NeuroImage (Nov 2021)

Precision magnetic field modelling and control for wearable magnetoencephalography

  • Molly Rea,
  • Niall Holmes,
  • Ryan M. Hill,
  • Elena Boto,
  • James Leggett,
  • Lucy J. Edwards,
  • David Woolger,
  • Eliot Dawson,
  • Vishal Shah,
  • James Osborne,
  • Richard Bowtell,
  • Matthew J. Brookes

Journal volume & issue
Vol. 241
p. 118401

Abstract

Read online

Optically-pumped magnetometers (OPMs) are highly sensitive, compact magnetic field sensors, which offer a viable alternative to cryogenic sensors (superconducting quantum interference devices – SQUIDs) for magnetoencephalography (MEG). With the promise of a wearable system that offers lifespan compliance, enables movement during scanning, and provides higher quality data, OPMs could drive a step change in MEG instrumentation. However, this potential can only be realised if background magnetic fields are appropriately controlled, via a combination of optimised passive magnetic screening (i.e. enclosing the system in layers of high-permeability materials), and electromagnetic coils to further null the remnant magnetic field. In this work, we show that even in an OPM-optimised passive shield with extremely low (<2 nT) remnant magnetic field, head movement generates significant artefacts in MEG data that manifest as low-frequency interference. To counter this effect we introduce a magnetic field mapping technique, in which the participant moves their head to sample the background magnetic field using a wearable sensor array; resulting data are compared to a model to derive coefficients representing three uniform magnetic field components and five magnetic field gradient components inside the passive shield. We show that this technique accurately reconstructs the magnitude of known magnetic fields. Moreover, by feeding the obtained coefficients into a bi-planar electromagnetic coil system, we were able to reduce the uniform magnetic field experienced by the array from a magnitude of 1.3±0.3 nT to 0.29±0.07 nT. Most importantly, we show that this field compensation generates a five-fold reduction in motion artefact at 0‒2 Hz, in a visual steady-state evoked response experiment using 6 Hz stimulation. We suggest that this technique could be used in future OPM-MEG experiments to improve the quality of data, especially in paradigms seeking to measure low-frequency oscillations, or in experiments where head movement is encouraged.

Keywords