In Vivo Measurement of Cerebral SPIO Concentration in Nonhuman Primate Using Magnetic Particle Imaging Detector

The purpose of this study is to develop a magnetic particle imaging (MPI) technique to directly measure time-varied cerebral superparamagnetic iron oxide nanoparticle (SPIO) concentration in rhesus macaques. A hand-held MPI detector was developed to monitor MPI signal changes at the third harmonics of the drive frequency in resting-state nonhuman primates. Phantom experiments were first performed to determine the sensitivity limits of the detector as a function of distance from the detector and SPIO concentration. The measured sensitivity profile was then used to reveal the most sensitive region of the detector. MPI detection was continuously performed to monitor MPI signal changes after two bolus injections of SPIOs in the rhesus macaque. We successfully developed a hand-held MPI to detect cerebral SPIO concentration changes in a living nonhuman primate. The detection limit of the MPI detector is about 125 ng iron. We reported on the in vivo measurement of cerebral SPIO concentration changes in rhesus macaque using a hand-held MPI detector. In vivo experiments showed the feasibility of the detector to sensitively measure MPI signals in a nonhuman primate brain.


I. INTRODUCTION
Cerebral blood volume (CBV) is considered as an important cerebral hemodynamic parameter for evaluating cerebrovascular alterations and metabolic functions in brain tissue [Hua 2019]. Abnormal dynamic CBV changes demonstrates the high relevance for brain diseases [Gonzalez 1995]. CBV also serves as a major modulator of the blood-oxygen-level-dependent (BOLD) effect resulting from neuronal activities [van Zijl 1998].
Various tracer-based methods have been proposed for the in vivo determination of CBV. For instance, an in vivo radioactive oxygen-15 tracer technique has been reported for measuring the CBV of rhesus monkeys [Eichling 1975]. In addition, the brain perfusion parameters of CBV can be determined by iodinated contrast agents using computed tomography [Heit 2016]. The concentration of cerebral superparamagnetic iron oxide nanoparticles (SPIOs) is considered to provide a direct reveal of CBV, since the blood-brain barrier can protect brain parenchyma from crossing the SPIO agent. Intravascular SPIOs are commonly used for obtaining maps of CBV induced by cortical activity in functional magnetic resonance imaging (fMRI) [Vanduffel 2001, Leite 2002. The use of intravascular SPIOs can reduce T2 and T2 * relaxation times in the blood to overcome the BOLD effect and, therefore, to improve the sensitivity of CBV functional imaging [Qiu 2012, D'Arceuil 2013, Baumgartner 2017]. However, the direct quantification of CBV using fMRI with IONs is challenging due to the simultaneous signal enhancement by the SPIOs and the BOLD effect [Leite 2002].
In this letter, we develop a magnetic particle imaging (MPI) technique to directly detect iron concentrations and thus quantify CBV versus time in rhesus macaque monkeys. We demonstrated that a hand-held MPI detector was capable of monitoring cerebral SPIO concentration-related MPI signal changes in living rhesus macaque. The phantom experiments showed that the lowest Fe concentration of 125 ng can be detected. In vivo experimental results suggested that the MPI technique has the great potential to be applied to obtain CBV maps of functional connectivity with high contrast-to-noise ratio for a nonhuman primate or even for humans [Graeser 2019].

A. Study Design
As previously reported, the first measurement of CBV changes in rodents using MPI detection is in Cooley [2018]. In this work, we perform in vivo measurement of cerebral SPIO concentration changes in a healthy rhesus macaque. Fig. 1 shows a schematic of the MRI imaging and MPI measurement in this study at each period. BOLD fMRI with a resting-state paradigm and T1 MRI scan were first conducted. After these baseline scans, a healthy monkey received intravenously two doses of commercial available SPIOs Mag3200 (DSPE-PEG2000-COOH iron oxide nanoparticles, 20 nm core, Nanoeast Biotech, Nanjing, Jiangsu, China) with a concentration of 1 mg Fe/ml. These two bolus injections of Mag3200 contained 4.5 and 5 mg iron respectively, resulting in cumulative doses of 4.5 mg Fe and 9.5 mg Fe. During the Mag3200 injection, the increase of MPI signal from cerebral SPIO concentration was monitored using a hand-held MPI detector for 15 min of each infusion. After MPI detection, the SPIO enhanced BOLD fMRI scan with the same imaging sequence for baseline MRI scan was performed.

B. Design of Hand-Held MPI Detector for Nonhuman Primate
The hand-held MPI instrument design of the drive and receive chains is schematically shown in Fig. 2. The drive coil is powered by an amplifier (AE Techron 7224, Elkhart, IN, USA) to generate an ac magnetic field with amplitude of 10 mT at the surface of the drive coil. In the drive side, a high-power third-order band-pass filter together with a resonant filter working at 25 kHz are used to match the drive coil and to suppress the total harmonic distortion. In the receive side, a 25 kHz second-order differential notch filter and a low-noise preamplifier (Stanford Research Systems SR560, Sunnyvale, CA) are used to reduce direct feedthrough of the drive fundamental harmonic and to amplify the MPI signal, respectively. Notably, the third harmonic component at 75 kHz of the MPI signal is merely extracted in the receive chain to alleviate the filtering requirement. Additionally, the experiment is performed with a 1 s duty cycle (18 ms of the measured data is discarded to avoid signal fluctuation, 12 ms of the third harmonic signal followed by 970 ms rest). All these recorded time point data are used to form the time-varying data of the measured MPI signals.

C. MPI Detector Sensitivity Calibration
The selection field for image spatial encoding is not considered in the current study of developing a relatively simple coil assembly for a hand-held detector. Alternatively, we used the measured sensitivity profile of the detector to localize MPI signals merged with anatomical MRI images of a macaque's head, as shown in Fig. 3. The sensitivity map was generated by measuring the third harmonic signal from a 12.5 µg Fe point source of SPIOs moving by a robotic translation stage with step size of 5 mm × 2 mm × 2.5 mm in a three-dimensional (3-D) grid (15 × 30 × 13). The hand-held MPI detector is most sensitive directly under the surface [yellow regions in Fig. 3(b)]. The sensitivity is decreasing as the distance from the surface increases [red regions in Fig. 3(b)]. Note that the measured third harmonic signal is acquired by taking a sensitivity-weighted sum of the voxels in the sensitivity map. Assuming a macaque's brain volume of 109.9 cm 3 (axes 7.0, 6.0, and 5.0 cm ellipsoid), the measured sensitivity map is considered that approximately 19% of the brain (a volume of 20.9 cm 3 ) is measured by the MPI detector.

D. Macaque CBV Experimental Procedure
The animal study was approved by the Institutional Animal Care and Use Committee at the Institute of Laboratory Animal Science, Chinese Academy of Medical Sciences. One healthy male rhesus macaque (weight 6.5 kg, 8 years) was used. A tracheostomy was performed, and femoral and venous cannulas were inserted. Before MRI experiments, the macaque was paralyzed with a 15 mg/kg intramuscular bolus injection of ketamine. After MRI experiments, the hand-held MPI detector was positioned tightly over the macaque's head, as shown in Fig. 2. The macaque received two bolus infusions of 4.5 and 5 mL polyethylene glycol (PEG)-coated SPIO (Mag3200, Nanoeast Biotech, Nanjing, Jiangsu, China) agent with a concentration of 1 mg Fe/mL, and each SPIO infusion was followed by an injection of 1 mL PBS. The total dose injected in the macaque was close to 1.46 mg Fe/kg. This dose was approximately 6.8 times less than the typical 10 mg Fe/kg SPIOs dose used in fMRI studies [Zhang 2020].

E. MRI Acquisition
All MRI scans were undertaken using a 3.0T Siemens MRI scanner (Trio) equipped with a 32-channel head surface coil. The macaque was positioned supine with its brain aligned with lasers to the isocenter of the MRI system. A T1 scan was acquired with a modified driven equilibrium Fourier transform sequence (TE = 3.4 ms, TR = 2200 ms, flip angle = 12°, matrix size = 384 × 384, field of view: 96 mm × 96 mm, voxel size: 0.8 mm × 0.4 m × 0.4 mm, 120 sagittal slices) to provide the anatomical information of the macaque brain. In vivo MPI data from the hand-held detector. (a) MPI raw data is shown during two 4.5 mg Fe bolus infusions. During the infusions the measured MPI signal (third harmonic) from the detector's sensitivity volume increases logarithmically with the concentration of SPIO in the blood. The total signal increase of the two infusions was approximately 25.4 µV. Infusion #1 and #2 yielded signal increases of ∼11.9 and ∼13.5 µV, respectively. (b) Distribution of standard deviation of the noise is shown from background noise, infusions #1 and #2. Each data point represents the standard deviation in 40 time series points during the SPIO infusion in the macaque. The P-value is determined using student's t-test. Statistical differences: * * * * p < 0.0001).
The T1 images were followed by resting-state functional imaging using a single-shot interleaved gradient-recalled echo planar imaging sequence covering the whole macaque brain (TE = 19 ms, TR = 2000 ms, flip angle = 90°, matrix size = 58 × 30, field of view = 96 mm × 96 mm, voxel size = 1.5 mm × 1.5 mm × 1.8 mm, 80 sagittal slices, repetition time = 240). After MPI data collection, the macaque was rescanned by the same imaging parameters as before the injection of SPIOs.

III. RESULTS
Phantom experiments were first performed using three 50 µL bulbs phantom of SPIOs with 250, 25, and 2.5 µg/mL concentrations of Fe. Each sample was alternately positioned at the distances d away from the receive coil center of the detector [see Fig. 4(a)]. Data were collected at each distance d varying from 0 to 10 mm with about 20 s measurement time series (1 s duty cycle). The magnitude of the third harmonic signal for the three samples roughly linearly decreases with the Fe concentration when samples are moved away from the surface center of the detector, as shown in Fig. 4(d).
Assuming a 5% blood concentration, we estimate that a 3 mm × 3 mm × 3 mm voxel in the brain volume of 109.9 cm 3 contains approximately 5 µL of blood. For a 10 mg Fe/kg SPIO dose in a 6.5 kg macaque with 62.1 ml/kg of total blood volume, a voxel with 5% CBV contains 805 ng Fe. Therefore, to detect a 25% neuroactivation-induced CBV change in this voxel, the detector must be sensitive to 201 ng Fe. Our results showed that we can detect a 125 ng Fe sample at the surface of the detector without averaging with a signal-to-noise ratio (SNR) of 6.8, which is lower than the expected 193 ng Fe concentration change resulting from a 25% neuroactivation-induced CBV in a phantom sample size. Fig. 5(a) shows the in vivo MPI signal measured by the hand-held detector positioned tightly over the macaque's head. Two bolus injections of SPIOs in the blood were represented by a logarithmic increase in the acquired MPI signal. The MPI signal increases for infusions #1 and #2 was about ∼11.9 and ∼13.5 µV, respectively. The SNR reached stability at the first 5 min and remained steady state in the following 10 min. This is consistent with the results reported in Smirnakis [2007]. These results validated the ability of our developed hand-held MPI detector to quantify the change of Fe concentration in the sensitive region of the detector. Intriguingly, the noise level also increases with the high concentration of Fe, as shown in Fig. 5(b). As reported in Cooley [2018], the increase in noise level may originate from complicated physiological fluctuations of signals that include neural activity and physiological modulations, such as respiration and cardiac pulsations.

IV. DISCUSSION
This letter describes the MPI detection of cerebral SPIO concentration in a living nonhuman primate. The sensitivity profile of the detector was measured and overlaid with T1-weigthted MRI images of the macaque brain to spatially localize the sensitive region under the macaque's skull. The sensitivity of the hand-held MPI detector was validated by phantom experiments with different SPIO concentrations. The in vivo rhesus macaque study demonstrated the feasibility of the monitoring of cerebral SPIO concentration changes with a high contrast-to-noise ratio compared to fMRI.
MPI is a relatively new imaging modality that has been first introduced in Gleich [2005]. The MPI signal originates from the nonlinear magnetization response of SPIOs located at a so-called field-free region (FFR) with an oscillating drive field and directly measured by a receiver coil. Unlike fully tomographic MPI scanners [Gleich 2005, Weizenecker 2009, Goodwill 2010, in our study, no selection field was generated for the FFR to spatially encode and map the signals to form SPIO concentration images, as in Cooley [2018]. Measured sensitivity profile calibration experiments showed that the most sensitive region of the detector was positioned on the scalp over the macaque's head (see Fig. 3). We also characterized that the measured noise in the MPI time series was increased with a high dose of SPIOs. Apart from the thermal noise sources, the physiological noise that originated from the intrinsic fluctuations in brain activity may partially contribute to the noise level of the MPI time series. Further investigation of physiological noise, especially resting state connectivity-related noise component, is required in the MPI time series. Intravascular contrast agent SPIOs, such as monocrystalline iron oxide nanoparticles (MIONs) or FDA-approved ferumoxytol, can enhance task-based fMRI signal by three-fold increase in contrastto-noise in the nonhuman primates' brain [Mandeville 2012]. In a resting-state study in nonhuman primates, Leite et al. [Leite 2002] showed only a 1.7-fold contrast increase compared to BOLD fMRI with an 8-10 mg/kg injection of a MION agent at 3 T. In this study, we also performed resting-state experiments for measuring macaque fMRI signals and functional MPI signals. We found that under the injected SPIOs dose of 1.46 mg Fe/kg to a 6.5 kg adult macaque, the CBV reflected by the T2 * intensity was only 3%, which is one-tenth lower than the typical requirement of the T2 * intensity of 30% in fMRI [Zhang 2020]. In contrast, the SNR of the cerebral SPIO concentration measurement in MPI can reach up to 134 with 10-fold lower dose than for BOLD fMRI. The PEG-coated particles revealed colloidal stability and long blood circulating time, which can increase the blood half-life time in this study. In previous mice studies, the PEG-coated particles showed an extension of blood half-life time from 105 min to 4.2 h . As shown in Fig. 5(a), the increase in the MPI signal indicates that the particles remain circulating in the cerebral blood vessels within the experiment.
Our study of MPI detection of cerebral SPIO concentration in living macaque has several limitations that can be further improved. First, we used commercially available SPIOs in the current study. The optimization of SPIOs is useful to improve the performance in MPI detection of cerebral SPIO concentration, such as larger core size, longer blood circulation time, and optimization of shape and composition of SPIO [Song 2020, Liu 2021, Lu 2021. Second, our hand-held MPI detector only obtains the weighted sum of the MPI signal within the sensitivity region of the receive coil. A tomographic MPI imaging system with selection field for nonhuman primates or even for humans [Graeser 2019] is needed to develop to form 3-D hemodynamic sensitivity images. Finally, no task or modulation of CBV experiments was performed in the current study. Other task-based functional MPI studies, such as somatosensory or visual stimulations for face recognition, will be exploited to validate the ability of MPI in functional imaging studies.

V. CONCLUSION
In conclusion, we have reported on highly sensitive detection of baseline iron levels in a rhesus macaque brain using a hand-held MPI detector. Various experiments showed the remarkable sensitivity of the detector to measure MPI signals. The development of a handheld MPI detector is a promising step toward CBV-based functional neuroimaging with MPI.