Matthew J. Brookes

Alpha band (8–13 Hz) electroencephalography (EEG) oscillations are central to cognition, but their generating mechanisms are poorly understood. Most laminar studies investigating origins of alpha oscillations rely on invasive animal intracranial recordings. To relate these findings to human alpha generation, non-invasive techniques need to be developed. Ultra-high-field (UHF, 7T) laminar functional magnetic resonance imaging (fMRI) enables cortical depth resolved brain responses, and, combined with simultaneous EEG, offers new insight into human alpha generation mechanisms. This work presents a framework for studying the generating mechanisms of electrophysiological signals non-invasively in humans using simultaneous EEG laminar fMRI. Data were acquired on 10 participants during an eyes-closed/eyes-open paradigm. In 9 of 10 participants, the quality of EEG and Blood Oxygenation Level Dependent (BOLD) fMRI data were sufficient to observe a significant negative correlation between EEG alpha power and the BOLD signal in visual cortex grey matter. To improve depth specificity and mitigate pial-surface BOLD signal increases due to draining veins, we applied “deveining”. The effects of each of the deveining analysis steps on the cortical depth profiles of the negative alpha–BOLD correlations were studied. The largest effect arose from excluding voxels in the tissue immediately surrounding veins. After deveining, the negative alpha–BOLD correlations were significantly weaker in the middle cortical depths than in superficial and deep depths. This depth dependence was absent when modelling the task with a boxcar regressor rather than EEG alpha power, suggesting specificity to spontaneous electrophysiological (alpha) fluctuations. In conclusion, we have established a method to non-invasively interrogate the origins of electrophysiological signals. Our alpha–BOLD depth profiles during the eyes-open/eyes-closed task (strongest correlations in superficial and deep layers) are consistent with activity in cortical depths typically associated with feedback/top–down processing, but we cannot definitively demonstrate causality from these data.

Objective . Magnetoencephalography (MEG) measures magnetic fields generated by current flow in neural assemblies, providing a non-invasive means to track brain function. In recent years, the MEG field has been revolutionised by a new class of field sensor called an optically pumped magnetometer (OPM). OPMs are small and lightweight, allowing for the construction of wearable and reconfigurable arrays of ‘triaxial’ field sensors (that measure magnetic fields along three axes simultaneously). These systems can be used to capture MEG data across the lifespan, from infancy to adulthood. With such multi-axis sensing arrays coming into routine use, here we aimed to determine how dense a sensor array is required to adequately capture the spatial variation of neuromagnetic fields across the scalp. Approach . We simulated neuromagnetic fields based on realistic head geometries of individuals aged 15 months to 34 years. We then used two independent approaches to determine the sensor spacing required to capture the highest spatial variation of the field across the scalp surface: one method was based on spherical harmonic decomposition of neuromagnetic fields, and the other based on evaluating sensor redundancy in densely sampled arrays. Main results . Both methods agree in showing that the required sensor separation increases with age, from ∼40 mm in a 1 year-old to ∼55 mm in an adult. Significance . The sensor spacings obtained in our simulations represent upper bounds for the required values: separations above these bounds will not adequately capture spatial field variation; however, reducing sensor separation below these limits will continue to improve performance. These findings are important for the future development of OPM-MEG technology, particularly the construction of systems for neurodevelopmental applications.

Neural fingerprinting seeks to identify individuals based on measurements of brain activity, exploiting the fact that aspects of brain function unique to an individual remain stable across repeated scans. Magnetoencephalography (MEG) is a powerful technique for fingerprinting. However, most MEG studies have used conventional (SQUID-based) MEG technology and typically rely on data aggregated over time, overlooking the rich temporal dynamics available in MEG. Here, using SQUID-MEG and the more recently introduced OPM-MEG, we showed fingerprinting was possible within and between modalities using static (time-aggregated) features; this is consistent with previous work. We further asked whether fingerprinting was possible based on network dynamics, estimated using a canonical hidden Markov model (CHMM). Results showed that the CHMM-derived networks provide a better fit to SQUID data than to OPM data; however, this effect was small and to be expected given the canonical networks were trained on SQUID data. We further showed that fingerprinting was possible within and between modalities using CHMM-derived state activation time courses and state power spectral densities. However, state summary statistics and transition probabilities only supported within modality fingerprinting. Our findings suggest that subject-specific information is preserved within CHMM states and across MEG technologies. This represents an important step towards advancing our understanding of brain dynamics, and particularly the brain networks delineated by the CHMM. The study also supports the future use of the CHMM for processing and interpretation of OPM-MEG data.

Tourette Syndrome (TS) is a neurodevelopmental hyperkinetic disorder characterised by involuntary tics. These tics are thought to arise due to hyperexcitability of the motor cortex due to disinhibition within the cortico-striato-thalamo-cortical pathway. Given the involvement of motor circuits in this disorder, we explored whether there is a difference in the oscillatory dynamics of voluntary movements in people with tic disorders. We recorded optically pumped magnetometer magnetoencephalography during cued voluntary finger abductions in individuals with tic disorders and age- and sex-matched neurotypical controls. We analysed the data both using conventional time frequency analysis and using hidden Markov modelling to explore the difference in beta burst characteristics and dynamics. Whilst no differences were seen between groups during conventional analysis, we demonstrated an increase in beta burst duration during the post-movement beta rebound within the contralateral motor cortex in individuals with TS suggestive of increased inhibition following movement. We also show evidence of increased contralateral sensorimotor functional connectivity and reduced connectivity from and between frontal regions, as measured using coincident beta bursts. This is suggestive of disrupted functional connectivity within frontal control networks in individuals with TS.

Beta oscillations are a fundamental feature of brain activity, linked to long-range connectivity within canonical networks, inhibition of sensorimotor cortices and the maintenance of a stable sensorimotor state in situations where the external world is predictable. The importance of beta oscillations in brain function is underscored by observations of their perturbation in neurological and psychiatric disorders. However, the precise role played by beta activity, particularly in mediating complex or skilful movements, remains incompletely understood. Here, we used a newly developed wearable optically pumped magnetometer-based magnetoencephalography (OPM-MEG) system to measure beta dynamics as participants learned to play a musical instrument. Twenty-two novice players took part in a study in which OPM-MEG data were recorded during two scanning sessions, while participants attempted to play a tune on a violin. Between the two sessions, participants received a violin lesson from an expert teacher. Results showed that robust data could be acquired during this naturalistic task, with beta oscillations decreasing in amplitude during movement and increasing upon movement cessation, as expected. Moreover, beta power whilst playing, in the motor and pre-motor areas, was significantly elevated after the lesson compared to before, and the movement-related modulation of beta amplitude was more pronounced after the lesson. These findings align with predictive coding models which suggest that beta amplitude should increase when individuals have greater certainty over the movements they carry out. Our study adds to an expanding literature on the role of beta oscillations and provides further evidence for the utility of OPM-MEG in naturalistic neuroscience.

Magnetoencephalography using optically pumped magnetometers (OPM-MEG) is gaining traction as a neuroimaging tool, with potential for improved performance and practicality compared with conventional instrumentation. However, OPM-MEG has so far lagged conventional-MEG in terms of the number of independent measurements of the MEG signal that can be made across the scalp (i.e., the number of channels). This is important since increasing the channel count offers improvements in sensitivity, spatial resolution, and coverage. Unfortunately, increasing channel count also poses significant technical and practical challenges. Here, we describe a new OPM-MEG system which exploits 3-axis sensors and integrated miniaturised electronic control units to measure MEG data from up to 384 channels. We also introduce a high-speed calibration method to ensure that the fields measured are high fidelity. We initially validate this system using a phantom: specifically, we localise a set of magnetic dipoles and compare the MEG-derived distances between them with the ground truth, demonstrating an accuracy of ~1 mm. We further show that the correlation between a mathematical model of the phantom and the magnetic fields measured is >0.998. Secondly, we demonstrate utility of our system for human MEG acquisition. Via measurement of visual cortex activity, we demonstrate improvements in sensitivity afforded by a higher channel count, and via a movie-watching experiment we show increased spatial resolution. In sum, we report the first OPM-MEG system with a channel count larger than that of typical conventional MEG devices. This represents a significant step towards OPMs becoming the sensor of choice for MEG.

Alpha band (8-13Hz) electroencephalography (EEG) oscillations play a key role in cognition, but their generating mechanisms are still poorly understood. Most studies investigating laminar origins of alpha oscillations have been conducted on animals using invasive intracranial recordings. To relate these findings to human alpha generation, non-invasive techniques need to be developed. Layer functional Magnetic Resonance Imaging (fMRI) at ultra-high field (UHF, 7T) allows for the interrogation of brain responses across cortical depths and combined with simultaneous EEG, provides the opportunity to gain new insight into human alpha generation mechanisms. This work establishes a framework to study the generating mechanisms of electrophysiological signals non-invasively in humans using simultaneous EEG layer-fMRI. Data were acquired on 10 participants during an eyes closed/eyes open paradigm. We showed that in 9/10 participants the quality of EEG and Blood Oxygenation Level Dependent (BOLD) fMRI data were sufficient to observe a significant negative correlation between EEG alpha power and the BOLD signal in visual cortex grey matter to the eyes open/eyes closed task. “Deveining” was performed to overcome the increase in BOLD signal toward the pial surface due to draining veins, and the effects of each of the steps in the deveining analysis on the cortical depth profiles of the negative alpha-BOLD correlations studied. The largest effect was dependent on the exclusion of voxels in the tissue immediately surrounding veins. Following deveining, the cortical depth profiles showed the negative alpha-BOLD correlations were significantly weaker in the middle depths compared with deep and superficial depths. When a boxcar rather than EEG alpha power was used to model the task, this depth-dependence was not seen, suggesting this was specific to spontaneous alpha-power modulations. In conclusion, we have established a method to non-invasively interrogate the origins of electrophysiological signals. Our alpha-BOLD depth profiles suggest the alpha signal to an eyes open-closed task is generated in superficial and deep layers suggesting top-down processes.

Cuboid shields formed from sheets of materials of high magnetic permeability are used to provide passive magnetic shielding in multiple applications, ranging from fundamental physics to functional neuroimaging. Current-carrying electromagnetic coils are often used in combination with passive shielding, since coils placed inside or outside the shield can be used to cancel unwanted residual magnetic fields, or to produce desired bias fields inside the shield. Understanding the effect of the high-permeability material on the time-varying fields produced by coils is important for implementing this hybrid active and passive shielding approach. Here, we use analytic expressions describing the interaction of magnetic fields with an infinite plane of high relative permeability (μr≫1) to characterize the behavior of shielding systems using coils attached to the inner or outer face of a cuboid shield in the regime (length scales of 0.01–1 m over a frequency range of 0–500 Hz) that is relevant for biomagnetic measurements. Theoretical predictions were compared with results from numerical simulations and experimental measurements made using a 55-cm-side cubic shield formed from 1.5-mm-thick sheets of MuMetal. The results confirm that the fields produced by internal coils are well predicted by the method of mirror images, with maximum deviations from perfect reflection of ∼2% in the magnitude and −0.02 rad in the phase of the field measured at ∼16 cm from an 8.5-cm-diameter circular coil driven at 440 Hz. The analytic expression indicates that the spatial variation of the shielded field from an external coil can be simply calculated from the two-dimensional Fourier transform of the field from the coil. This was confirmed by simulations and experimental measurements made using an 8.5-cm-diameter coil driven at frequencies of 0.1, 1, 5, and 10 Hz. This simple Fourier relationship facilitates the design of external coils that generate target-field variations inside the shield. Successful implementation of this approach was demonstrated experimentally using an external coil that was designed to produce the same field inside the shield as an unshielded 8.5-cm-diameter coil. Numerical simulations also confirmed the prediction that the shielded field distant from a circular coil is equivalent to that from a magnetic monopole located at the coil center. These findings offer new insight for the design and implementation of magnetic shielding using active and passive systems, particularly for measurements of biomagnetism.

Magnetoencephalography using optically pumped magnetometers (OPM-MEG) is gaining significant traction as a neuroimaging tool, with the potential for improved performance and practicality compared to conventional instrumentation. However, OPM-MEG has so far lagged conventional-MEG in terms of the number of independent measurements of magnetic field that can be made across the scalp (i.e. the number of channels). This is important since increasing channel count offers improvements in sensitivity, spatial resolution and coverage. Unfortunately, increasing channel count also poses significant technical and practical challenges. Here, we describe a new OPM-MEG array which exploits triaxial sensors and integrated miniaturised electronic control units to measure MEG data from up to 384 channels. We also introduce a high-speed calibration method to ensure the that the fields measured by this array are high fidelity. The system was validated using a phantom, with results showing that dipole localisation accuracy is better than 1 mm, and correlation between the measured magnetic fields and a dipole model is >0.998. We then demonstrate utility in human MEG acquisition: via measurement of visual gamma oscillations we demonstrate the improvements in sensitivity that are afforded by high channel density, and via a moviewatching paradigm we quantify improvements in spatial resolution. In sum, we show the first OPM-MEG system with a channel count larger than that of typical conventional MEG devices. This represents a significant step on the path towards OPMs becoming the sensor of choice for MEG measurement.

Source modelling in magnetoencephalography (MEG) infers the spatial origins of electrophysiological signals in the brain. Typically, this requires an anatomical MRI scan of the subject’s head, from which models of the magnetic fields generated by the brain (the forward model) are derived. Wearable MEG—based on optically pumped magnetometers (OPMs)—enables MEG measurement from participants who struggle to cope with conventional scanning environments (e.g., children), enabling study of novel cohorts. However, its value is limited if an MRI scan is still required for source modelling. Here we describe a method of warping template MRIs to 3D structured-light scans of the head, to generate “pseudo-MRIs”. We apply our method to data from 20 participants during a sensory task, measuring induced (beta band) responses and whole-brain functional connectivity. Results show that the group average locations of peak task-induced beta modulation were separated by 2.75 mm, when comparing real- and pseudo-MRI approaches. Group averaged time–frequency spectra were also highly correlated (Pearson correlation 0.99) as were functional connectome matrices (0.87) and global connectivity (0.98). In sum, our results demonstrate that source-localised OPM-MEG data, modelled with and without an individual MRI scan, can be comparable. While individual MRI scans remain the “gold standard” for OPM-MEG modelling, our method will be useful for future studies where MRI data capture is challenging.