To image the nanoparticle distribution, a strong static magnetic gradient field is applied across the entire target. This selection field saturates the particles' magnetization everywhere besides a single field-free point (FFP). Drive coils then superimpose an oscillating magnetic field, and the nanoparticles' response is detected by a series of receive coils.

As superparamagnetic nanoparticles exhibit a nonlinear response to an oscillating field, the detected signal contains higher-order harmonics that can be exploited for imaging. Only particles whose magnetic response is not suppressed by saturation, i.e. those in the direct vicinity of the FFP, will generate harmonics that contribute to the measured signal. Scanning the FFP through the sample thereby enables reconstruction of a full tomographic image of particle distribution.

Since its introduction in 2005, MPI has undergone much development – in terms of hardware evolution, nanoparticle design and optimization of image reconstruction methods. At the SPIE Medical Imaging conference, held earlier this year in Lake Buena Vista, FL, a dedicated MPI conference session examined the latest progress in these key areas.

One sided approach

MPI scanners currently being investigated are based upon a set-up in which two pairs of transmit coils and two pairs of smaller receive coils are arranged symmetrically in a square configuration. The resulting two-dimensional field-of-view (FOV) is situated in between the coils, a design that imposes size limitations on the object being imaged.

Researchers at the University of Lübeck in Germany are taking a different approach. They are developing a single-sided MPI scanner, in which all coils required to generate the magnetic fields and receive the nanoparticle response are positioned on one side of the imaging target. Such a set-up means that the target no longer has to be small enough to fit inside the scanner.

"Our vision is to create a device that's seen in the same way as an ultrasound transducer," explained Thorsten Buzug, director of the University's Institute of Medical Engineering. "You just put the patient to one side and the field is produced in the body without needing coils on the other side."

To create the single-sided device, the researchers redesigned the MPI scanner geometry to comprise two concentric transmit coils. When direct currents in opposite directions are applied to each coil, superimposition of the induced magnetic fields creates FFPs on either side of the coil axis, one of which can be used for imaging. Applying an alternating current to one or both of the transmit coils generates the drive field and enables axial scanning of the FFP.

For two-dimensional imaging, the FFP must also be scanned in an orthogonal direction to the axial motion, which requires additional transmit coils. Buzug and colleagues simulated various single sided coil geometries, and found that a set of two D-shaped coils produced a magnetic field that can push the FFP left and right, enabling steering of the FFP though two-dimensional space. The current design has a penetration depth of a few centimetres, which the researchers are working to improve.

Ramping sensitivity

Buzug's team is also working on a means to increase the sensitivity of MPI by replacing the FFP with a field-free line (FFL). Scanning the region of interest with a FFL can increase the signal-to-noise ratio by an order of magnitude over FFP scanning.

To do this, the researchers had to come up with a means to generate, rotate and translate the FFL. Buzug explained that initial designs proposed to achieve this used 16 Maxwell coil pairs on a ring for FFL provide generation and rotation, plus two further pairs for translation. However, such a system can't actually be built as its power requirements are simply too high.

Instead, the researchers have come up with a design in which just three Maxwell coil pairs can produce a rotatable FFL, while a four-coil-pairs design resulted in the highest field quality. This system is expected to provide a significant increase in sensitivity at a moderate rise in electric power loss (compared with an equivalent FFP scanner). The team is now evolving the FFL design further and has built a prototype for initial testing.

Particle optimization

The other key component in the MPI process is the nanoparticle itself. At the SPIE meeting, Matthew Ferguson of the Krishnan Group at the University of Washington (Seattle, WA) took a close look at the issue of nanoparticle design. "There is clear evidence of the direct influence on imaging properties from the particles themselves," he explained.

While MPI resolution is intrinsically limited by the gradient strength, it is also dependent upon the size of the nanoparticle's magnetic core, and generally increases with core size. Likewise, MPI signal strength increases with nanoparticle size, with an optimum maximum size for each individual imaging set up.

As well as being optimized in size, nanoparticles for use in MPI should have minimal variability in volume distribution, as well as a magnetic relaxation time that's fast enough to respond to the excitation field. With these constraints in mind, Ferguson and colleagues are developing new types of biocompatible iron-oxide (Fe3O4) nanoparticles. The team has already created particles with superior performance to that of Resovist, a commercial tracer comprising superparamagnetic iron-oxide nanoparticles.

To create larger nanoparticles with a uniform and controllable size, the Washington researchers synthesize the particles in organic solvents. The particles are then transferred into aqueous solution by functionalization with biocompatible amphiphilic polymers.

The researchers characterized the properties of two magnetite nanoparticle samples. The first had an effective magnetic core size of 22.4 nm in diameter, a distribution in core size of ±7 nm, and an iron concentration of 1.35 mgFe/ml. The second had a core diameter of 20.1 nm, a ±5 nm distribution, and an iron concentration of 6.1 mgFe/ml. The corresponding values for Resovist are 14 nm, ±7 nm and 1.99 mgFe/ml.

MPI signals from the nanoparticles were measured using a custom-built MPI spectrometer that transmits at 25 kHz and measures up to 40 harmonics. The larger particles showed improved performance compared with Resovist, with more detectable harmonics and an increase in signal intensity of around one order of magnitude at various field strengths.

The team also measured the intrinsic spatial resolution of the nanoparticles in a DC biasing field. The intrinsic resolution of the magnetite particles was better than that of Resovist by more than 25% at higher harmonics. For the third harmonic – the largest and the one used for signal detection in their setup – image resolution at a gradient strength of 1.3 T/m was 12.7 mm for Resovist, 10.8 mm for the 22 nm particle, and 7.7 mm for the 20 nm particle. At 2.6 T/m, the corresponding figures were 6.3, 5.4 and 3.8 mm, respectively.

Ferguson noted that these initial results indicate the potential for sub-millimetre spatial resolution from the new nanoparticles. The 22 nm magnetite particle, for example, exhibited a resolution of 0.4 mm at a gradient strength of 6 T/m (for the 37th harmonic).

Last April, Ferguson, together with Kannan Krishnan and Amit Khandhar (also part of the Krishnan Group), founded a start-up company called LodeSpin Labs to commercialize their particles. "We are continuing to optimize particle characteristics such as size and uniformity to drive performance," he concluded.

• Check back on the 27 April for the second article in this two-part series, which will take a look at the evolution of x-space MPI.