Search

MagnetoEncephaloGraphy (MEG) provides a measure of electrical activity in the brain at a millisecond time scale. From these signals, one can non-invasively derive the dynamics of brain activity. Conventional MEG systems (SQUID-MEG) use very low temperatures to achieve the necessary sensitivity. This leads to severe experimental and economical limitations. A new generation of MEG sensors is emerging: the optically pumped magnetometers (OPM). In OPM, an atomic gas enclosed in a glass cell is traversed by a laser beam whose modulation depends on the local magnetic field. MAG4Health is developing OPMs using Helium gas (4He-OPM). They operate at room temperature with a large dynamic range and a large frequency bandwidth and output natively a 3D vectorial measure of the magnetic field. In this study, five 4He-OPMs were compared to a classical SQUID-MEG system in a group of 18 volunteers to evaluate their experimental performances. Considering that the 4He-OPMs operate at real room temperature and can be placed directly on the head, our assumption was that 4He-OPMs would provide a reliable recording of physiological magnetic brain activity. Indeed, the results showed that the 4He-OPMs showed very similar results to the classical SQUID-MEG system by taking advantage of a shorter distance to the brain, despite having a lower sensitivity.

The performance of EEG source reconstruction has benefited from the increasing use of advanced head modeling techniques that take advantage of MRI together with the precise positions of the recording electrodes. The prevailing technique for registering EEG electrode coordinates involves electromagnetic digitization. However, the procedure adds several minutes to experiment preparation and typical digitizers may not be accurate enough for optimal source reconstruction performance (Dalal et al., 2014). Here, we present a rapid, accurate, and cost-effective alternative method to register EEG electrode positions, using a single digital SLR camera, photogrammetry software, and computer vision techniques implemented in our open-source toolbox, janus3D. Our approach uses photogrammetry to construct 3D models from multiple photographs of the participant's head wearing the EEG electrode cap. Electrodes are detected automatically or semi-automatically using a template. The rigid facial features from these photo-based models are then surface-matched to MRI-based head reconstructions to facilitate coregistration to MRI space. This method yields a final electrode coregistration error of 0.8 mm, while a standard technique using an electromagnetic digitizer yielded an error of 6.1 mm. The technique furthermore reduces preparation time, and could be extended to a multi-camera array, which would make the procedure virtually instantaneous. In addition to EEG, the technique could likewise capture the position of the fiducial markers used in magnetoencephalography systems to register head position.