The electroencephalogram (EEG) and magnetoencephalogram (MEG) are directly and instantaneously coupled to the currents across cortical neuronal membranes which mediate information processing. They are widely used for both clinical diagnosis and for investigating the neural mechanisms of cognition with excellent temporal resolution. The goal of this application is to advance our understanding of the relationships between brain imaging signals at the macroscopic levels ? EEG and MEG - and the underlying circuits and cellular activity at the fine-grained scales. To address this goal, we propose to develop sophisticated computational neural models that are consistent with the large amount of data we already have concerning the synaptic and active currents in cortical neurons, their connections with each other and with the thalamus and the other brain structures, their organization in layers and columns, and areas, and their interconnections between areas (Bazhenov). This model will generate characteristic sleep and wake activity including oscillations and less organized rhythms. We will combine the neural model with biophysical models to calculate the consequent population phenomena (Halgren, Bazhenov). At a local level, transmembrane currents combined with cellular architecture and arrangement result in current source density, local field potentials and equivalent current dipoles, whose spatial arrangement, correlation and phase produce MEG and EEG. Transmembrane currents are calculated using populations of realistic multi-compartment neurons (Bazhenov), which are realistically mapped to actual reconstructed cortical surfaces, and propagated to extracranial sensors using realistic biophysical models (Halgren). We will refine and test the model using novel empirical analyses of large numbers of intracranial micro- and macro-array recordings, quantifying their amplitude, coherence, and phase-lag (Cash, Davis and Pati). To complete the loop, we will develop a novel inverse solution approach to acquire local level population activity from EEG/MEG data (Halgren, Dale) and we will link this activity back to the neuron and synapse level processes using computer models (Bazhenov). By establishing bidirectional links between circuits and cellular level activity generated by the model and predicted EEG and MEG signals validated by data, we will derive a set of predictions regarding what human EEG and MEG measurements can tell us about underlying cellular and synaptic level activity in empirical studies. We propose to use combined neural and biophysical modeling, confirmed with extensive intracranial recordings, as a framework allowing the principled quantitative integration of the many pertinent anatomical, physiological and neurobiological findings. The articulated model will thus embody our best current understanding of how EEG and MEG are generated. In addition, we will obtain for the first time the essential cortical information needed to empirically model the neural origins of EEG and MEG.
The EEG and MEG are complex phenomena that are the result of a particular summation and cancellation of billions of interacting neurons. We will create a computational neural model, that reproduces the essential elements of EEG and MEG from channels and synapses, through neurons and circuits, to columns, areas and the entire cortex, to identify the link between underlying circuits and cellular activity and the resulting macroscopic EEG/MEG recordings. This link is crucial for interpreting in neural terms the most common noninvasive imaging method, during diagnostic procedures and translational research, leading to more effective treatments of neuropsychiatric disorders, including advances in non-invasive electromagnetic approaches to their amelioration.
