MPI_Tx_Rx_Coils
The combination of a transmit (also referred to as "drive" and "Tx") and receive ("Rx") coils form the basis of signal production and acquisition in MPI. Drive coils excite the SPIONs within the FOV with high amplitude AC magnetic fields (~8mT at 25kHz) and the receive coil acquires the signal while attenuating signal being generated directly from the drive coils.
Figure 1: A 3/4 rendering of the Tx/Rx assembly showing the coil arrangement and structural support. The rendering in Inventor caused the structure to appear hollow-- this is not the case, they are 3-D printed with infill.The angled holes are for set-screws to hold it axially and centered in the bore.
For the following coils, we have an overall free bore diameter of 50mm. This is to ensure a rat head can fit into our scanner.
To excite the particles within the FOV, there is a singular goal which is to create a drive field at sufficient amplitude. The word "sufficient" is intentionally left vague as the drive amplitude is heavily informed by the application. For instance, if you planning on modifying the small-bore imager for minimum power such as Graeser et al. does for their system, a drive field of 6-8mT might be "sufficient"(1). By driving higher amplitudes you gain SNR, but interestingly has been shown to lower spatial resolution(2). Further, if you build the MPS system, it is feasible to characterize the particles you plan to use and then optimize the excitation field amplitude based on those data.
The everpresent design limitation is heating. Heat is the name of the game with power electromagnet design. With high amplitudes, you get substantial thermal drift (especially without water cooling) and potentially coil damage if the heat really gets out of hand. Our current drive coil is cooled by wrapping tubing around the outer layer and pumping cool water through those tubes. Malhotra et al. describe an interesting method for water cooling in their 2019 INSPECT paper(3).
The most significant influence on the heating optimization question is wire diameter (assuming Litz)-- thin wire allows for higher turn density which means lower requisite current for a given magnetic field. But for a given current, heat density is proportional to diameter to the fourth power (Resistance is proportional to diameter squared, power is proportional to resistance squared), yet there is a geometric aspect to design to consider as well such as proximity effect and ensuring the coolant can sufficiently access the coils. The drive coil should be designed in tandem to the filter, because a higher "matching ratio" will enable an overall much more efficient transfer of power. See the next section for discussion on the filter.
With that said we have designed the following drive coil (The mechanical design has changed! The wire spacing/arrangement is the same, but the reader should reference the Inventor files for the mechanical aspects.):
The purpose of the filter is to allow for the drive coil to be powered by a maximally pure and sufficiently powerful current. The power comes from an amplifier (AE Techron 7224 or 7548) which, without a proper filter, would severely degrade the signal quality and be incapable of driving the current directly. Indeed, the limiting factor for the small-bore imager for a significant period of time was this filter. Please see our more thorough write-up on this element of the design-- its importance cannot be overstated
The Rx coil in some ways has more subtle design decisions involved. Below is our design but keep in mind there are certainly options to explore. For one, wire diameter is a pretty major question for different (yet similar) reasons. There is the tradeoff of turns versus resistance again but this time the resistance plays a role because eventually, thermal noise can limit the sensitivity. In addition to the resistive noise, the inductive impedance of the Rx coil (and filter) influences the performance of the preamplifier, and so should be "noise matched". A good discussion of that is published by Zheng et al. (4)
Early in the building process the Rx filter was a component that was significantly debated. Either it can be designed to resonate at F3 to improve the sensitivity of the system (by essentially amplifying the signal at that harmonic), although this approach rejects the other harmonics which contain spatially relevant information. The other approach more often used within the field of MPI is to use a broadband receive chain and a notch filter to reduce drive feedthrough. A third option is to receive broadband without notching out the drive frequency, and attenuating it purely by geometric decoupling ('Geometric decoupling' refers to the gradiometer canceling out the homogeneous field at the drive frequency.) with active cancellation further attenuating drive coil feed-through.
We utilize a design which notches the fundamental and passes all frequencies up until about 300kHz, after which an anti-aliasing filter heavily attenuates the signal.
A reference for a successful filter design is here-- presented at IWMPI
We have developed a low-noise amplifier ("Pre-amp") in house to be used with our imager. The preamp is loosely based on the design published in (4), in that it takes multiple op-amps in parallel and sums the signal.
(1) Graeser et al. Human-sized magnetic particle imaging for brain application, 2019
(2) Croft et al. Low drive field amplitude for improved image resolution in magnetic particle imaging, 2015
(3) Malhotra et al. Tracking the Growth of Superparamagnetic Nanoparticles with an In-Situ Magnetic Particle Spectrometer (INSPECT)
(4) Zheng et al. Optimal Broadband Noise Matching to Inductive Sensors: Application to Magnetic Particle Imaging, DOI: 10.1109/TBCAS.2017.2712566