Direct MEG/EEG detection using a novel MRI approach

Johns Hopkins University, Baltimore MD

Investigators

Abstract

This R21 grant responds to RFA-EY-17-001 ?BRAIN Initiative: New Concepts and Early-Stage Research for Large-Scale Recording and Modulation in the Nervous System ?for unique and innovative technologies in an even earlier stage of development ? including new and untested ideas in the initial stages of conceptualization? where preliminary data would not be available?. Electroencephalography (EEG) and magnetoencephalography (MEG) which measures the magnetic fields associated with the neuronal currents are the only tools currently available to noninvasively measure electrical activity in human brain. They provide a wealth of information on brain electrical activity in epilepsy, dementias, autism, depression, stroke, and trauma for clinical diagnosis and surgical planning. Yet the neuronal currents detected by EEG electrodes are confounded by intervening brain, CSF, skull and scalp tissues, while MEG requires superconducting magnetometers and a quiet magnetic environment. Moreover, the signal sources in EEG and MEG are not spatially-encoded per se and pose an inverse problem without a unique solution. While magnetic resonance imaging (MRI) plays a key role in The Brain Initiative with fiber track mapping and functional MRI (fMRI) methods and it aught to be sensitive to EEG/MEG signals, attempts to use MRI to measure neuronal electrical activity have thus far proved inconclusive or null. The MEG signals range from 10-15 to 10-12 Tesla(T) during epileptic spikes, while MRI?s sensitivity to neuronal sources is estimated at only about 10-10 T. Support is sought to develop an entirely new MRI approach to directly measure MEG-modulated MRI signals using a spatial encoding scheme that provides exceptionally high intrinsic signal-to-noise ratios (SNR) and speed. The new method of localization?spectroscopy with linear algebraic modeling or SLAM?uses a greatly-reduced SNR-optimized gradient encoding set to directly localize signals from small volumes of brain tissue that can be arbitrarily segmented and resized post-acquisition, from scout MRI. Tests show that SLAM can deliver an SNR of ?105, yielding a calculated sensitivity of 10-14-10-15T for the brain water signal in ~10ml volumes in ~30ms acquisitions (~30Hz bandwidth). Mechanistically, MEG/EEG modulation of MRI signals arises via the zero/low-frequency dipole-dipole component of the spectral density function, which is directly proportional to the spin-spin relaxation rate, 1/T2. The plan is to continuously apply T2-sensitive SLAM MRI at ~30 Hz to detect MEG/EEG modulations of ~10-14T signals over a meaningful ~30Hz EEG range in healthy human volunteers. Simultaneous MRI-compatible EEG will be used for reference to identify signals, detect artifacts and optimize detection of relevant signals. Studies to detect resting-state, visual- and audio-evoked potentials are planned. If successful, the grant would deliver noninvasive MRI-based localization and sizing of neuronal signal sources which might then be correlated directly with fMRI, fiber-tracking and other MRI metrics.