley & Sons, Inc.

MRI-FL modality

Fluorescence imaging with high sensitivity is the earliest modality to be introduced into IONP to ameliorate its contrast sensitivity. Initially, the fluorescence dye or quantum dots were directly conjugated with IONP to form the MRI-fluorescence dual-modality CAs [74,[248], [249], [250]]. Xu and co-authors used IONP as seeds to grow CdSe QDs on its surface and obtained hybrid IONP with magnetic and fluorescence property [251]. This hybrid IONP could successfully achieve fluorescence imaging on cellular level, while the existed energy transfer between the fluorescent domain and IONP result in a fluorescence quenching and limit its application as a MRI-fluorescence dual-modality CA in vivo. Since this quench effect is highly dependent on the distance between the fluorescent domain and IONP, one can improve its fluorescent property by increasing this distance. Lee et al. attached IONP on dye-doped silica nanoparticles to form “core-satellite” nanostructure with MRI and fluorescence imaging capacity (Fig. 16a–f) [73]. The fluorescence signal of this core-satellite nanostructure is enhanced by 1.7 times compare to the directly dye conjugated IONP with the assistance of separating effect of silica, which reduces quenching effect between IONP and fluorescent dye. Moreover, the core-satellite nanostructure with assembled IONP show increased T2 relaxivity with the r2 value of 397 mM−1s−1. The perfect MRI and fluorescent contrast ability endow this nanostructure to conduct sensitive imaging of sub-millimeter cellular clusters.

Fig. 16

IONP based MRI-FL and MRI-PA dual-modal CA. (a) Schematic diagram for the synthesis of DySiO2-(Fe3O4)n nanoparticles. TEM images of (b) DySiO2, (c) IONP, and (d) DySiO2-(Fe3O4)n nanoparticles. (e) Analyses of T2 relaxation rates and T2-weighted MRI images of DySiO2-(Fe3O4)n nanoparticles and free IONP. (f) Photoluminescence spectra of DySiO2-(Fe3O4)n nanoparticles and dye linked IONP. Reproduced with permission [73]. Copyright 2006, John Wiley & Sons, Inc. (g) TEM and EDX mapping images of CP-IO nanocomposites. (h) T2-weighted MR images and (i) T relaxation rate analyses of CP-IO nanocomposites. (j) Plots of the simulated temperature distribution in CP-IO, CP, and MIX nanoparticles at 100 ns. (k) Plots of the simulated PA signal transmission from surface of nanoparticles to surrounding water environment. Reproduced with permission [79]. Copyright 2018, John Wiley & Sons, Inc.

MRI-PA modality

As a relatively new imaging technique, PA imaging exhibits high spatial resolution and fast real-time scan. Over the past decades, there are numerous materials, including gold [69,[252], [253], [254]], indocyanine green (ICG) [78,255], polypyrrole (PPy) [80], and diketopyrrolopyrrole (DPP) [79], have been applied to incorporating the advantages of PA and MRI into a single nanostructures. Thawani et al. combined ICG and IONP to develop a stable nano-cluster [78]. The as-prepared nanostructure exhibited obvious enhancement effect on both MRI-PA imaging and realized effective MRI-PA assistant tumor detection on mice model. Interestingly, IONP was discovered to increase the signal of traditional PA imaging CA. Recently, Liu and co-workers integrated DPP-based conjugated polymer (CP) and IONP into close proximity by amphiphilic polymer to fabricate MRI-PA dual-modal CA (Fig. 16g–k) [79]. With the maintained T2 relaxivity of IONP, the obtained product can produce remarkable enhanced PA signal. A 45% PA signal intensity amplification is observed in the obtained nanocomposites compare to the bare CP due to the addition of IONP, which boosts the heat generation and heat dissipation pathways in the production of PA signal.

MRI-PET/SPECT dual-modality

PET or SPECT, those signal is generated by gamma-rays emitted from decaying radioisotopes (e.g. 64Cu and 124I for PET and 99mTc and 131I for SPECT), show high sensitivity but relatively poor spatial resolution. Combination of PET or SPECT with MRI have been proved to improve the contrast imaging efficiency of IONP [[256], [257], [258]]. 124I have been successfully linked on MnIONP through chemical conjugation (Fig. 17a–c) [259]. The resultant hybrid CA maintain equivalent contrast effect corresponding to the CA of single MRI and PET modality. This unique feature endow this dual-modality CA with high spatial resolution and high sensitivity to perform lymph node imaging. On the assistant of a fused MRI-PET image, researchers can easily differentiate the tiny brachial lymph from axillary lymph node with the diameter of a few millimeters. Meanwhile, many attempts have been conducted to fabricate IONP based MRI-SPECT dual-modality CAs. IONP radiolabeled with 99mTc is reported as a MRI-SPECT CA (Fig. 17d–h) [66]. The 99mTc labeled IONP with retained in vivo MRI contrast property shows SPECT contrast imaging. Coupled with excellent stability, this CA provides sufficient signal to successfully achieve liver and spleen imaging on mice model. On the basis of long half-life, 125I have been investigated as the SPECT imaging moiety to construct IONP based MRI/SPECT dual-modal CAs.

Fig. 17

IONP based MRI-PET/SPECT dual-modal CAs. (a) Illustration of preparation of 124I linked MnIONP. PET/MRI images of SLNs in a rat at 1 h post injection of 124I linked MnIONP into the right forepaw in (b) coronal and (c) transverse view. Reproduced with permission [259]. Copyright 2008, John Wiley & Sons, Inc. TEM images of (d) IONP and (e) 99mTc radiolabeled IONP. T2-weighted MRI images (f) before and (g) 15 min postinjection of 99mTc radiolabeled IONP. (h) SPECT-CT image of the same mice in a similar view 45 min postinjection. Reproduced with permission [66]. Copyright 2011, American Chemical Society.

MRI-CT modality

Considering CT is advantageous in regard to its high resolution and ease of forming 3D visual reconstruction of tissue of interest, introducing CT contrast moiety onto IONP have been used to improve its imaging efficiency [[260], [261], [262], [263]]. The common strategy is synthesis of Au nanomaterials hybrided IONP. Zhao et al. successfully synthesized strawberry-like Fe3O4–Au hybrid nanoparticle with enhanced X-ray attenuation and magnetic properties as an accurate MRI-CT dual-modal CA to distinguish the grade of live disease (Fig. 18a–e) [264]. Except to Au, the possibility of fabrication IONP based MRI-CT dual-modal CAs with other CT contrast moiety have been investigated as well. Multifunctional Fe3O4/TaOx core/shell nanoparticle have been synthesized by Hyeon and co-workers (Fig. 18f–i) [265]. On the basis of its low cost and high biocompatibility, TaOx is more promising as the CT contrast moiety to construct MRI-CT dual-modal CA compared to Au nanostructure. Fe3O4/TaOx core/shell nanostructure exhibites remarkable enhancement effect in both X-ray CT and MRI phantom images, which make it as an effective MRI-CT dual-modal CA in the detection of newly formed blood vessels of and microenvironment of tumor.

Fig. 18

IONP based MRI-CT dual-modal CAs. (a) TEM images of Fe3O4–Au hybrid nanoparticles. (b) X-ray attenuation assay and (c) T2 ralaxation rate analysis of Fe3O4–Au hybrid nanoparticles. (d) T2-weighted MRI and (e) CT images of rat pre and post injection in transversal view. Reproduced with permission [264]. Copyright 2015, Elsevier Ltd. (f) CT phantom images and HU values of Fe3O4/TaOx nanoparticles. (g) T2-weighted MRI images and r2 values of Fe3O4/TaOx nanoparticles. In vivo (h) CT image and (i) T2-weighted image of rat 24 h after injection of Fe3O4/TaOx nanoparticles. Reproduced with permission [265]. Copyright 2012, American Chemical Society.

IONP based responsive MRI CAs

The basis of MRI CA based disease diagnosis is the signal difference caused by the different accumulation amount of CA in lesion and surrounding tissue. Although surface modification could optimize the in vivo behavior of CA and improve their accumulation in lesion, the mis-accumulation in normal tissue could not be avoided Besides, the traditional CAs are “always on” systems that generate MR signal enhancement regardless of their location, which may result in poor target to background signal difference. Recently, responsive MRI CA have been designed to respond to specific changes in the surrounding physiological microenvironments and specific physical conditions [94,[266], [267], [268], [269], [270]]. Compared to the traditional MRI CA with fixed contrast capacity in both normal and lesion tissue, responsive MRI CAs are designed to switch “on” and “off” T1 or T2 signal change in response to a specific stimulus of lesion. Generally, responsive MRI CA could be divided into three types, those are active, recovery, and switchable. Discussion on the recent development of strategy to construct IONP based responsive MRI CA to assist doctors to achieve accurate and early lesion diagnosis is urgent and meaningful.

Active

Compared to single IONP, assembled architecture shows remarkably increased T2 contrast ability. Modifying nanoparticles with specific molecular have been proved to be an effective strategy to implement their self-assembly under stimuli [271]. This unique phenomenon provided opportunity for IONP to complete the conversion from single state to assembled state and achieve T2 active MRI imaging. Jasanoff and Co-worker engineered IONP with fused C2 domains of synaptotagmin 1 (C2AB), which could naturally respond to the concentration change of Ca2+ (Fig. 19a–e) [272]. Based on the two binding sites for Ca2+ in C2AB, this IONP could aggregate in the presence of Ca2+ and promote its T2 contrast. Along with the increasing of Ca2+ concentration from 0 to 1.2 mM, the r2 value elevates from 151 to 261 mM−1s−1. Base on the calcium-dependent r2 change, this nanomaterial could be considered as a potential CA to monitor the dynamic change of Ca2+ concentration in brain. Additionally, IONP based nanostructure have been reported to respond to matrix metalloproteinase (MMP) enzymes. In this work, IONP bearing complementary azide and alkyne click moieties are individually prepared [273]. To achieve MMP response, a MMP responded peptide was conjugated on the surface of IONP to block the activity click reaction. In vitro study indicates that the two individual IONP successfully forms nanocluster at the present of MMP and result in a 160% increase in T2 activity. Recently, a cancer biomarker, glutathione (GSH), responsive IONP based active probe was developed (Fig. 19f–h) [274]. With the existence of GSH, the disulfide bond on the surface of IONP is reduced, inducing the aggregate of IONP. The status conversion from single state to aggregation state results in the interlocked responses of both T1 and T2 signals and is utilized to quantitatively map the GSH within brain gliomas.

Fig. 19

IONP based intelligent CA with responsive signal active capacity. (a) Schematic cartoon illustrate the sensor design based on lipid coated IONP and C2AB. (b) AFM images of IONP based calcium-responsive nanostructure after exposure to 0 or 1 mM Ca2+. (c) r2 values and (d) T2-weighted MRI images of IONP based calcium-responsive nanostructure after exposure to different Ca2+ concentrations. (e) r2 changes observed in HEPES buffer over multiple cycles of calcium or EDTA addition. Reproduced with permission [272]. Copyright 2018, Nature Publishing Group. (f) Schematic drawings to show molecular mechanism of GSH-induced agglomeration of intelligent probe. (g) Temporal evolution of ΔR1 and ΔR2 for intelligent probe and peptide modified IONP during the incubation with GSH. (h) GSH concentration dependent ΔR1 and ΔR2 for intelligent probe and peptide modified IONP. Reproduced with permission [274]. Copyright 2021, John Wiley & Sons, Inc.

Recovery

As a typical T2 CA, IONP have been proved to quench the T1 relaxivity of Gd chelate [275]. It has been reported that conjugating Gd chelates on IONP with bio-response linker could achieve T1 relaxivity recovery (Fig. 20a and b) [276]. In this study, the Gd chelate was conjugated on the silica coated IONP by MMP-2 responded peptides. To amplify the signal, dendrimers are introduced to increase the number of conjugated Gd chelate. MMP-2 successfully cleaves the peptide, resulting in release of Gd chelate from the local magnetic field generated by IONP and T1 signal recovery. Based on the bio-response procedure, the detection limit of these systems to MMP-2 could be reduced to 0.5 nM achieving MMP sensing both at cellular level and xenograft tumors. T2 relaxivity of IONP could be limited by a dense T1 CAs shell as well. Introducing an environment responsive shell with T1 contrast on the surface of IONP can initially shield its T1 and T2 relaxivity and achieve signal recovery under stimuli. Kim et al., developed a T1/T2 dual mode recovery imaging probe by coating IONP with redox-responsive paramagnetic Mn3O4 nano shell (Fig. 20c–g) [277]. Due to the strong silencing effect between Fe3O4 core and Mn3O4 shell, T1 and T2 contrast of this nanostructure are remarkably limited. Upon introduction to a tumor intracellular reducing environment, the Mn3O4 shell decomposed to release free Mn2+ ions. This structure alteration resulted in exposure of the interior Fe3O4 core to aqueous environment and recovery of T1 and T2 relaxivity.

Fig. 20

IONP based intelligent CA with responsive signal recovery capacity. (a) Schematic illustration of MMP-2 detection using IONP based on T1 relaxivity recovery. (b) T1-weighted MRI images of IONP based T1 recovery system incubated with MMP-2. Reproduced with permission [276]. Copyright 2017, American Chemical Society. (c) Schematic illustration of the redox-responsive activatable nanoshell. (d) T1-and (e) T2-weighted MRI images of responsive system after treatment of GSH and non-treatment. Relaxivity plot of (f) T1 and (g) T2 relaxation rate analyses. Reproduced with permission [277]. Copyright 2016, Elsevier Ltd.

Switchable

Recently, IONP based environment responsive MRI CA with modal switchable capacity have been developed to further increase the signal difference between lesion and normal tissue. There are two modal of switchable responsive MRI contrast imaging: T2 contrast switch to T1 contrast (mode I) and T1 contrast switch to T2 contrast (mode II). The strategy to develop mode I MRI CAs is construction IONP based nanocluster or nanoaggregation by responsive ligand. This kind of CA exist as nanocluster in blood circulation and normal tissue while collapse to dispersive nanoparticle, resulting in the T2 contrast ability decrease and T1 contrast capacity increase. On contrary, mode II CA is designed by in situ assemble of specific modified dispersive nanoparticles. Generally, ultrasmall IONP, typical with the size below 5 nm, have been considered as T1 CAs. With specific surface modification, these nanoparticles form the nanoaggregation when it reach to the tumor tissue and can be readily enhance T2 signal [[88], [89], [90]].

T/Tswitchable

Lu et al. fabricated a pH-sensitive IO nanocluster through linking IONP with i-motif DNA-derived pH-responsive linkers (Fig. 21a–g) [278]. Along with the decrease of pH value from 7.4 to 5.5, the state of this nanocluster converts from assembly to disassembly, resulting in drastic decrease in T2 relaxivity while increase in T1 relaxivity. This pH dependent MRI contrast enable this nanocluster to successfully achieve T2/T1 switchable contrast imaging and detected tiny hepatocellular carcinoma. Besides to the pH-response T2/T1 switchable CAs, redox environment have been applied as another trigger to accomplish sensitive T2/T1 switchable MRI imaging [88]. Ultrasmall IONP were cross-linked into nanocluster through the S–S bond of Cys. On the basis of the fact that the reductive environment of tumor can rupture the S–S bond, this nanocluster would dissociated into single IONP at tumor site. The state conversion lead to this nanocluster change from T2 dominate to T1 dominate and undergo T2/T1 switchable MR imaging of tumor.

Fig. 21

IONP based modality switchable CA. (a) Schematic illustration of intelligent system based modality switchable diagnosis of HCC. TEM images of (b) small IONP, (c) intelligent system in PBS (pH 7.4), and (d) intelligent system in MES (pH 5.5). (e) Kinetic analysis of intelligent system disassembly upon the pH change from 7.4 to 6.5/5.5. (f) T1-weighted images and (g) relaxivity of intelligent system in PBS (pH 7.4) and MES (pH 6.5 and 5.5). Reproduced with permission [278]. Copyright 2018, American Chemical Society. (h) Schematic illustration of the synthesis of IONP based intelligent system for enhanced retention and tunable T1/T2-weighted MR imaging of inflammatory arthritis. (i) T1 relaxation rates, (j) T2 relaxation rate, and (k) T1-weighted, (l) T2-weighted MR image of IONP based intelligent system under 405 nm laser irradiation (1.0 W cm−2) for different time. Reproduced with permission [279]. Copyright 2019, John Wiley & Sons, Inc.

Compared to the nanocluster, ultrasmall IONP can easily extravasate from the tumor vasculature and diffuse into tumor tissue. Thus, IONP based switchable MRI CA show single state in delivery process and aggregation state at tumor have been investigated. Mao and co-authors discover that IONP with the size of 3.5 nm can effectively extravasate through tumor vessel and penetrates into tumor [90]. Based on the tight interstitial space, interstitial pressure, and acidic intratumoral environment, this ultrafine IONP could form cluster. The MRI investigation reveals that ultrafine IONP shows T1 dominate contrast in the tumor vasculature and followed by emerging as T2 CAs in tumor after injection. To further increase the controllability of this clustering process and MRI contrast modal conversion, light sensitive T1/T2 switchable CA has been developed recently (Fig. 21h-l) [279]. Li et al. modified ultrasmall IONP with light-addressable unit DA via PEG spacer. The generated Fe3O4-PEG-DA shows high T1 relaxivity with the value of 3.83 mM−1s−1 and could be used as a sensitive T1 CA. Once laser irradiation with different duration time is applied, IO nanocluster with different aggregation degree can be formed. After 12 min irradiation, the size of IO cluster could reach 798.4 nm and result in the obviously increase of T2 relaxivity (from 9.04 to 31.60 mM−1s−1) while reduction of T1 relaxivity (from 3.83 to 1.61 mM−1s−1), showing a typical T1/T2 switchable process.

Optimization in vivo behavior of IONP

Apart from r1 or r2 value of MRI CA, the sensitivity and accuracy of MRI contrast imaging are also determined by the contrast efficiency of CA in vivo. Simply, the contrast efficiency of CA is dependent on the signal difference between normal tissue and lesion, which is most straightforward determined by the amount of CA in lesion. The surface ligand of IONP could affect its circulation behavior, accumulation amount in lesion, and uptake of tumor cells [96], thus IONP with optimized surface structure show perfect in vivo behavior and achieve high efficient contrast imaging.

Polymers

Polymers, including natural and synthetic polymer, functionalized IONP have drawn much attention due to the improvement on the structural stability, pharmacokinetics, and biodistribution of IONP.

Natural polymer

Natural polymeric ligands, such as polysaccharides and protein, commonly containing multiple active groups strongly binding with IONP for stability improvement in harsh biological environments. Dextran, a typical polysaccharide, has been extensively used for coating IONP with enhanced stability and functionality. The Dextran cross-linked IONP show negligible changes in size and morphology in the blood circulation [280]. Notably, the saccharides are natural signal molecular on the cell surface, modification IONP with polysaccharides exhibits targeting ability to some specific issue. Kamruzzaman et al. observed that lactobionic acid (LA) modified IONP showed the capacity to target hepatocytes [281]. The uptake amount of LA modified IONP is significantly higher than unmodified and maltotrionic acid modified IONP. Further in vivo targeting capacity analyses indicate LA modified IONP only result in the signal changes in liver cells, revealing its capacity to achieve specific imaging of liver. Recently, a great effort has been devoted to functionalization of IONP with proteins with improved stability, prolong circulation time, and targeted imaging. Protein coating have been proved to improve physical stability and reduce aggregation of IONP. Casein coated IONP remained stable for months without any aggregation or size change after incubation in medium containing serum [182]. Besides, Quan and co-worker discovered that coating IONP with HSA could reduce the efficiency of IONP removed by reticulo endothelial system (RES) and prolong its circulation time in blood [282]. The circulation half-life of IONP with HSA coating is about 87 min, which is significantly longer than that without HSA modification (3 min). Since antibody is a natural protein, coating IONP with protein can further achieve targeted imaging and improve diagnosis sensitivity. Chekhonin et al., synthesized monoclonal antibodies against VEGF (mAbVEGF) conjugated BSA-modified IONP and assessed its ability to achieve targeted tumor diagnosis [283]. It appears that this antibody modified IONP successfully enable visualization of glioma microvessels with sensitive detection of neoangiogenic areas after administration.

Synthetic polymer

To improve the stability of IONP, coating IONP with synthetic polymer, such as polyethylene glycol (PEG), poly(lactic-co-glycolic acid) (PLGA) and Polyvinylpyrrolidone (PVP) have got more and more attention. Among all synthetic polymers, PEG as a US FDA approved ligand is extensively used to modify IONP due to its advantage on decrease the adsorption of protein in serum and prolonging its circulation time in the body [99,284]. Weller and co-author investigated phagocytosis of PEG modified (PEGylated) IONP by J774 macrophages (Fig. 22a–d) [199]. They found that the uptake amount of PEGylated IONP was significantly lower than that of clinically used Resovist (commercial IONP based MRI CAs), revealing a obvious decrease of unspecific uptake into cells of RES and endowing IONP with long circulation time. In another study, PEG coated 3 nm IONP displayed lower toxicity but longer circulation time than clinically used GBCAs [46]. Additionally, these unique features endow PEGylated IONP with the ability to achieve dynamic time resolved MR angiography in rats. Because of its high stability, biodegradability, and biocompatibility, another synthetic polymer, PLGA, have been widely applied in construction IONP based MRI CA. Wang et al., developed IO loaded PLGA-mPEG nanoparticles as MRI CAs [285]. Due to their optimized surface structure, this IO loaded PLGA-mPEG nanoparticle show higher contrast effect and longer half-life circulation time in comparison with Resovist. Modifying PLGA coated IONP with target motif can further improve contrast efficiency of IONP. Wang and co-workers functionalized PLGA-coated IONP by Arg-Gly-Asp (RGD) peptide [286]. Since RGD peptide has a tendency to bind activated platelets at the thrombus site, this MRI CA shows high affinity with thrombi and serves as a sensitive CA for early thrombi detection. Additionally, PVP attracted much interest in construction IONP based MRI CA because of non-toxicity, low cost, and antiviral properties. Chen and co-workers synthesized PVP coated IONP by a thermal decomposition method [287]. The PVP coated IONP exhibits high solubility and stability in various buffer and serum. They discover that macrophages take up greater amounts of large core PVP coated IONP than Feridex (clinical used MRI CAs), which results in the slightly higher contrast signal caused by PVP coated IONP than Feridex.

Fig. 22

Effect of surface ligand on in vivo contrast efficiency of IONP. Prussian blue staining of J774 macrophages after incubation with IONP coated by PEG with the molecular weight of (a) 350, (b) PEG 1100, (c) PEG 2000, and (d) Resovist. Reproduced with permission [199]. Copyright 2009, American Chemical Society. Prussian blue staining of HUVECs incubated with (e) PEGylated IONP, (f) RGD modified IONP, (g) RGD modified IONP plus free c(RGDyK), and (h) control group for 12 h at Fe concentration of 100 μg/mL. Scale bar: 20 μm for all images. (i) In vivo biodistribution of Fe in major organs. (j) T1-weighted MRI images of mice bearing tumor before and after administration of RGD modified IONP. Reproduced with permission [298]. Copyright 2016, Ivyspring International Publisher. (k) TEM image and schematic cartoon of surface structure of IO@ZDS. (l) Chromatograms of IONP in serum binding test. (m) UV–Vis spectrum of IO@ZDS with the increase of storage time. Reproduced with permission [300]. Copyright 2012, American Chemical Society. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Target ligand

IONP modified with normal small molecular is easily uptake by macrophage and clear by RES, which decreases its contrast efficiency in vivo. To increase the accumulation of IONP in lesion, especially tumor, much attempt have been performed by coating IONP with targeted motif. Targeting ligands, including antibodies [109,288,289], peptides [290,291], aptamers [292,293], folic acids [294,295], and hyaluronic acid [296,297] have been used to modify IONP to achieve targeted MR contrast imaging. Jia et al. modified ultra-small IONP by c(RGDyk) molecular (Fig. 22e–j) [298]. Owing to the RGD modified surface could specifically recognize tumor angiogenesis, the RGD modified IONP shows high accumulation in tumor site, which results in remarkable MR signal enhancement at hepatic tumor and low detection limit with the size of 2.2 mm. Despite each targeting ligand enables IONP with targeting ability, the type of ligand can significantly affect its targeting capacities. For example, due to the lack of consistent covalent bonding site antibodies, which is difficult to controllable attach on the surface of IONP [299]. This defect limits the presentation of antibody binding sites and lower, even hinder, its binding activity, resulting the partial loss of target capacity of IONP.

Zwitterionic molecular

When IONP enters the physiological environment, serum proteins rapidly adsorb to its surface and form protein corona. The protein corona usually consists various proteins, including adhesion mediators, signaling and transport proteins, and apolipoproteins, which could improve the uptake efficiency of IONP by macrophages in MPS [95,97]. Additionally, the protein corona alters the surface interface between each particle and result in aggregation. These defects dramatically decrease the circulation time of IONP and its accumulation in lesion, especially tumor. Previous research indicated that zwitterionic molecular, that contain either zwitterionic group or a mixture of anionic and cationic groups, could effectively reduce the non-specific protein adsorption [105,[301], [302], [303]]. Bawendi and Co-author found that the size of zwitterionic dopamine sulfonate coated IONP (IO@ZDS) are similar when incubated with 1 PBS, 10% FBS, and 20% FBS. These results indicate the low nonspecific affinity of IO@ZDS towards to serum proteins due to its nearly neutral charge (Fig. 22k-m) [300]. Another study investigated the circulation fate of zwitterionic dopamine sulfonate coated gadolinium doped IONP (GdIO@ZDS) in vivo [220]. The T1-weighted signal change in heart is relative tiny after injection of GdIO@ZDS at 10, 30, and 60 min, indicating the slow elimination of GdIO@ZDS and long circulation time.

Improve biocompatibility

Although it is generally considered IONP are biocompatible in comparison to other metal oxide nanoparticles, there still remains concern on this aspect. Bare IONP have been found to generate reactive oxygen species and resulted in the in vitro cytotoxicity. Therefore, coating IONP with a more biocompatibility shell can effectively improve its biocompatibility. Since different surface coating layers show different effects on cell and different behavior in the body, various attempts have been performed to construct IONP with high biocompatibility through adjusting coating layer. Normally, the coating layer of IONP could be divided into two types, those are inorganic shell and organic shell.

Inorganic shell

Coating IONP with a biocompatible inorganic shell provides protection against release of iron ions and reactivity of IONP. After coating with the inorganic shell, the interaction activity to surrounding biological entities change from IONP to the inorganic shell. Consequently, the chemically inert shell prefer to develop IONP with high biocompatibility. Recently, gold and silica coated IONP have been discovered to be nontoxic due to their bio-inert property. Gold nanomaterial with excellent biocompatibility have been widely used in biomedical application. The cytotoxicity of gold nanoparticle have been investigated on retinal pigment epithelial (ARPE-19) cell line [304]. It appears that gold nanoparticle with small sizes shows high biocompatibility even the exposure concentration up to 5 mg/mL. Therefore, coating IONP by gold may increase its biocompatibility. Esparza and co-worker assessed the cytotoxic behavior of gold coated IONP [305]. Cytotoxicity tests indicate that the apoptosis rate of MDCK cells treated by gold coated IONP is about 0.71%, which is significantly lower that cell treated by IONP (40.33%). Gu and co-workers further assessed the in vivo biosafety of gold coated IONP using spleen-deficient rats. They found that coating IONP with gold can effectively increase the toxicity grade I concentration from 2.5 mg/mL of naked IONP to 5 mg/mL [306]. Further short-term genetic toxicity assessed by micronuclei and comet assay indicate that gold coated IONP show lower mutation and DNA damage level compare to naked IONP.

Silica is generally recognized as safe by the US FDA. In recent years, various silica coated IONP with high biocompatibility have been used in biomedical application. However, it has been found that silica coated IONP show significant influence on the survival of mNSCs mouse stem cells, which may be ascribed to the release of free iron ions in cell cytoplasm after lysosomal degradation [307]. Further modification have been discovered to be an effective strategy to improve the biocompatibility of silica coated IONP. Injumpa et al. incubated PEG modified silica coated IONP, silica nanoparticles, and IONP with macrophage cells and investigated their cytotoxicity [308]. Compared to the IONP, PEG modified silica coated IONP produces less cytoxicity. More importantly, PEG modified silica coated IONP show fewer effect on the secretion of pro-inflammatory cytokines (TNF-α and IL-6) than IONP. It should note that the cytotoxicity of silica coated IONP is highly dependent on its size. Lisi and co-workers find that sub 5 nm silica coated IONP do not alter stem cell characteristics and interfere with the commitment potential [309]. In vivo study over 7-week period reveal negligible acute and chronic toxicity after systemically administered this silica coated IONP in mice, showing excellent cytocompatibility.

Organic shell

Biocompatibility polymer coated IONP have been found to be relatively nontoxic and utilized to improve the biocompatibility of IONP. Natural polysaccharide, especially dextran, with superior biocompatibility has been considered as a promising candidate to improve biocompatibility of IONP. Muller et al. found that dextran coated IONP displayed no effect on cell viability and increasing in cytokines or superoxide production. No toxicity was observed after incubation human monocyte-macrophages with dextran coated IONP at concentration up to 1 mg/mL over 72 h [310]. Recently, another polysaccharide, hyaluronic acid (HA), with CD-44 target activity have been developed to confer biocompatibility. Atrei et al. observe a significant increase in cell viability of NIH3T3 cells exposed to the HA modified IONP respect to bare IONP. Moreover, morphometric analyses indicate that HA modified IONP show little effect on the morphology change than bare IONP, revealing an improved biocompatibility [311]. As another natural production, protein, have been used to stabilize IONP and improved biocompatibility. Roig and co-worker developed IONP outfitted with albumin (BSA) corona and assessed their cytotoxicity in adherent and suspension cells and model organism Caenorhabditis elegans [312]. In comparison to the IONP without coating, BSA coated IONP are efficiently protected in lysosome and lumen of C. elegans. Based on the high stability, BSA coated IONP are more biocompatible than the uncoated ones on cellular level and in C. elegans. Apart from natural polymer, some artificial biocompatible polymers are used to improve the biocompatibility of IONP. PVA coated IONP were found to promote BT-474 cell viability at concentrations up to 100 μg/mL. Besides, no significant ROS generation and morphology change was detected in BT-474 cells incubated with PVP coated IONP at the concentration of 50 μg/mL [313]. These results indicate that the artificial synthesized polymeric shell may also induce certain cytotoxicity in cells, although it shows high biocompatibility. In future, the long-term toxicity of IONP coated by various biocompatible shell should be systemic studied to figure out the relationship between the surface coating shell and biocompatibitity and develop IONP with no toxicity in preclinical and clinical study.

Conclusion and perspective

IONP with unique magnetic property and high biocompatibility have been widely used as MRI CAs for a long time. To realize early and accurate lesion MRI detection, various attempts have been performed to optimize the structure of IONP. In this review, we summarize the recent progress to construct IONP based high-performance MRI CA through improve their contrast ability and efficiency. This review is started with a comprehensive discussion on the strategies to increase the contrast ability by rising r2 or r1 value of IONP. As a typical T2 CA, we firstly generalized the parameters, which affect T2 relaxivity of IONP. i) Ms. On the basis of classical theories, T2 relaxivity of IONP is proportional to its Ms, which is highly dependent on its crystal structure. One can increase Ms of IONP by improving the crystallinity, dopant, and forming unique structure to obtain IONP with high T2 relaxivity. ii) Effective radius. The effective radius is responsible for the field perturbation area for the outersphere protons and proportional to the T2 relavixity in MAR. Therefore, T2 relaxivity of IONP could be promoted by increasing its effective radius through rising crystal size, forming specific morphology, and assembling to nanocluster. It should note that when the particles or cluster size reach to a certain limit, T2 contrast of IONP reaches to its maximum value and fulfills the SDR. iii) Proton diffusion behavior. Proton diffusion behavior in the magnetic field gradients have been discovered to influence the T2 relaxivity of IONP. The anisotropic structure and proper coating layer could generate strong inhomogeneous magnetic field and suitable diffusion environment, which can increase proton diffusion efficiency and T2 relaxivity of IONP. Compared to the systematical study on T2 relaxivity of IONP, the research on its T1 relaxivity is at the initial stage. Reduction the size of IONP, which can decrease the Ms value and increase the number of magnetic Fe ions on its surface, have been applied as the few exist methods to fabricate IONP based T1 CAs decades ago. With the development of synthetic method, more and more novel attempts have been performed to optimize the structure of IONP to improve its T1 relaxivity in the past ten years. By surveying the literature, we have covered a number of new structural optimization of IONP to ameliorate its T1 relaxivity. i) Increasing q value of IONP. INOPs with hollow structure and anisotropic morphology obtained by controllable synthesis could provide increased iron ion on the interfere surface between IONP and surrounding aqueous environment, which could rise its q value and result in the T1 relaxivity elevation. ii) Increasing unpaired electrons and improving τs. The existence of ferrous ions with small amount of unpaired electrons and low τs in IONP limits its T1 relaxivity. The unpaired electrons and τs of IONP could be increased by introducing other magnetic ions with sufficient unpaired electrons or long τs, such as europium, manganese, gadolinium, and copper. iii) Optimizing surface structure. T1 relaxivity of IONP is dependent on the chemical exchange efficient between IONP and water proton at the interface. Molecular weight, hydrophilic, and efficiency on charge-transfer between electronic and magnetic interactions of surface ligand have been considered as key parameters to optimize water diffusion, retention, and interaction with the magnetic centers. This beneficial effect could further improve chemical exchange efficiency between IONP and water proton and contribute to its r1 value.

After detailed discussion on contrast ability improvement, this review then summarized the recent attempt to increase the contrast efficiency of IONP in vivo from three different directions. This part starts with the discussion on recent advances of fabricating IONP with dual-modal contrast. By precise controlling the crystal structure and complementary combination of various materials, IONP with dual-modality imaging capability are emerging as versatile platform to provide comprehensive diagnostic information in disease imaging. Then IONP with the capacity to response to specific changes in the surrounding physiological microenvironment have been discussed on three different manners, those are active, recovery, and switchable. Compared to the traditional MRI CA with fixed contrast capacity in both normal and lesion tissue, this microenvironment responsive MRI CA could switch “on” and “off” T1 or T2 signal change in response to a specific stimulus of lesion. Ultimately, comprehensive understanding of strategies to increase the accumulation of IONP based CA in the lesion site, especially tumor site are present. In our surveyed publications, modified with PEG, target motif, and zwitterionic molecular are most applied method to increase the positive or active target behavior of IONP. This surface functional motif can optimize the circulation behavior and increase accumulation amount of IONP in lesion, which elevate the imaging efficiency of IONP.

Nevertheless, in spite of the remarkable progresses, there are still some obstacles ahead toward further clinical translation. Despite a number of studies have proved IONP exhibits low toxicity and high biocompatibility, biocompatibility study of IONP engineered by other cations is necessary. Additionally, all strategies are performed in the laboratory, which could not consider some practical consideration in translation to clinical, such as large-scale synthesis, long-term storage, and cost in time and money. However, we still believe that IONP based high-performance MRI CA hold great promise in further clinical early and accurate diagnosis. We hope that this comprehensive review could shed light on the development of next generation IONP based MRI CA.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.