Ultra Low Field team at CAI. From left to right: Dr Viktor Vegh, Reuben Pellicer Guridi, Dr Michael Vogel, Shrinath Kadamangudi, Professor David Reutens and Aiden Carey

Why ultra low field?

Magnetic resonance imaging (MRI) and nuclear magnet resonance (NMR) spectroscopy are core investigative methods. Applied in a broad range of applications, MRI/NMR is the most versatile non-invasive and non-radiating imaging technique to obtain high resolution anatomical and functional images for medical and neuroscience, or evaluate molecular structures in material science, chemistry and biology. However, strong magnetic fields above 1 Tesla are required generated only by superconducting magnets. This makes MRI/NMR instruments bulky, heavy and very expensive to purchase, operate and maintain, limiting their use to major research or clinical facilities.

In light of these limitations, ultra-low field-NMR/MRI, with applied magnetic fields of less than 10 mT, was developed about 10 years ago. As an alternative imaging and scanning solution to high field instrumentation ULF-MRI/NMR has the potential to exploit a number of features unique at this field regime, such as

  • acquisition of images with higher contrast-to-noise ratios (CNR)
  • imaging in the presence of metals
  • concurrent use with other modalities like magnetoencephalography (MEG)
  • no susceptibility artefacts

Probably the most promising feature is the possible Larmor frequency overlap and match with the rates of change of slow chemical and biological processes such as diffusion, protein folding, and neuronal activity. This provides new opportunities to study these slowly evolving dynamic processes and allows to explore new imaging contrast mechanisms.

(a) SPMA design for generating the magnetic field environment for ULF relaxometry. (b) SPMA prototype design with COMSOL, with additional permanent magnets, exemplifying encoding (instead of gradients) field generation with permanent magnets. Constructed SPMA prototype to generate (c) sample pre-polarisation and (d) measurement field. 

What are the key challenges in ULF today?

The sample signals at ULF-NMR/MRI are very weak, due to the low applied magnetic field and subsequently small achievable sample magnetisation which limits achievable signal strength. Currently ULF research focus on three practical approaches to boost the signal-to-noise ratio (SNR):

  • Implementation of highly sensitive magnetometers, and
  • Sample pre-polarisation, applying a strong (~0.1T) pulsed magnetic field before the measurement.

However, the use of these magnetometers like superconducting quantum interference devices (SQUID’s) and atomic magnetometers (AM) with compact resistive coils for sample pre-polarisation are suboptimal solutions, since these sensors require protective circuits and shielded environments. Moreover, the presence of highly conductive material might interfere with the magnetometers. Also, strong sample pre-polarisation results in irreversible energy dissipation into heat, requiring cooling devices and possible sample protection


ULF Research Aims at CAI

Comparison of Bp and Bm generated by SPMA prototype (left hand side) with numerical simulation
a) Field direction of Bp indicated by array of needles (top, inset, red circle) and surface plot of COMSOL (bottom inset, blue circle)
b) Field direction of Bm measured (top inset, red circle) and similated (bottom inset, blue circle)

Our research aim at the Centre for Advanced Imaging (CAI) is develop ULF-MRI without superconductive and resistive technology towards a truly portable and economic imaging solution. Our research is focused on:

  • Optimise NMR signal yields by determining optimal sensor locations and orientations relative to sample size and shape

We studied numerically the temporal magnetic field evolution during sample demagnetisation across a range of sample sizes and shapes [1]. Figure 1 details how sample shape and size influence the magnetic field distribution and therefore determines optimal location and orientation of the magnetometer. We demonstrated that an optimal sample shape and size exist in terms of maximal signal strength for different sensor orientations. For the various sensor locations tested the variation in signal strength can be as large as 90 %, highlighting the importance of numerical studies in the design of ULF-NMR systems.

  • Developing a novel coil based magnetometer, optimised for the ULF regime (> 450 KHz)

With our in-house developed global optimisation method a tuned prototype magnetometer was built with a measured sensitivity of less than 4 fT/(Hz)1/2 very close to the numerical prediction.

  • Designing and constructing a dynamic permanent magnet based ULF-MRI instrument

Halbach arrays, an arrangement of individual permanent magnets, are known to produce stronger and more homogeneous fields compared to equivalent resistive coils. We extended the static Halbach array concept by the dynamic small permanent magnet array (SPMA) in which multiple magnetic field configurations for ULF are generated by mechanically rearranging the magnets [3]. A manual prototype was built and tested showing a very good match with the numerical results. Three patent (A-C) applications originating from the SPMA technology were filed.

  • Sensitivity of prototype coils. Sensitivity comparison between measured and estimated values for a tuned voltage-to-voltage coil with ideal and lossy capacitators, see right hand side
    Developing novel image reconstruction methods and approaches

Standard MRI image reconstruction is based on fast Fourier transforms (FFT) to ensure fast acquisition but it relies on highly linear gradient fields for spatial encoding. However, the encoding fields generated by our permanent magnet array are non-linear, hence, it requires a different (for instance, back projection) image reconstruction method. We are currently exploring new ways with non-linear fields to (a) enhance locally spatial resolution to enhance image quality and (b) acquisition speed by using other data analysis methods like machine learning and deep neuronal networks.