Field Free Line Magnetic Particle Imaging Characterisation and imaging device up-scaling

First presented in 2005, magnetic particle imaging (MPI) is a new imaging modality able to acquire and reconstruct the distribution of a magnetic tracer included in a non-ferromagnetic volume. In 2009, Weizenecker et al. presented a video showing a tracer distribution flowing through the beating heart of a mouse. It was acquired in a 20.4x12.0x16.8 mm^3 volume at 46 frames per seconds. The spatial encoding was done using a field free point (FFP) moved along a Lissajous trajectory. Since then, many efforts have been done to improve the technology. To further increase the sensitivity of the acquisition process, it has been proposed to replace the FFP by a field free line (FFL). In 2014, Weber et al. presented the first continuously rotating FFL scanner, acquiring and displaying 50 circular images per seconds within a Field of View (FOV) with a diameter of 25 mm.

However, to image a tracer distribution in the human body, up-scaled version of those scanners have to be realised. In this thesis, the design of two FFL scanners having a bore diameter of 173 mm and 500 mm are investigated, to highlight and investigate some of the challenges awaiting the up-scaling process.

The first part of this work defines precisely the acquisition sequence and the imaging properties of the MPI systems.
It is shown that the coils designed in this work have magnetic-field topologies which cannot be considered as ideal.
This, in turn, has an impact on the reconstruction of the tracer distribution.
Using a model based approach which assumes perfectly straight lines, as the filtered back projection, leads to artefacts in the reconstructed images.
Using a system-function reconstruction approach with properly resolved spectra suppresses those artefacts, but the acquisition rate is divided by two.
To cover a larger FOV, a new FFL sequence using focus fields is introduced.
It requires the use of a system function reconstruction, as no model-based reconstruction are available for a generic sequence.
The first reconstruction shows artefacts, which have to be investigated in further studies.

The second part of this work is focused on safety limitations associated with MPI scanners. The main foreseen safety risk associated with an MPI scanner is the peripheral nerves stimulation (PNS). As no precise data exist on the PNS thresholds for the frequencies and magnetic field-strengths usable in MPI, an approximation is proposed to evaluate the PNS risk of MPI sequences. Applying this relation to a human-sized FFL scanner imaging an FOV of 400x405 mm^2 10 times per second, it is found that the patient will probably experience PNS. Further experimental validations have to be carried out to determine more precisely the PNS thresholds for such sequences. Afterwards, the sequences may be adapted to avoid PNS in the patient. A first example of such an adjustment is also presented.

The third part of this work is focused on the construction of the electromagnets used in an FFL scanner with a bore diameter of 173 mm. A boundary element method formulation is found to be adequate to optimise all the necessary coils. A scanner design with a gradient on the line of 0.8 T.m^(-1) and two 15 mT drive fields has been implemented. The fields strengths are mainly limited by the cooling infrastructure available and the technical complexity associated with all the parts of the scanner. The actual design requires around 40 kW of electrical power. The rotation frequency of the line is actually limited by the filters, power factor correction elements and the shielding of the different elements.