Assessment of iron nanoparticle distribution in mouse models using ultrashort‐echo‐time MRI

Abstract

Microscopic magnetic field inhomogeneities caused by iron deposition or tissue‐air interfaces may result in rapid decay of transverse magnetization in MRI. The aim of this study is to detect and quantify the distribution of iron‐based nanoparticles in mouse models by applying ultrashort‐echo‐time (UTE) sequences in tissues exhibiting extremely fast transverse relaxation. In 24 C57BL/6 mice (two controls), suspensions containing either non‐oxidic Fe or AuFeO _x nanoparticles were injected into the tail vein at two doses (200 μg and 600 μg per mouse). Mice underwent MRI using a UTE sequence at 4.7 T field strength with five different echo times between 100 μs and 5000 μs. Transverse relaxation times T ₂* were computed for the lung, liver, and spleen by mono‐exponential fitting. In UTE imaging, the MRI signal could reliably be detected even in liver parenchyma exhibiting the highest deposition of nanoparticles. In animals treated with Fe nanoparticles (600 μg per mouse), the relaxation time substantially decreased in the liver (3418 ± 1534 μs (control) versus 228 ± 67 μs), the spleen (2170 ± 728 μs versus 299 ± 97 μs), and the lungs (663 ± 101 μs versus 413 ± 99 μs). The change in transverse relaxation was dependent on the number and composition of the nanoparticles. By pixel‐wise curve fitting, T ₂* maps were calculated showing nanoparticle distribution. In conclusion, UTE sequences may be used to assess and quantify nanoparticle distribution in tissues exhibiting ultrafast signal decay in MRI.

Microscopic magnetic field inhomogeneities caused by iron deposition or tissue‐air interfaces may result in rapid decay of transverse magnetization in MRI. UTE sequences may be used to detect and quantify the distribution of iron‐based nanoparticles in mouse models in tissues exhibiting extremely fast transverse relaxation.

Abbreviations

FoV

field of view

FSE

fast spin‐echo

ICP‐MS

inductively coupled plasma mass spectroscopy

mGRE

multi‐echo gradient‐echo

RoI

region of interest

T _E

echo time

UTE

ultrashort echo time

ZTE

zero echo time

1. INTRODUCTION

Among the broad spectrum of nanoparticles available for biomedical applications, iron‐based nanoparticles have attracted significant attention due to their unique magnetic properties and excellent biocompatibility. These attributes render them highly promising for various therapeutic (eg magnetic particle heating, ¹ drug delivery, ² blood purification ³ ) and diagnostic applications. ⁴ In particular, their utilization as contrast agents for MRI has been frequently investigated and led to several clinically approved products. ⁵ Due to their biocompatibility, they are an attractive alternative to gadolinium‐based MRI contrast agents, which face considerable toxicity concerns. ⁶

Depending on oxidation state and size, iron‐based nanoparticles attain various magnetic states, ranging from paramagnetic to superparamagnetic or ferromagnetic. These characteristics can cause shortened T ₁, T ₂, or T ₂* relaxation times as well as phase and susceptibility changes by influencing the magnetic environment of surrounding hydrogen protons. For example, when superparamagnetic magnetic iron oxide or ferromagnetic iron nanoparticles are injected in vivo, they induce distortions of the magnetic field around them, resulting in shortening of the T ₂ and T ₂* relaxation times of tissues in which they accumulate. The reduction of T ₂* relaxation time provides a negative enhancement, resulting in the generation of a hypointense signal that appears dark on T ₂*‐weighted images. ⁷ Iron‐based nanoparticles not only impact on T ₂/T ₂* relaxation times but also reduce the longitudinal relaxation time T ₁, which may result in signal increase for low concentrations.

Research in the field has so far focused predominantly on the development of novel iron‐based nanoparticles with optimized magnetic characteristics. While the advancement in this regard has started to stagnate, the understanding of the biodistribution of nanostructures in general is still limited. ⁸ This is surprising, as the low targeting efficiencies of nanoparticles are frequently considered as the bottleneck in the development of highly efficient diagnostic and therapeutic nanosystems. ⁹

Radiology and especially MRI provide promising approaches to this issue. Iron deposits, eg as a consequence of hemochromatosis, are routinely assessed via exponentially decaying signals seen in multi‐echo gradient‐echo sequence (mGRE) MRI. ¹⁰ , ¹¹ The measurement of T ₂* transverse relaxation times is widely recognized as the primary approach to quantify the distribution and concentration of iron‐based deposits in the liver. ¹² Hence, such methods are in principle well suited for the biodistribution analysis of iron‐based nanoparticles, although their concentrations are typically lower.

Standard MR sequences offer echo times (T _E) in the range of a few milliseconds for spin‐echo sequences and down to 1 ms for gradient‐echo sequences. Signals arising from tissues with a very short T ₂, well below 1 ms, are therefore not visible using standard sequences, as the signal has already decayed by the time of acquisition. In the image, these tissues appear dark and indistinguishable from air cavities or noise. With conventional MRI sequences, such as fast spin‐echo sequences (FSE), which use T _E values longer than 1 ms, there is no possibility to encode the decaying signal of short‐ and ultrashort‐T ₂/T ₂* tissues before it has reached zero or near zero. A short T ₂ can usually be found in tissue with strong couplings of solid materials such as teeth, ligaments, tendons, and bones in the human body.

Moreover, iron content quantification via R ₂*‐MRI using mGRE imaging in general is compromised towards high iron concentrations or at higher fields due to ultrafast signal decay. ¹³ , ¹⁴ In the lung, mGRE of the pulmonary parenchyma is especially challenging since it is hampered by multiple challenges, primarily related to (1) patient respiratory‐ and circulation‐related motion, (2) low proton density, and (3) extremely fast signal decay caused by microscopic magnetic field inhomogeneities between alveoli and air. The aforementioned limitations of mGRE‐based iron concentration estimation could be overcome by using ultrashort‐echo‐time (UTE) imaging. ¹⁵ , ¹⁶ UTE imaging allows for very short delays (≤100 μs) between signal excitation and data acquisition, thereby enabling the detection of tissues with very short transverse relaxation times, such as cortical bone ¹⁷ , ¹⁸ and lungs. ¹⁹ , ²⁰

This study aims to demonstrate the feasibility of UTE‐based T ₂* mapping for semi‐quantitative measurement of iron concentration in liver, spleen, and lung of mice injected with nanoparticles.

2. MATERIAL AND METHODS

2.1. Study design

The study was approved by the local veterinary office of the canton of Zurich. C57BL/6 mice (8‐10 weeks old, female) were injected with 200 μL of non‐oxidic Fe or AuFeO _x nanoparticle dispersion at two different concentrations (1 mg/mL (low) and 3 mg/mL (high)) in the tail vein, resulting in a dose of 200 or 600 μg per mouse, respectively. For each condition, four mice were injected and two untreated mice served as control (no sham injection). Additionally, nanoparticle distribution was also studied in two tumor‐bearing mice, where liver metastases were induced using a well established protocol. ²¹ Finally, to investigate potential alterations in biodistribution due to knocked‐down macrophages, nanoparticle distributions were investigated in wild type mice compared with macrophage‐depleted mice. All MRI measurements of the above mice were performed 24 h after nanoparticle injection. In the sequence optimization (13 additional mice), a set of echo times for the UTE sequence was initially tested (T _E = 20 μs, 100 μs, 500 μs, 1,500 μs). In the final protocol, five different echo times were used (see below).

2.2. Animal surgery

Animal surgery was carried out as previously described. ²¹ Interventions were performed by a surgeon with extensive experience in experimental microsurgery. Isoflurane inhalation (Attane, Minrad I, Buffalo, NY) 2‐3% mixed with pure oxygen was used for anesthesia induction; intraoperative analgesia was administered via subcutaneous application of buprenorphine (0.1 mg/kg body weight). For metastasis induction, median laparotomy was performed after fixation of the animal with tape on a heating pad. The liver was mobilized by cutting the falciform ligament and the membrane between caudate and left lateral lobe with microsurgical scissors. After display of the portal vein, 1 × 10⁵ MC‐38 tumor cells, prepared in 100 μL PBS (phosphate buffered saline), were injected intraportally with a 29‐gauge insulin syringe. The needle was then removed and hemostasis achieved by gentle pressure with cotton swabs and application of small pieces of TachoSil (Baxter, Deerfield, IL), if necessary. The abdomen was closed with two layered continuous sutures with 5‐0 prolene. Mice recovered on a warmed heating pad, and food and water were provided 1 h after the operation. Postoperative analgesia with buprenorphine was administered via drinking water for three days. Animals underwent MRI one week after liver tumor induction. Macrophages were depleted using anti‐CFSR1 antibody (300 μg/mouse; Bio X Cell, Lebanon, NH) 12 h prior to nanoparticle injection. Mice were sacrificed after the MRI examination.

2.3. Nanoparticle production

Dry and scalable flame spray pyrolysis was used for the nanoparticle synthesis, as described elsewhere in detail for AuFeO _x ²² and Fe. ²³ The non‐oxidic iron nanoparticles consist of iron carbide (Fe₃C) and are encapsulated by a thin carbon shell of a few nanometers. The AuFeO _x nanoparticles consist of a Janus‐like Au/Fe₂O₃ core structure (30 at.% Fe = mol_Fe/[mol_Fe + mol_Au]) hermetically covered by an inert SiO₂ coating (13 wt% SiO₂ = m _SiO2/(m _SiO2 + m _Au + m _Fe2O3). The as‐prepared nanoparticles were surface functionalized with polyethylene glycol via silane chemistry. Typically, 10 mg AuFeO _x nanoparticles were dispersed in 17.5 mL ethanol (puriss p.a., Sigma Aldrich, MO, USA) via strong ultrasonication (Sonics Vibra‐Cell, Sonics & Materials, Inc., Newtown, CT, USA, 8 kJ, power 90%, pulse on/off 10 s/10 s) for 90 s. Thereafter, 1.25 mL NH₃ (25%, analytical reagent, Riedel‐de Haen, Honeywell Specialty Chemicals, Seelze, Germany) was added and the dispersion again sonicated for 30 min. Then 50 μL APTES ((3‐aminopropyl)triethoxysilane, 99%, Sigma Aldrich) was added, the reaction was started and kept at 25 °C for 1 h under shaking (800 rpm). Afterwards, the particle dispersion was washed via centrifugation (15 min, 8500 rpm) twice with ethanol and twice with ultrapure water and then dispersed in 2 mL of borax buffer to a concentration of 5 mg/mL. The resulting dispersion of amine functionalized nanoparticles was then sonicated for 30 min. Afterwards 100 mg mPEG‐SC (5 k, methoxypoly(ethylene glycol) succinimidyl carbonate, Biochempeg, Watertown, MA, USA) was dissolved in 2 mL borax buffer and mixed with the freshly sonicated particle dispersion to initiate the reaction. The resulting mixture was weakly sonicated (Branson 1800 ultrasonic cleaner, Brookfield, CT, USA) for 5 min before being placed in a shaker for 24 h (25 °C, 800 rpm). Afterwards, the particles were washed via centrifugation (15 min, 8500 rpm) once in borax buffer and once in ultrapure water before storing them at 1 mg/mL in saline (0.9%) for further experiments.

2.4. Microscopy and elemental analysis

Tissue samples were fixed in formalin and post‐fixed in 0.2% glutaraldehyde. They were paraffin embedded and sectioned using standard histology sample preparation protocols established at Sophistolab (Muttenz, Switzerland). Tissue sections were hematoxylin and eosin or F4/80 stained and imaged by optical microscopy (Zeiss, Oberkochen, Germany) at different magnifications. A whole slide scanner (ScopeM, ETH Zurich) was used to obtain overview images. Inductively coupled plasma spectrometry was used for iron and gold quantification in tissue after digestion using an Agilent 7900 (Santa Clara, CA, USA).

2.5. MRI

All MRI examinations were carried out in a Bruker 4.7 T BioSpec 47/40 (Bruker BioSpin, Ettlingen, Germany) equipped with a circular polarized ¹H mouse whole body transmit‐receive RF coil. The animal bed was equipped with a pad with continuous warm water flow supply to keep the mouse warm. Mice were anesthetized during MRI with 0.6% isoflurane (1‐chlor‐2,2,2‐trifluorethyl‐difluoromethylether) (Attane; Minrad I, Buffalo, NY) mixed with 100% oxygen. The animal's eyes were protected from drying by applying sterile ophthalmic ointment (Vit. A Crème, Bausch & Lomb Swiss, Steinhausen, Switzerland). A gradient‐echo localizer was applied in three spatial directions to provide anatomical overview. A transverse oriented T2w FSE sequence (T _R 4410 ms/first echo 15 ms; effective T _E 45 ms; echo train length 8; matrix 384 × 384; field of view (FoV) 45 × 45 mm; slice thickness 1 mm, three averages) was used to depict organ morphology. For visualization of fast relaxing tissue and for relaxation measurements, a three‐dimensional UTE sequence was applied with five subsequent acquisitions with T _E = 100 μs, 500 μs, 1500 μs, 3000 μs, and 5000 μs, respectively. For spin excitation, a non‐selective RF block pulse of 20 μs duration was applied. Further protocol parameters of the UTE sequence were the following: repetition time T _R = 8 ms; matrix 128 × 128 × 128; FoV 45 × 45 × 45 mm³; spatial resolution 0.35 × 0.35 × 0.35 mm³; flip angle 18.9°; one averages; acquisition time 6 m 50 s for each echo time. The standard reconstruction algorithm of the Bruker UTE3D sequence was applied for image reconstruction.

2.6. Data analysis

Data analysis was performed on a stand‐alone personal computer using custom‐made routines in the MATLAB programming language (MathWorks, Natwick, MA). Signal intensity curves and parametrical maps of T ₂* relaxation times were computed by mono‐exponential fitting. A nonlinear least square fit to the signal intensities was performed based on the Levenberg‐Marquardt algorithm using the following equation:

$$S\left({\mathit{TE}}_i\right)=S\left(0\right)*e^{-\frac{{\mathit{TE}}_i}{T2*}}+N$$

with S denoting the signal intensity, T _Ei meaning the different times to echo, T ₂* the apparent transverse relaxation time, and N the noise (three variables fitted).

For signal intensity curves, a polygonal region of interest (RoI) was manually placed to encompass as much of the investigated organ tissue as possible. RoI analysis was performed three times in each mouse in three different slices. In the lungs, the subpleural area was chosen to avoid signal contamination from large vessels. Moreover, RoI measurements were made on both sides of the lung. In liver parenchyma, RoI measurements were made three times in the lower and upper parts of the liver.

For statistical analysis, mean values and standard deviations were computed. The Kruskal‐Wallis test was used to test for significance between two groups. P values less than 0.05 were considered statistically significant. All statistical analyses were performed using commercially available software (SPSS, Release 17.0, Chicago, IL).

3. RESULTS

3.1. Qualitative assessment of image quality

FSE sequences with T ₂‐weighted contrast (T _E 45 ms) and UTE sequences (T _E between 100 μs and 5000 μs) were acquired. Representative images are shown in Figure 1. In mice without nanoparticle injection, the FSE sequences show intermediate signal intensity of the liver, allowing for assessment of liver parenchyma. In mice treated with Fe nanoparticles, a complete signal loss to background noise can be observed in liver parenchyma caused by the magnetic field inhomogeneities induced by nanoparticles. In the lungs, regardless of whether nanoparticles have been applied or not, a complete signal loss is observed in the FSE sequence. Applying UTE sequences, in iron‐loaded liver, iron‐loaded spleen, and lungs, signal above the noise level can be observed. Visually on MR images iron deposition was only observable in liver and spleen parenchyma, and from image impression alone nanoparticle deposition in the lung was not detectable. Liver metastases were detectable in both FSE and UTE sequences (Figure 2). However, in the liver loaded with Fe nanoparticles the normal parenchyma could not be assessed due to signal loss, whereas in UTE sequences both liver metastases and iron‐loaded liver parenchyma were visible. No uptake of nanoparticles was seen in the metastases.

FIGURE 1.

FSE (T _E = 45 ms) (A, B, D) and UTE sequence (T _E = 100 μs) (C, E, F) MR images of metastasis‐free mice with and without administration of Fe nanoparticles (low dose). A‐C, The liver parenchyma, which darkens due to a complete signal loss caused by the accumulated Fe nanoparticles. D, E, Images of the lung. No signal can be measured in lung parenchyma using an FSE sequence due to the fast signal decay caused by microscopic magnetic field inhomogeneities. Using a UTE sequence with echo time T _E = 100 μs, signal above the background noise level can be measured in iron‐loaded liver, spleen, and lung parenchyma. Typical RoI measurements for computation of signal curves are indicated by the dashed lines. Li indicates liver, Lu lung, and S spleen. NP, nanoparticles

FIGURE 2.

FSE and UTE sequence MRIs of mouse liver metastases after Fe nanoparticle administration (low dose). A T ₂ hyperintense metastasis (arrow) can be seen in the FSE image with complete signal loss of the surrounding iron‐loaded liver parenchyma. In the UTE sequences, both liver parenchyma and metastasis are clearly visible using T _E of 500 μs and 1500 μs, respectively. In the T ₂* map, the metastases show notably longer T ₂* relaxation time compared with liver parenchyma loaded with nanoparticles

3.2. Signal intensity curves

Example T ₂* signal intensity curves are shown in Figure 3. From the curves it can be seen that an excellent curve fitting based on the five signal intensities of different echo times between 100 μs and 5,000 μs was possible in untreated mice as well as for both nanoparticle compositions and for both injected numbers of nanoparticles in liver, lung and spleen. A sequence with even shorter echo time of 20 μs resulted in worse signal‐curves due to a signal loss at this echo time compared with the later echo at 100 μs, which may be due to hardware limits of the ADC converter. No reliable signal curves could be computed from tissues with T ₂* relaxation times longer than the liver (eg muscle tissue). Investigation of such tissues would have required the inclusion of even longer echo times in the MRI protocol. In contrast, signal curves with suitable quality could be obtained from the lung tissue in all mice (treated and untreated).

FIGURE 3.

T ₂* signal relaxation curves (measurement points and corresponding mono‐exponential fits) for liver (top), lung (middle) and spleen (bottom) parenchyma. The computed transverse relaxation times are shown in the legends

3.3. Quantitative T ₂* transverse relaxation times

An overview of the measured T ₂* values is visualized in Figure 4; numeric mean values and standard deviations are provided in Table 1. No statistical difference was found between the left and right sides of the lungs; therefore, T ₂* values from the two sides were pooled. Liver T ₂* values were shorter close to the diaphragm compared with the more caudal parts of the liver (data not shown), which is most likely due to macroscopic magnetic field inhomogeneities close to the diaphragm (tissue‐air interface). Dependent on the nanoparticle type and dose, a reduction of T ₂* values can be seen in liver, spleen, and lung tissue. Macrophage depletion was performed to increase the nanoparticle deposition in the lungs, resulting in a similar reduction of T ₂* time with the low number of Fe‐based nanoparticles in the lung as compared with the high number of nanoparticles without macrophage depletion. In some of theT ₂* values of Figure 4, a relatively large error bar can be seen, which may be due to inaccurate T ₂* quantification in tissues with long transverse relaxation time, small size of the evaluated organ or incomplete injection of nanoparticles in the tail vein.

FIGURE 4.

Mean and standard deviations for T ₂* transverse relaxation times of liver, spleen, and lung for untreated mice, mice injected with AuFeO _x or Fe nanoparticles in high and low concentrations, and mice treated by macrophage depletion before injection of low concentrations of AuFeO _x or Fe nanoparticles. It can be seen that Fe nanoparticles result in stronger reduction of transverse relaxation time compared with AuFeO _x nanoparticles, and the higher number of nanoparticles results in stronger reduction of T ₂*. In the liver, an effect of Fe nanoparticle deposition can clearly be observed. Macrophage depletion causes a reduction of T ₂* in the lung with low concentration similar to high concentration without macrophage depletion, which may be attributed to higher nanoparticle deposition in the lung due to lower accumulation in liver macrophages

TABLE 1.

Mean T ₂* relaxation times and standard deviations measured in the liver, kidney, and lung after application of different nanoparticle types and amounts

T 2 * relaxation times [μs]
Treatment	Liver	Spleen	Lung
Untreated	3418 ± 1535	2170 ± 728	861 ± 208
AuFeO x low	6438 ± 4193* (p = 0.04)	1634 ± 202 (p = 0.26)	527 ± 153* (p < 0.01)
AuFeO x high	3836 ± 2024 (p = 0.50)	2276 ± 686 (p = 0.93)	663 ± 101* (p < 0.01)
AuFeO x m‐depl.	3118 ± 1521 (p = 0.68)	1594 ± 426* (p = 0.04)	539 ± 84* (p < 0.01)
Fe low	1771 ± 2729* (p < 0.01)	453 ± 280* (p < 0.01)	507 ± 106* (p < 0.01)
Fe high	228 ± 67* (p < 0.01)	299 ± 97* (p < 0.01)	413 ± 99* (p < 0.01)
Fe m‐depl.	720 ± 173* (p < 0.01)	193 ± 78* (p < 0.01)	409 ± 89* (p < 0.01)
Mean RoI size [voxels]	93 ± 45	40 ± 17	44 ± 18

^*Statistically significant versus untreated.

3.4. Parametrical T ₂* maps

T ₂* maps were calculated in a pixel‐by‐pixel manner; example images are displayed in Figures 5 and 6. From the generated T ₂* maps with good image quality, it can be seen that the curve fit is stable enough to allow pixel‐wise evaluation.

FIGURE 5.

UTE images (left) and corresponding T ₂* parametrical maps (right) of the liver for high (top) and low (bottom) numbers of injected Fe nanoparticles. L, liver; S, stomach

FIGURE 6.

UTE images (left) and corresponding T ₂* parametrical maps (right) of the basis of the lungs without injected nanoparticles and of the spleen (blue arrow) with low number of injected Fe nanoparticles

3.5. Ex vivo characterization of the nanoparticle content in tissues

After mouse sacrifice, liver, spleen, and lungs were harvested and analyzed using histology, including macrophage staining with F4/80antibodies. Additional metal quantification was performed using inductively coupled plasma mass spectroscopy (ICP‐MS). Observation of F4/80 antibody‐stained sections shows the presence of black deposits in macrophages in the organs from the mice exposed to both magnetic nanoparticles, with a dose effect Figure 7). These black spots can then be related to the presence of nanoparticles, and thus confirm the presence of the Fe and AuFeO _x particles in the liver and the spleen. In contrast, no differences have been observed between the lungs from the control and nanoparticle‐exposed mice. This indicates that UTE sequences can potentially detect nanoparticle deposits with improved sensitivity compared with histology alone. To quantify the amount of metal, ICP‐MS of liver and spleen tissue was used in three healthy mice, showing first the presence of gold only in the tissues exposed to AuFeO _x nanoparticles, and second an increase of the amount of iron above the endogenous level after injection of both types of nanoparticle (Figure 8).

FIGURE 7.

Imaging of histology sections of liver (A, B) and spleen (C, D) after tissue staining with F4/80 antibody. Micrographs in A and C present sections of organs harvested from vehicle injected animals (A, C); B and D display micrographs of animals injected with a high dose of Fe (B, D). In the latter, red arrows indicate the presence of characteristic black depositions. Scale bars: 50 μm

FIGURE 8.

Quantification of the iron (and gold) content in the livers and spleens from mice injected with vehicle solution, low dose of Fe nanoparticles or low dose of AuFeO _x nanoparticles. Results displayed for one mouse per condition

4. DISCUSSION

In this study we utilize UTE imaging to better visualize typically MRI invisible organs with a short T ₂*, such as lungs, and liver after application of iron‐based nanoparticles providing an alternative method for investigating the biodistribution of nanoparticles, as seen in our case with an Fe nanoparticle loaded tumor‐containing liver and lung. The sensitivity for the detection of magnetic nanoparticles as molecular imaging probes is enhanced due to the increased signal yield, and a semi‐quantitative measurement of nanoparticle concentration is provided by computation of T ₂*.

Using FSE, we observed complete signal loss in the liver parenchyma after nanoparticle injection and deposition. We did not find any signal in lungs with FSE after nanoparticle deposition due to fast signal decay caused by microscopic magnetic field inhomogeneities owing to tissue‐air interfaces in the alveoli of lungs. Using a UTE sequence with echo time T _E = 100 μs, signals above the background noise level can be measured in Fe nanoparticle loaded livers with differentiated metastasis, spleen, and lung parenchyma.

It is known that tissues with low background signals and nanoparticle induced susceptibility artifacts require fast signal acquisition starting immediately after the RF excitation. In UTE pulse sequences, the RF excitation is performed before gradients are ramped up and data acquisition begins after a short, non‐negligible delay time. Data acquisition occurs while the readout gradient is being ramped on. Since the readout gradients are turned on at the beginning of data acquisition, the sampling pattern of k‐space is radial, filled from the center out, and is nonlinear in time. UTE can be applied for 2D imaging, ²⁴ , ²⁵ with a slice selective gradient during the RF pulse while encoding in two directions, or for 3D imaging, without any slice selective gradients and encoding in all three directions. Slice selective 2D radial acquisition may be used to image brain tissue, ²⁶ cortical bone, ²⁷ , ²⁸ and tendons. ²⁹ An acquisition‐weighted stack of spirals sequence has been proposed for high‐resolution 3D UTE imaging. ³⁰ Typically, radial trajectories are used, although other center‐out trajectories such as spirals, ³¹ twisted projections, ³² or cones ³³ are also feasible. However, it must be pointed out that center‐out techniques for lung imaging are limited by the required rather long readout.

UTE shares with the other radial sampling techniques the advantage of being less sensitive to motion artifacts compared with cartesian sampling. ³⁴ One challenge of the UTE sequence is that acquiring data during gradient ramping can lead to major image distortions originating from eddy currents and time delays within the gradient system. Zero‐T _E imaging may overcome this problem. A disadvantage of UTE imaging is that it typically requires a relatively long scan time compared with conventional T ₂*‐weighted GRE. However, UTE imaging is advantageous and potentially more sensitive compared with T ₂ and T ₂* imaging using standard sequences when nanoparticle concentrations are high in the tissue, but may not be beneficial when the tissue nanoparticle concentration is low. In particular, in tissues with long relaxation time inaccuracies in T ₂* quantifications may occur with UTE sequences, as the curve shape and the noise level cannot reliably be estimated.

Zero echo‐time (ZTE) or UTE pulse sequences use specialized acquisition and reconstruction techniques to enable detection of ultrashort‐T ₂ components in vivo, and rely on beginning data acquisition as soon as possible after completion of the RF pulses. ³⁵ Radial imaging, as in UTE and ZTE, is particularly robust to movement artifacts. A certain amount of image blurring must nevertheless be expected if motion is present during the acquisition. Thus, the need of the day is for techniques such as UTE and ZTE, which are able to compensate for motion, resulting in the best image quality possible. Moreover, such techniques can be useful to determine the influence of motion over parameter quantification in motion corrupted datasets.

Different tissues and organs have different T ₂* values, and the different statuses of the same tissues and organs also have different T ₂* values. Thus, the T ₂* value has the potential to reflect a change in the biochemical components of the counterpart and may be used for the early diagnosis and quantitative description of some diseases, eg to detect and quantify excess iron deposition in various organs such as the liver, heart, spleen, pancreas, and pituitary gland. ³⁶ In T ₂ *‐weighted sequences, the lesion, structures, or areas of dephasing are shown as dark areas, leading to their detection or characterization. ³⁷ The T ₂ *‐based contrast mechanism is also used in other MR applications such as susceptibility weighted imaging, MR perfusion imaging, functional MRI, and iron overload imaging. ³⁸ Iron overload can be caused by disorders of iron absorption, such as hereditary hemochromatosis and thalassemia intermedia, defects in heme metabolism, or long‐term transfusion therapy. ³⁹ , ⁴⁰

Iron nanoparticles, as magnetically active agents, primarily shorten transverse relaxation times T ₂ and T ₂*, which leads to prominent signal decrease or “negative contrast” of targeted tissue on T ₂‐ and T ₂*‐weighted images. UTE imaging enables capturing signal of the shortened T ₂ and T ₂*, allowing for obtaining signal from tissues with rapidly decaying transverse relaxation, as observed in the current study. Here we saw that Fe nanoparticles result in stronger reduction of transverse relaxation time compared with AuFeO _x nanoparticles. This can be explained by the higher saturation magnetization of the Fe nanoparticles (140 emu/g) ⁴¹ compared with the AuFeO _x system (43.5 emu/g). ²² Also, as expected, a higher concentration of nanoparticles results in stronger reduction of T ₂*. In the liver, an effect of nanoparticle deposition can clearly be observed. Interestingly a reduction of T ₂* for lung parenchyma in vivo after nanoparticle deposition has been shown for the very first time here in this study.

The lung parenchyma properties that are important in the context of its visualization with MRI are its low density and the susceptibility differences between tissue and air. The tissue density in a healthy human lung is around 0.1 g/cm³, ⁴² which is about 1/10 that of other tissues. As the MR signal is directly proportional to the tissue proton density, the MR signal arising from the lung is 10 times weaker than that of other tissues, even prior to any relaxation. The inherently low signal to noise ratio (SNR) is one of the main limitations that makes structural proton MRI of the lung challenging, particularly when acquisition times must be within reasonable limits and high spatial and/or temporal resolution is required. ⁴³ The second important property is the high number of tissue‐air interfaces in the lung. Oxygen present in the air is paramagnetic and lung tissue is diamagnetic, leading to a magnetic susceptibility difference at lung‐air interfaces. Since between two areas of different susceptibility there exists a small static magnetic field, this results in a highly inhomogeneous local magnetic field on a spatial scale smaller than the size of a typical voxel. This leads to a rapid signal dephasing in gradient‐echo imaging, and thus to short T ₂*. Thus, gradient‐echo MRI of lung parenchyma becomes highly challenging and requires pulse sequences with peculiarly low T _E. Moreover, continuous motion is present in the thoracic and abdominal area, due to the presence of respiratory movements and the heartbeat. This requires countermeasures for avoiding artifacts and image blurring, which may happen when these movements occur on a time scale shorter than the acquisition time of the image. ⁴⁴

The T ₂* relaxation time of tissues is dependent on many parameters, eg the voxel size, the voxel geometry, the field strength, the number of echoes, the distribution of the measured echoes, and the technique for calculating T ₂* (fitting algorithm). In general, the larger the voxel size and the higher the field strength, the faster the transverse signal decay. The treatment of noise can strongly influence the calculated T ₂* depending on whether a fixed or variable noise is used in the fitting algorithm. Using a 3D UTE sequence at 4.7 T with an isotropic spatial of 0.35 × 0.35 × 0.35 mm³ with five different echo times and variable fitting of noise, we measured a T ₂* of normal lung of 860 μs. Olsson et al ⁴⁵ at the same field strength used a larger voxel size of 0.31 × 0.31 × 1.56 mm³, only two echo times, and an algorithm with fixed noise measuring 460 μs. Guo et al ⁴⁶ used 7 T field strength, a spatial resolution of 0.47 × 0.47 × 0.47 mm³ and six different echo times, measuring a T ₂* of 395 μs. Balasch ⁴⁷ measured 200 μs at 11.7 T using a spatial resolution of 0.2 × 0.2 × 0.2 mm³, 10 echo times, and local noise measurement.

Our study had several limitations. First, only a small number of animals have been used, with four mice dedicated to each concentration and nanoparticle type. Due to rules of animal welfare, the 3R principles (reduce, replace, refine) must be applied, and we believe that with the applied study design we were able to demonstrate the main aims of our study. However, we cannot exclude that using more animals in each branch of the study would have been achieved even better results. Second, after UTE protocol optimization and finalizing the post‐processing routines, we cannot exclude that even better settings for both the UTE sequence (with less background blurring) and the post‐processing algorithms could have been found. However, the final settings used in this study were shown to provide consistent and reliable results. Third, only five different echo times of the UTE sequence have been acquired, which is due to the approved animal protocol restricting the possible time of the MRI examination protocol due to cooling of the animals. Fourth, all experiments have been carried out at 4.7 T, and potentially better image quality and higher quantification accuracy could have been achieved with higher field strength or other types of improved hardware. In particular, it would have been desirable to acquire signal at even shorter echo time such as 100 μs, which was not possible with our hardware set‐up. A shorter transmit/receive switch might be a solution, as described by Sheth et al. ⁴⁸ Fifth, for accurate iron quantification, a calibration curve would be needed correlating the tissue iron content to the corresponding T ₂*. Due to differences in the intrinsic tissue‐specific T ₂* relaxation times, a separate curve would be required for each organ to be evaluated. In the current study, we abstained from exact calibration of the iron‐T ₂* curves, as this would require a large number of animals injected with different numbers of nanoparticles. Finally, a technique called quantitative susceptibility mapping was recently described to be combined with UTE imaging (UTE‐QSM), which might be superior for iron quantification compared with T ₂* mapping.

In conclusion, we have demonstrated that UTE imaging can be used for obtaining increased signal in tissues loaded with iron‐based nanoparticles and consequently short transverse relaxation time. In particular, detection of nanoparticles loaded in lung parenchyma becomes feasible. A semi‐quantitative measurement of nanoparticle concentrations can be obtained by calculating T ₂* from UTE imaging of different echo times.

Boss A, Heeb L, Vats D, et al. Assessment of iron nanoparticle distribution in mouse models using ultrashort‐echo‐time MRI. NMR in Biomedicine. 2022;35(6):e4690. doi: 10.1002/nbm.4690