1076 IEEE TRANSACTIONS ON MEDICAL IMAGING, VOL. 31, NO.

5, MAY 2012

Projection X-Space Magnetic Particle Imaging

Patrick W. Goodwill*, Justin J. Konkle, Bo Zheng, Emine U. Saritas, and Steven M. Conolly

Abstract—Projection magnetic particle imaging (MPI) can im- Due to the extraordinary contrast and penetration of the MPI
prove imaging speed by over 100-fold over traditional 3-D MPI. In imaging modality, there are many applications for which a 2-D
this work, we derive the 2-D x-space signal equation, 2-D image projection MPI image would be highly desirable. This is be-
equation, and introduce the concept of signal fading and resolu-
tion loss for a projection MPI imager. We then describe the design cause only two dimensions need be scanned rather than three,
and construction of an x-space projection MPI scanner with a field and so projection scanning is inherently faster by a factor equal
gradient of 2.35 T/m across a 10 cm magnet free bore. The system to the number of pixels in the projection direction. Hence, pro-
has an expected resolution of 3.5 8.0 mm using Resovist tracer, jection MPI is about two orders of magnitude faster than 3-D
and an experimental resolution of 3.8 8.4 mm resolution. The MPI. This increased speed will be critical for real-time angiog-
system images 2.5 cm 5.0 cm partial field-of views (FOVs) at 10
frames/s, and acquires a full field-of-view of 10 cm 5.0 cm in 4 raphy, stem cell imaging, and cancer detection using dynamic
s. We conclude by imaging a resolution phantom, a complex “Cal” contrast enhancement. Indeed, we believe that MPI’s preferred
phantom, mice injected with Resovist tracer, and experimentally mode of operation may be fast projection imaging, akin to pro-
confirm the theoretically predicted x-space spatial resolution. jection and fluoroscopic X-ray.
Index Terms—Angiography biomedical imaging, magnetic par- Key to forming a projection image is a field free line (FFL)
ticle imaging (MPI), magnetic particles. main magnet [9]. A FFL is similar to the field free point (FFP)
used in 3-D MPI scanners, except the FFL has close to zero field
magnitude within a line-like region as opposed to a point-like re-
gion. There are many techniques for generating a field free line,
I. INTRODUCTION which is equivalent to the quadrupolar magnetic field used in
linear accelerators since the 1950s [10]. The first work to pro-
AGNETIC particle imaging (MPI) is a new medical
M imaging modality with great promise for high-contrast
angiography, cancer imaging, in vivo cell tracking, and inflam-
pose a FFL MPI magnet used simulations to show that a rotating
and shifting a FFL could increase the sensitivity of the MPI tech-
nique by an order of magnitude [9]. This work was followed by
mation imaging in humans and small animals. Knopp et al., who simulated a series of improved main magnet
First published in 2005 [1], the MPI technique has progressed designs capable of electronically rotating the FFL [11], [12]. In
rapidly. Recent hardware developments include: real-time MPI addition, Knopp et al. built and characterized using hall effect
in a mouse [2]–[4], methods for single-sided MPI using strongly measurements a nonrotating FFL electromagnet design with a
nonlinear magnetic field gradients [5], and narrowband and gradient strength of 0.1 T/m [13]. All the prior literature on FFL
x-space images of mice and phantoms [6]–[8]. magnets [9], [11]–[13] proposes translating and rotating the FFL
Current state-of-the-art MPI is intrinsically 3-D due to the to produce a 3-D image with improved SNR; we instead pro-
design of the main magnet [3], [8]. These main magnet designs pose that using a FFL for a projection scan enables new classes
form a single point in three-dimensions termed the field free of real-time tracer studies and high sensitivity tracer imaging.
point (FFP), where the magnetic field magnitude is weaker than Here, we build the first permanent magnet FFL gradient for
the saturation magnetization of a magnetic nanoparticle tracer. MPI and provide the first MPI images taken with a FFL. The de-
An image is formed using a FFP by performing a voxel-wise sign inspiration for the FFL magnet used in this paper is the stan-
scan of a 3-D volume. To date, all experimental MPI scans have dard quadrupole magnet used in linear accelerators, which we
been performed with a FFP. approximate by constructing a four-rung magnet using
NdFeB permanent magnets. The number of rungs is a tradeoff
Manuscript received October 24, 2011; revised January 10, 2012; accepted between field homogeneity and access. Because the projection
January 12, 2012. Date of publication February 03, 2012; date of current version direction of a quadrupole magnet is axial, we require
May 02, 2012. This work was supported in part by CIRM Tools and Technology access through the side of the magnet structure to introduce an
Grant RT1-01055-1, and a UC Discovery grant. The contents of this publication
are solely the responsibility of the authors and do not necessarily represent the animal or phantom. A four-rung approximation to the
official views of CIRM or any other agency of the State of California. This work distribution offers a full 180 of open access. FFL movements
was supported in part by the National Institute of Biomedical Imaging and Bio-
engineering under Grant 1R01EB013689. . The content is solely the responsi-
are performed using high power electromagnets and mechani-
bility of the authors and does not necessarily represent the official views of the cally using a robot. The design concept and completed magnet
National Institute of Biomedical Imaging and Bioengineering or the National are shown in Fig. 1.
Institutes of Health. Asterisk indicates corresponding author.
*P. W. Goodwill is with the Department of Bioengineering, University of
In this paper, we begin by deriving analytical expressions in
California, Berkeley, CA 94720 USA (e-mail: goodwill@berkeley.edu). x-space for the projection MPI signal, system point spread func-
J. J. Konkle, E. U. Saritas, and S. M. Conolly are with the Department of tion, and magnet homogeneity requirements. We then design
Bioengineering, University of California, Berkeley, CA 94720 USA. and build an x-space MPI projection scanner, which we use to
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. measure the system’s native point spread function, image a com-
Digital Object Identifier 10.1109/TMI.2012.2185247 plex “Cal” phantom, and image a mouse injected with tracer.
0278-0062/$31.00 © 2012 IEEE
Authorized licensed use limited to: Gebze Technical University. Downloaded on June 29,2023 at 13:22:50 UTC from IEEE Xplore. Restrictions apply.
GOODWILL et al.: PROJECTION X-SPACE MAGNETIC PARTICLE IMAGING 1077

A. Projection X-space MPI Signal Equation

The x-space theory of projection MPI builds on our previous
work on multidimensional MPI. In multidimensional MPI, we
derived the signal and image equations by assuming the FFP has
a unique position in three dimensions. Maxwell’s equations in
a source free region imposes physical constraints on the achiev-
able magnetic fields; as described in [8] and [21] the achiev-
able magnetic field gradient can be written as a parameterized
gradient matrix with diagonal symmetry and trace
times the spatial vector. Specifically the magnetic field
as a function of position can be written

(1)

with one free design parameter (0,1). Fortunately, a FFL

Fig. 1. First projection MPI scanner based on a quadrupole magnet built using gradient for projection MPI along the direction can be con-
NdFeB magnets. (a) Diagram showing all the magnets in the Berkeley FFL structed while obeying Maxwell’s equations. Suppose we de-
scanner. The NdFeB magnets are mounted in a G10 fiberglass composite and
aluminum frame, which generates a FFL collinear to the y axis (dotted red line). sign the FFL magnet so that (or ) and so that
NdFeB magnet magnetization orientation is shown with arrows. The FFL slow the cross terms , , and are zero. This describes a
shift electromagnets enable shifting the FFL along the x axis. The resonant xmit physically realizable field, since the matrix has zero trace and
coil shifts the FFL rapidly along the z axis. The recv coil receives the nanopar-
ticle signal. (b) Completed projection MPI scanner. diagonal symmetry

II. MPI FFL THEORY

Three MPI techniques have been shown experimentally. The For this important case, the location of the free point would
most published technique is harmonic-space MPI, which uses a be undefined as the gradient matrix is singular. Note that this
system matrix. The system matrix is comprised of Fourier com- describes a FFL along the direction.
ponents of the temporal signal for every possible location of a To enable x-space theory with this gradient matrix, we first
point source [1], [2], [4], [14]–[17]. Reconstruction is achieved define the concept of a FFL. We define the FFL as the line across
through regularization and matrix inversion techniques such which the magnetic field magnitude is zero, which we assume
as singular value decomposition or algebraic reconstruction. here is straight but could in fact be curved. For this case, we
The second technique is a narrowband technique, which re- write the gradient matrix in a reduced, nonsingular form.
constructs harmonic images into a composite image using a Let us now assume that the FFL is along the axis. Then, the
modified Wiener deconvolution [6], [7]. The third technique, magnetic field can be written
which we use in this paper, is x-space MPI [8], [18], [19].
X-space MPI is a new theoretical approach to MPI signal pro-
cessing and reconstruction that enables acquisition of a native
MPI image, and does not require a system function, harmonics, We can now shift the center of the field-of-view (FOV) using
or pre-characterization of the nanoparticles or imager. We de- homogeneous magnetic fields
scribed the x-space technique in detail in our first two x-space
papers [8], [18]. X-space offers several advantages over har-
monic-space MPI matrix reconstructions. Specifically, x-space
MPI reconstruction offers experimentally proven linearity and where and are time varying, homogeneous mag-
shift invariance (LSI), as well as real-time image reconstruction netic fields that point in the and instrument axes. Giving the
speed [8]. Importantly, x-space reconstruction makes no attempt gradient a convenient negative sign, the total magnetic field as
to deconvolve the MPI signal to improve resolution over the res- a function of time can be written as
olution determined by the physics of the nanoparticles and field
gradient, and thus avoids the significant noise gain of deconvo-
lution (see the noise gain in [20]).
Solving for , we see that the instantaneous location of the FFL
The x-space technique relies on three major hypotheses: 1)
is moved by the application of homogeneous magnetic fields.
that magnetic nanoparticles adiabatically align with an applied
The instantaneous FFL center is located at the 2-D position
magnetic field, 2) that the FFP location is unique, and 3) that
the loss of low frequency information due to filtering the funda-
mental is recoverable. In this section, we extend x-space theory
to work with projection MPI. We first describe the signal and The inverse of is guaranteed to exist since the determinant of
image equations, which differ from the 3-D case [8]. the gradient matrix is nonzero
Authorized licensed use limited to: Gebze Technical University. Downloaded on June 29,2023 at 13:22:50 UTC from IEEE Xplore. Restrictions apply.
1078 IEEE TRANSACTIONS ON MEDICAL IMAGING, VOL. 31, NO. 5, MAY 2012

for all nontrivial FFL gradients. Physically, this means that there It is important to understand that this fast x-space image recon-
exists a unique FFL in the uniform region of homogeneous field struction does not deconvolve the PSF from the recon-
and linear region of gradient field . structed image. Standard deconvolution methods may be added
Provided that the static field gradient is linear, the FFL moves after this final reconstruction step, but these will inevitably re-
linearly with the addition of magnetic fields in the and axes. duce the final image SNR.
It is important that the magnetic fields and have
good homogeneity. This is because the addition of any magnetic B. FFL Homogeneity Requirements
field along the field free line (perpendicular to both and ) The uniformity of the FFL is vital for projection imaging.
will cause the magnetic field experienced by the sample to not Nonideality of the FFL results in two signal effects: signal
pass through zero, and will lead to signal fading and resolution fading and resolution loss. For low levels of nonideality, we
loss artifacts. These artifacts are discussed in detail below in approximate the loss of signal in a simple fading approximation
Section II-B. as the resolution loss is less noticeable. For larger levels of
Our results from 3-D multidimensional MPI directly apply to nonideality, the resolution of the system can drop significantly.
projection MPI when we use the revised gradient matrix defi- Here we describe both effects.
nition [8]. Substituting into the generalized signal equation, we 1) Fading: Fig. 2 shows how the point spread function
find that the signal equation remains a convolution changes for a point source passing through a nonideal field
free line away from the magnet isocenter. Because the FFL is
(2) nonideal, particles near the ends of the FFL never pass through
a field region with zero magnetic field. This partial passage
through the Langevin curve leads to signal dropout near the
where is the receive coil sensitivity matrix, ends of the FFL. We have designed the Berkeley FFL main
is the magnetic moment of a single nanoparticle, magnet so that there is minimal signal dropout ( 5% fading)
is the magnetic nanoparticle density, using Resovist (Bayer–Schering) tracer.
is the saturation field of the magnetic nanoparticle, We can describe the nonideality of the FFL as a gradient along
is the point spread function, and is the FFL the direction of the FFL. In our magnet, the FFL and conse-
velocity vector. However, this equation differs from the 3-D quently the undesired gradient is along the axis, [which is
case. First, the convolution does not occur in three dimensions equivalent to in (1)]. Assuming the field free line is along
as before, but in the plane perpendicular to the FFL. Second, we the axis, the maximum undesired magnetic field at any posi-
do not convolve the point spread function with the nanoparticle tion along the axis can be written
density, but the projection of the nanoparticle density along the
FFL, . We can express the projection of the nanoparticle (4)
density as an integral
We now introduce the term signal fading to describe the loss of
image intensity due to FFL inhomogeneity. Let us assume we
are measuring a point source at some position along the axis.
Then, the peak signal of a point source scanned at some posi-
The resulting units of the nanoparticle projection density func- tion along the axis can be estimated using the Normal signal
tion has the units of . The point spread function envelope. From [8], the normalized peak signal as a function of
equation remains the same as in the general 3-D case a static offset field is

(5)

where is the remaining signal (see Fig. 3). For example, if

, this describes a 10% fading from ideal. Equation
(5) predicts 2% fading for Resovist tracer used in the Berkeley
where denotes the Langevin function. The projection reduces projection scanner ( at 2 cm, Resovist
the dimensionality of the matrix PSF so that . , ). Of course, any FFL magnet
We reconstruct the projection MPI image by dividing the raw design will suffer from some degree of fading, and a moderate
collinear signal by the instantaneous velocity in x-space fol- amount (say 10% fading) will very likely be well tolerated clin-
lowed by gridding of the signal to the instantaneous FFL lo- ically, similar to (the much greater) fading of the MRI signal
cation in 2-D space. As proven below, this extremely fast pro- with distance from an RF surface coil.
cessing step obtains a 2-D convolution of the point spread func- 2) Resolution Loss: Magnet inhomogeneity reduces system
tion with the ideal projection of the nanoparticle density resolution because the nanoparticle no longer passes through the
full Langevin curve (see Fig. 2). Unfortunately, the widening of
the full-width at half-maximum (FWHM) is difficult to express
in a closed form solution. We can instead calculate the increase
(3) in FWHM numerically using the full x-space theory, which we
show graphically in Fig. 3.
Authorized licensed use limited to: Gebze Technical University. Downloaded on June 29,2023 at 13:22:50 UTC from IEEE Xplore. Restrictions apply.
GOODWILL et al.: PROJECTION X-SPACE MAGNETIC PARTICLE IMAGING 1079

Fig. 2. Nonideality in a FFL causes signal loss and widening of the PSF. We can approximate small FFL nonidealities as a reduction in signal amplitude, or signal
<
fading. The Berkeley magnet achieves 5% fading.

can take a substantial amount of time when all human safety

considerations (limits on and SAR) as well as power
supply considerations are enforced. Thus, reducing the number
of dimensions in an image reduces the number of voxels,
which results in a concomitant increase in imaging speed.
This increase in speed could enable the application of MPI
to varied applications. For example, projection MPI would be
ideal for real-time angiography with speeds approaching X-ray
fluoroscopy, since the FFL can be moved just as fast as a FFP
in a 3-D scanner. For perspective, the FFL in the Berkeley
projection scanner is moving at more than 1200 m/s rms. If we
instead desire to increase SNR rather than speed, projection
scanning enables multiple image averages without increasing
Fig. 3. Increasing levels of FFL inhomogeneity worsen fading and resolution.
scan time over a 3-D scan of the same area.
For FFL inhomogeneity values of H=H < 2, resolution loss and fading are
III. MAGNET DESIGN AND CONSTRUCTION
approximately equal in magnitude.
In this section, we describe the magnet design criteria and
system construction. The overall design concept was to con-
For moderate amounts of fading up to which struct a side-access quadrupole magnet. To achieve the magnetic
corresponds to 20% fading and resolution loss, the FWHM field gradients necessary for moderate resolution MPI we de-
widening is approximately equal to the fading. For example, if cided to use NdFeB permanent magnets with no field return.
there is 10% fading, there is approximately 10% loss in resolu-
tion. As the FFL nonideality increases beyond 20% fading and A. Magnet Design Criteria
resolution loss, resolution loss begins to accelerate when com- Gradient Strength: The native resolution of MPI poten-
pared to the fading. tially improves cubically with the nanoparticle diameter and
While the main magnet described in this paper has minimal linearly with the gradient strength [8], [17], [18]. To reach
resolution loss with Resovist SPIO tracer at the far extents of a target resolution, we must tailor the gradient strength to
the FFL, as new nanoparticles for MPI are developed the reso- the specific nanoparticle used as a tracer. Our previous work
lution loss can become a stronger design criteria. For example, a with a small-scale MPI imaging system with a 6 T/m gradient
theoretical nm SPIO nanoparticle and Resovist tracer achieved 1.6 mm resolution [8], with the
in the Berkeley FFL scanner would have over 20% resolution distribution of particle diameters in Resovist behaving experi-
loss at the edges of the FOV unless we shim the main gradient mentally equivalent to a nm nanoparticle. Thus, with
magnet. Resovist tracer and linear scaling of the resolution with the
gradient strength, we would require a 9 T/m gradient to achieve
C. Two Order of Magnitude Speed Improvement millimeter resolution. Unfortunately, this gradient strength is
The clearest benefit to a projection scan is an increase in challenging to achieve with water-cooled copper gradients.
speed. For a 3-D image, the FFP must scan every voxel in the Indeed, it is even difficult to achieve this gradient strength with
received image. The number of voxels in a typical medical high-grade NdFeB magnets. Hence, we decided to build a 2.35
imaging scan can be large , and so a 3-D scan T/m gradient, which was the largest gradient strength we could
Authorized licensed use limited to: Gebze Technical University. Downloaded on June 29,2023 at 13:22:50 UTC from IEEE Xplore. Restrictions apply.
1080 IEEE TRANSACTIONS ON MEDICAL IMAGING, VOL. 31, NO. 5, MAY 2012

Fig. 4. Main magnet is constructed in two halves using custom assembly jigs. The forces between the six NdFeB main magnet pieces during assembly are estimated
>
at 2500 newtons. (a) Jig to assemble the six NdFeB magnets rectangles into a main magnet. (b) Completed main magnet half.

build given our internal manufacturing capabilities and FOV NdFeB Magnet Construction: The physical structure was
requirements. designed in CAD software (Solidworks, Dassault Systèmes
According to x-space theory, a gradient of 2.35 T/m has an Solidworks Corp, MA) and constructed by the Berkeley Elec-
expected resolution of 3.5 8.0 mm when imaging Resovist trical Engineering Machine shop. To prevent eddy currents, the
tracer, which behaves like a nanoparticle with a lognormal mag- major components that experience time varying magnetic fields
netic core diameter distribution of nm [19]. This ex- are constructed of G10 fiberglass composite, which is strong,
pected resolution is for sample excitation and signal reception nonmagnetic, and nonconductive. The magnets were assembled
in only one instrument axis (the axis in this imager). The ex- with custom jigs (see Fig. 4) to overcome the significant mag-
pected resolution could be improved to 3.5 3.5 mm by sample netic forces estimated to be 2500 newtons between the twelve
excitation in two axes ( and in this imager), which we have rectangular NdFeB magnets during construction. The NdFeB
not yet implemented in order to limit system complexity. magnets are electrically and magnetically isolated from the
We note that increasing the gradient strength improves the transmit and receive coils by a 2-mm-thick copper eddy current
resolution of the system at the expense of longer scanning times. shield machined from high conductivity oxygen-free copper.
This is because maximum excitation strength (in mTpp) is lim- FFL Shift Magnets: The slow FFL shifting magnets are com-
ited by magnetostimulation and SAR heating of the patient [18]. mercially produced, off-the-shelf water-cooled electromagnets
Then, a strong gradient reduces the imaging speed two-fold (GMW Magnet 5403, GMW, San Carlos, CA) driven by a cur-
since increasing the gradient strength necessarily reduces the rent-controlled linear power amplifier (AE Techron 7780, AE
mean FFL velocity as well as increases the number of voxels in Techron, Elkhart, IN). The electromagnets achieve 60 mT at
each image. the center of the imaging bore with an peak power of approxi-
Field-of-View (FOV): The magnet FOV is defined by the bore mately 5 kW. This can electronically shift the FFL by up to 2.5
opening available to the sample, the maximum we can shift the cm in in less than 100 ms. Both the shift distance as well as
field free line using fast and slow FFL shifting magnets, and the speed could be increased with the use of a more powerful am-
limits of robot movements. We have chosen a free bore of 4 cm plifier.
to match the strength of the FFL shifting amplifier in the axis Transmit/Recv Subsystems: The transmit filter system is
as well as to enable mouse imaging. The FOV down the bore is constructed using a voltage-controlled linear power amplifier
limited only by robot movement. (AE Techron LVC5050) connected to a third-order lowpass
FFL Shift Linearity: As we shift the FFL, it is crucial that filter driving a resonant, water-cooled transmit coil tuned to 22
the FFL remain undistorted and homogeneous across the field kHz. The transmit chain achieves a 60 mTpp excitation strength
of view. The homogeneity of the shift field is important in the at a 3 kW continuous power level, for a pFOV of 2.5 cm in
FFL axis ( axis), as any field components along the FFL will the axis. The transmit and receive coil formers are printed
cause signal fading, as discussed above. The circular bore sim- using a rapid prototyping system ZPrinter 150 (Burlington,
plifies this since the length of the FFL can shorten as it shifts MA) and impregnated in epoxy. The receive coil is wound
from the center. Inhomogeneity in the shift axes ( axes) also in a gradiometer-like configuration so that it shares minimal
leads to distortion of the field free line. Since the nanoparticles mutual flux with the transmit coil. The transmit and receive
used in our system have a mean core diameter 20 nm, the ho- coils are easily changed as they are loosely inserted in the
mogeneity requirements during FFL shifting are not difficult to copper eddy current shield. The received signal is notch filtered
meet. For higher core diameter particles the design of the main at 22 kHz, preamplified (SR560, SRS, Sunnyvale, CA), high
FFL and slow shifting magnets could require shimming or a pass filtered (SRS SIM965), and amplified with a second stage
more complex FFL magnet. (SRS SIM911) before digitization.
Control Console: A data acquisition and control card
B. Scanner Construction (NIDAQ-6259, National Instruments, Austin, TX) synchro-
nously controls the resonant transmit coil, FFL slow movement
The completed magnet next to a diagrammatic representation magnet, mechanical stage translation, and data acquisition.
of the orientation and location of the NdFeB magnets that pro- Control, data analysis, and reconstruction are all completed in
duce the main field gradient is shown in Fig. 1. custom software (MATLAB, Mathworks, Natick, MA).
Authorized licensed use limited to: Gebze Technical University. Downloaded on June 29,2023 at 13:22:50 UTC from IEEE Xplore. Restrictions apply.
GOODWILL et al.: PROJECTION X-SPACE MAGNETIC PARTICLE IMAGING 1081

IV. METHODS
Simulation: Design of the magnet configuration was com-
pleted using a surface current model [22]. The surface current
model enables an accurate estimation of the magnetic fields pro-
duced by NdFeB magnets by modeling the field produced using
an equivalent surface current. Since the magnetic permeability
of NdFeB is near unity at , the surface current model
yields better than one percent accurate results at positions in-
side the bore of the magnet. This enables modeling the gradient
strength for arbitrary permanent magnet gradients simply and
rapidly without finite element modeling.
Magnet Characterization: Following construction, the
magnet was characterized with axial and transverse Hall Effect
Probes (Model 475, LakeShore Cryotronics, Westerville, OH)
mounted into a custom probe mount. The hall effect probes
were stepped throughout the FOV to measure all three field
components.
Imaging Pulse Sequence: The pulse sequence used to image
the phantoms is shown in Fig. 5. In the imaging sequence, a res-
onant transmit coil moves the FFL at 22 kHz across a 1.25 cm
region in the axis. Slow FFL shifting magnets enable slower
Fig. 5. 10 FPS pFOV pulse sequence: A representative portion of the pulse se-
movement of the FFL up to 2.5 cm in the axis. This enables quence used in the Berkeley Projection MPI scanner designed to observe dB/dt
the acquisition of an independent partial field-of-view (pFOV) limits in a mouse while achieving fast partial FOV frame rates. The pulse se-
of 2.5 cm 5.0 cm at 10 frames/s. The shift speed and dis- quence enables scanning a partial FOV of 2.5 cm x 5.0 cm at 10 frames/s. [Top]
Movement in the x axis is performed using a pair of water cooled solenoids that
tance in the axis are limited by the voltage and current, re- 6
shift up to 2.5 cm in approximately 100 ms. [Middle] Rapid sinusoidal move-
spectively, of the driving amplifier. Increasing both would en- ment at 22 kHz occurs in the z axis using a resonant, water cooled electromagnet
able faster scanning as well as a larger FOV. We step the patient 6
that shifts up to 1.25 cm. [Bottom] The sample is translated at 2.5 cm/s in the
table continuously during the scan in the axis to acquire a full z axis using a mechanical stage to enable scanning the full FOV.
FOV of arbitrary length.
Reconstruction: Each pFOV region is reconstructed by di-
system for resolution, linearity, and ability to image complex
viding the received signal by the instantaneous velocity of the
phantoms in a projection format.
FFL, and gridding the resulting signal to the instantaneous posi-
tion of the FFL [8], [18]. For a given pFOV, we grid a A. Magnet Characterization
pixel image. The resulting pFOVs are reconstructed
into a full FOV as described in our previous work [8] by recov- In Fig. 6 we see the gradient strength along each axis through
ering the lost fundamental signal, as detailed in the next para- the isocenter of the bore. The magnet axes can be seen labeled
graph. in Fig. 1, where is vertical along the field free line, is the
All MPI scanning methods must filter out the fundamental horizontal imaging axis, and is down the magnet bore. As ex-
tone (and any nanoparticle signal at the same frequency) since pected, the gradient along the field free line is small and stays
the received signal is contaminated by direct feedthrough from below 0.08 T/m across the 4 cm imaging FOV, which corre-
source to detector. Fortunately, it is possible to recover the sponds to 1.6 mT field error at the ends of the FFL. The actual
imaging information lost in this operation, which translates system homogeneity may be better than this value as we believe
directly to recovering each pFOV’s dc offset. We recover this that the hall effect sensor rotational alignment dominates our
lost information by using continuity and by assuming zero measurement error. For Resovist, this level of homogeneity re-
signal outside the FOV [8]. sults in less than 5% fading.
Mouse Imaging: Mice were prepared by injecting in a tail In Fig. 7, we see the characterization of the main gradient
vein with 20 undiluted Resovist tracer, followed by sacri- gradient compared to the magnetic field predicted by the surface
fice after 30 s or sacrifice after 5 min. Mice were sacrificed by current model. The field measurements agree with the surface
asphyxiation in a precharged chamber followed by cervical current simulations to better than 1%.
dislocation. Mice were prepared according to a mouse protocol B. Imaging Results
approved by the ACUC at UC Berkeley.
In Fig. 8, we see a test of the resolution of the projection
scanner. The measured resolution using Resovist tracer is 3.8
V. RESULTS AND DISCUSSION
mm in (down the bore), and 8.4 mm in (transverse to the
We characterized the magnet with two techniques. First, we bore), which matches reasonably with the expected system res-
measured the magnetic fields achieved by the main FFL magnet olution of 3.5 8.0 mm.
and compared them to the theoretically expected fields from the Fig. 8 also tests the linear shift-invariance (LSI) properties of
surface current model. Second, we characterized the imaging the projection scanner. The system is LSI only after the dc offset
Authorized licensed use limited to: Gebze Technical University. Downloaded on June 29,2023 at 13:22:50 UTC from IEEE Xplore. Restrictions apply.
1082 IEEE TRANSACTIONS ON MEDICAL IMAGING, VOL. 31, NO. 5, MAY 2012

Fig. 6. Field gradient in the principal axes of the completed FFL magnet, measured using a hall effect probe and numerical differentiation. The system achieves
6
a 2.35 T/m gradient strength in x and z , and less than 0.08 T/m in y . Across a 4-cm FOV, this amounts to a undesired magnetic field component of only 1.6 mT
at the edges of the FOV.

Fig. 7. Comparison of the Hall effect field measurements and the Biot–Savart simulation of the FFL magnet. The surface current model adequately models the
NdFeB magnets to better than 1%.

recovery algorithm [8] is applied. In medical imaging, LSI prop- inhomogeneity of approximately 20%, which we plan to fix in
erties are not guaranteed. For example, X-ray computed tomog- future revisions to the transmit and receive coils.
raphy (CT) imaging is a canonical example of a useful imaging In Fig. 10, we see a photo of a complex MPI phantom as
technique that is not, strictly speaking, LSI. However, making a well as a projection MPI image of the phantom. As can be
system approximately LSI is fundamental to medical imaging. seen in the image, the x-space MPI technique results in native
CT linearizes the X-ray signal by taking its logarithm. Ultra- MPI images over large FOVs. Unlike system function recon-
sound scales the returned signal by a function that is exponen- struction techniques [17], x-space image reconstruction requires
tial with time. And in x-space MPI we must velocity correct our no precharacterization of the imaging system and nanoparticle
received signal and we must also recover the low frequency data tracer. System function reconstruction hinges upon the assump-
in order to guarantee that our signal varies linearly with quantity tion that nothing will change in the MPI signal between the cal-
and position of tracer throughout the full FOV. ibration study (performed in saline) and the actual scan in more
In Fig. 9, we see a demonstration of the projection capabilities viscous blood. Indeed, it is well known that viscosity varies sub-
of the Berkeley FFL scanner. This image was taken by physi- stantially between arterial and capillary system. This could be
cally moving a point source along the projection axis , as well another challenge for the system matrix reconstruction. System
as down the bore. The resulting image shows how the shape of function techniques are used by other MPI imaging groups in-
the point spread function changes little as a function of position cluding [1], [4], [15].
along the projection. As predicted by x-space fading theory, the There are two primary artifacts in the native MPI image in
signal does not fade near the edges of the FOV. In fact, the signal Fig. 10(b): baseline recovery artifacts, and signal dropout arti-
increases near the edges of the FOV due to a receive coil facts. The baseline recovery artifact is due to noise and system
Authorized licensed use limited to: Gebze Technical University. Downloaded on June 29,2023 at 13:22:50 UTC from IEEE Xplore. Restrictions apply.
GOODWILL et al.: PROJECTION X-SPACE MAGNETIC PARTICLE IMAGING 1083

Fig. 10. Projection MPI image of a complex Cal phantom. (a) The physical
“Cal” phantom is filled with diluted Resovist (one part Resovist, nine parts DI
water). (b) Native x-space MPI image. Total acquisition time: 8 s (N =2aver-
2
aged frames). FOV: 10 cm 5.0 cm.

Fig. 8. Experimental resolution test of line sources separated below, above,

and strongly above the native resolution of the system. (a) The phantom was
constructed using laser cut channels filled with diluted tracer (one part Resovist,
nine parts DI water). (b) Native x-space MPI image with total acquisition time of
2
4 s. FOV: 10 cm 5.0 cm. (c) Line scan across position x =0 . The measured
resolution is 3.8 mm in the z axis (down the bore), and 8.4 mm in the x axis
(transverse to the bore). The experimental resolution reasonably matches the
2
expected system resolution of 3.5 8 mm.

Fig. 9. Image of a point source physically rastered in the y axis (vertical pro- Fig. 11. (a) Full-body MPI projection scan of a mouse injected in a tail vein
jection axis) and z axis (down the bore). The signal increases at the ends of the L g
with 20  (556  Fe) undiluted Resovist tracer and sacrificed after 30 s. The
line because the B receive coil sensitivity is not perfectly homogeneous. g
resulting MPI image shows a MPI-MRI visible fiducial (50  Fe), as well as
the outlines of the mouse brain, heart, and liver. (b) Full-body MPI projection
scan of an identically prepared mouse, but with a pause of 5 min after injection
and before sacrifice. In the projection MPI scan, it is apparent that the Resovist
drift in the received signal that result in incorrect baseline tracer is rapidly filtered from the blood stream by the liver. The MRI scan (taken
post-sacrifice) shows a dark liver as Resovist is a T agent. Total scan time: 8 s.
stitching. We can see this baseline artifact between the “C” and MPI FOV = 6 220 cm cm. Both images are scaled so that the 50 g fiducial
the “a,” where baseline recovery has added a slight background has a signal intensity of 1.
( 7% of full scale) to the image. Further averaging ( for
this image) reduces baseline recovery artifacts. There is also a
signal dropout artifact present at the top and bottom of the “C.” lightly deconvolved with Wiener filtering [23] to visually clean
We believe the signal dropout occurs due to reduced image up the background in the image due to the low resolution of
resolution in the transverse direction inherent to the collinear the system. Later, the animals were scanned with a 3T Siemens
point spread function [8]. MRI scanner. The MPI and MRI images were co-registered with
In Fig. 11, we see full-body MPI projection images of two MPI-MRI visible fiducials. The most striking feature of the MPI
mice injected in a tail vein with 20 undiluted Resovist. Fol- scans is their contrast, particularly when compared with the MRI
lowing a tail injection, one animal was sacrificed after 30 s, and images. The MPI technique produces perfect contrast, as tissue
the second after 5 min. The animals were imaged immediately has no MPI signal. The image of a mouse sacrificed after 30 s
in the Berkeley projection MPI scanner. The MPI images are shows clear perfusion of the tracer to the brain and heart, and
Authorized licensed use limited to: Gebze Technical University. Downloaded on June 29,2023 at 13:22:50 UTC from IEEE Xplore. Restrictions apply.
1084 IEEE TRANSACTIONS ON MEDICAL IMAGING, VOL. 31, NO. 5, MAY 2012

limited tracer in the liver. After 5 min circulation, the dominant course, important experimental considerations must be tested
location of the tracer is in the liver, and lower tracer levels in the with the larger nanoparticles, including increased relaxation ef-
brain. In the MRI images, the SPIO tracer causes a nonspecific fects and potential loss of super-paramagnetism.
dropout in the 30-s mouse and a full dropout of the liver
after 5 min. Since Resovist is a liver targeting agent for liver VI. CONCLUSION
cancer diagnosis using MRI, the rapid filtering of the tracer by In this paper, we have analytically described, designed,
the liver is not surprising. constructed, and tested a projection MPI imager. The system
Fig. 8, 10, and 11 demonstrate the high contrast inherent to reconstructs using our recently developed x-space systems
MPI. We believe this contrast is ideal for rapid angiography, theory, which enabled reconstruction without precharacteri-
which has the simple goal of obtaining a high contrast image zation of the imaging tracer, harmonics, or matrix inversion
of the blood vessels. The MPI technique may also find use in techniques. X-space MPI further ensures that the resulting
other techniques that aim to see a tracer with high contrast and image is Linear and Space-invariant, which is apparent when
sensitivity, such as dynamic contrast enhancement (DCE) de- imaging a complex “Cal” phantom.
tection of tumors or imaging of labeled stem cells. For these We believe this paper is the first to present experimental pro-
techniques it may be desirable to co-register an MPI image with jection MPI images in phantoms as well as in mice. Moreover,
a tissue image, and so MPI could be combined with a second we believe the image of the “Cal” phantom is an advance to the
imaging modality such as MRI or CT in a manner analogous to state of the art, both in image quality and the large field of view.
PET-CT or SPECT-CT (as demonstrated in Fig. 11). These results are encouraging as we develop sub-millimeter (na-
The improvement in imaging speed when using a projection tive resolution) x-space MPI systems, as well as wider-bore
imager over a 3-D imager with the same gradient strength is lim- x-space MPI scanners.
ited by the native resolution of the system. For example, with Projection MPI is a promising new imaging format for MPI
the current system resolution of 3.8 mm, a projection across the that could improve imaging speed by two orders of magnitude
5 cm FOV results in a speedup of 26 times when assuming a with no loss of SNR. The improvement in speed would be
2x oversampling of 1.9 mm per pixel. While this speedup falls useful in real-time applications such as MPI “fluoroscopy,”
short of the two order of magnitude speed improvement for an dynamic contrast enhancement, and interventional MPI. MPI
ideal clinical FFL scanner, this system is just a starting point as “fluoroscopy” holds particular promise as a safe rapid angiog-
we continue to develop new generations of projection MPI scan- raphy technique because the safety profile of SPIO nanoparticle
ners. The current system already images pFOVs at 10 frames/s tracers [27], [28]. Alternatively, we could improve 2-D SNR by
(FPS), which could be improved through the use of a more pow- an order of magnitude over 3-D MPI by using the faster acqui-
erful amplifier or a resonant circuit. Imaging speed for the full sition rate to increase averaging time. This mode of acquisition
FOV is presently limited by the mechanical translation of the would be particularly well suited for cell tracking with labeled
sample in the axis. Full FOV imaging time could be improved SPIO reporters where sensitivity is paramount.
through the construction of an additional set of electromagnets
to move the FFL over large distances in the axis instead of ACKNOWLEDGMENT
using mechanical movement. The authors would like to thank W. Yang and O. Carrasco-
While we believe that the image quality is an advance in Zevallos for their excellent assistance.
the state of the art, it is important to consider that the 3.8 mm
resolution of the Berkeley FFL scanner when imaging Reso- REFERENCES
[1] B. Gleich and J. Weizenecker, “Tomographic imaging using the non-
vist tracer pales when compared with state-of-the-art micro-CT linear response of magnetic particles,” Nature, vol. 435, no. 7046, pp.
and micro-MRI scanners. Typical resolution in these preclin- 1214–1217, Jun. 2005.
ical techniques are in the 100- m range or finer. Fortunately, [2] B. Gleich, J. Weizenecker, and J. Borgert, “Experimental results on fast
2-D-encoded magnetic particle imaging,” Phys. Med. Biol., vol. 53, pp.
there are a number of techniques we can use to increase the res- N81–N84, Feb. 2008.
olution of x-space MPI. First, deconvolution methods may be [3] J. Weizenecker, B. Gleich, J. Rahmer, H. Dahnke, and J. Borgert,
employed to improve MPI spatial resolution [1], [17], [20], but “Three-dimensional real-time in vivo magnetic particle imaging,”
Phys. Med. Biol., vol. 54, no. 5, pp. L1–L10, Feb. 2009.
these risk significant deconvolution artifacts and noise amplifi- [4] J. Weizenecker, J. Borgert, and B. Gleich, “A simulation study on the
cation. Gains in resolution may also be achieved by increasing resolution and sensitivity of magnetic particle imaging,” Phys. Med.
the gradient strength of the system at the expense of longer Biol., vol. 52, no. 21, pp. 6363–6374, Nov. 2007.
[5] T. F. Sattel, T. Knopp, S. Biederer, B. Gleich, J. Weizenecker, J.
scan times and reduced SNR. For example, we have built two 6 Borgert, and T. M. Buzug, “Single-sided device for magnetic particle
T/m FFP main field gradients for 3-D imaging of that are large imaging,” J. Phys. D: Appl. Phys., vol. 42, no. 2, p. 022001, Dec. 2008.
enough to image mice [7], [8]. Last, we believe spatial reso- [6] P. Goodwill, G. Scott, P. Stang, and S. Conolly, “Narrowband mag-
netic particle imaging,” IEEE Trans. Med. Imag., vol. 28, no. 8, pp.
lution may be less of a challenge in the future as nanoparticle 1231–1237, Aug. 2009.
research groups work to tailor magnetic nanoparticles for MPI [7] P. Goodwill and S. Conolly, “Narrowband magnetic particle imaging
[24]–[26]. Because the linear spatial resolution of MPI scales in a mouse,” in World Scientific: Proc. 1st Int. Workshop Magn. Particle
Imag., 2010.
cubically with particle diameter, larger diameter nanoparticles [8] P. Goodwill and S. Conolly, “Multi-dimensional x-space magnetic par-
offer extraordinary promise for improving spatial resolution. ticle imaging,” IEEE Trans. Med. Imag., vol. 30, no. 9, pp. 1581–1590,
For example, by moving to 34-nm diameter nanoparticles and Sep. 2011.
[9] J. Weizenecker, B. Gleich, and J. Borgert, “Magnetic particle imaging
imaging at a 6 T/m field strength, we theoretically expect to im- using a field free line,” J. Phys. D: Appl. Phys., vol. 41, no. 10, p.
prove spatial resolution 20-fold to 190 m spatial resolution. Of 105009, May 2008.
Authorized licensed use limited to: Gebze Technical University. Downloaded on June 29,2023 at 13:22:50 UTC from IEEE Xplore. Restrictions apply.
GOODWILL et al.: PROJECTION X-SPACE MAGNETIC PARTICLE IMAGING 1085

[10] E. D. Courant, M. S. Livingston, and H. S. Snyder, “The strong-fo- [20] T. Knopp, S. Biederer, T. F. Sattel, M. Erbe, and T. M. Buzug, “Pre-
cusing synchroton—A new high energy accelerator,” Phys. Rev., vol. diction of the spatial resolution of magnetic particle imaging using
88, no. 5, pp. 1190–1196, 1952. the modulation transfer function of the imaging process,” IEEE Trans.
[11] T. Knopp, M. Erbe, S. Biederer, T. F. Sattel, and T. M. Buzug, “Effi- Med. Imag., vol. 30, no. 6, pp. 1284–1292, Jun. 2011.
cient generation of a magnetic field-free line,” Med. Phys., vol. 37, no. [21] M. Bernstein, X. Zhou, J. Polzin, K. King, A. Ganin, N. Pelc, and G.
7, p. 3538, 2010. Glover, “Concomitant gradient terms in phase contrast MR: Analysis
[12] T. Knopp, T. Sattel, and S. Biederer, “Field-free line formation in a and correction,” Magn. Reson. Med., vol. 39, pp. 300–308, 1998.
magnetic field,” J. Phys. A: Math. Theor., 2010. [22] E. P. Furlani, Permanent Magnet and Electromechanical Devices: Ma-
[13] T. Knopp, M. Erbe, T. F. Sattel, S. Biederer, and T. M. Buzug, “Gener- terials, Analysis, and Applications. New York: Elsevier, 2001.
ation of a static magnetic field-free line using two Maxwell coil pairs,” [23] R. C Gonźalez and R. Eugene Woods, Digital Image Processing.
Appl. Phys. Lett., vol. 97, no. 9, p. 092505, 2010. Upper Saddle Rive, NJ: Prentice Hall, 2008.
[14] T. Knopp, S. Biederer, T. Sattel, J. Weizenecker, B. Gleich, J. Borgert, [24] R. M. Ferguson, K. R. Minard, and K. M. Krishnan, “Optimization of
and T. M. Buzug, “Trajectory analysis for magnetic particle imaging,” nanoparticle core size for magnetic particle imaging,” J. Magn. Magn.
Phys. Med. Biol., vol. 54, no. 2, pp. 385–397, Dec. 2008. Mater., vol. 321, no. 10, pp. 1548–1551, 2009.
[15] T. Knopp, T. F. Sattel, S. Biederer, J. Rahmer, J. Weizenecker, B. [25] K. M. Krishnan, “Biomedical nanomagnetics: A spin through possibil-
Gleich, J. Borgert, and T. M. Buzug, “Model-based reconstruction for ities in imaging, diagnostics, and therapy,” IEEE Trans. Magn., vol. 46,
magnetic particle imaging,” IEEE Trans. Med. Imag., vol. 29, no. 1, no. 7, pp. 2523–2558, Jul. 2010.
pp. 12–18, Jan. 2010. [26] R. M. Ferguson, K. R. Minard, A. P. Kandhar, and K. M. Krishnan,
[16] T. Knopp, S. Biederer, T. Sattel, J. Rahmer, J. Weizenecker, B. Gleich, “Optimizing magnetite nanoparticles for mass sensitivity in magnetic
J. Borgert, and T. Buzug, “2-D model-based reconstruction for mag- particle imaging,” Med. Phys., vol. 38, no. 3, pp. 1619–1626, Mar.
netic particle imaging,” Med. Phys., vol. 37, p. 485, 2010. 2011.
[17] J. Rahmer, J. Weizenecker, B. Gleich, and J. Borgert, “Signal encoding [27] E. Neuwelt, B. Hamilton, C. Varallyay, W. Rooney, R. Edelman, P. Ja-
in magnetic particle imaging: Properties of the system function,” BMC cobs, and S. Watnick, “Ultrasmall superparamagnetic iron oxides (US-
Med. Imag., vol. 9, no. 1, p. 4, 2009. PIOs): A future alternative magnetic resonance (MR) contrast agent for
[18] P. W. Goodwill and S. M. Conolly, “The x-space formulation of the patients at risk for nephrogenic systemic fibrosis (NSF)?,” Kidney Int.,
magnetic particle imaging process: 1-D signal, resolution, bandwidth, vol. 75, no. 5, p. 465, 2009.
SNR, SAR, and magnetostimulation,” IEEE Trans. Med. Imag,, vol. [28] M. Lu, M. H. Cohen, D. Rieves, and R. Pazdur, “FDA report: Fer-
29, no. 11, p. 1851, Nov. 2010. umoxytol for intravenous iron therapy in adult patients with chronic
[19] P. Goodwill, A. Tamrazian, L. Croft, C. Lu, E. Johnson, R. Pidaparthi, kidney disease,” Am. J. Hematol., vol. 85, no. 5, pp. 315–319, May
R. Ferguson, A. Khandhar, K. Krishnan, and S. Conolly, “Ferrohydro- 2010.
dynamic relaxometry for magnetic particle imaging,” Appl. Phys. Lett.,
vol. 98, 2011.

Authorized licensed use limited to: Gebze Technical University. Downloaded on June 29,2023 at 13:22:50 UTC from IEEE Xplore. Restrictions apply.