DATE:

Friday, May 1, 2020

TIME:

3:00 PM

LOCATION:

Virtual Zoom Conference

SPEAKER:

Dr. Mark Pflieger, Cortech Solutions, Inc., Computational Science Research Center & Center for Clinical and Cognitive Neuroscience, San Diego State University

ABSTRACT:

The inverse problem of mapping an individual’s extracranial EEG potentials and MEG fields back to their current generators in the cortex and other brain structures has no unique solution (Helmholz). Consequently, many electromagnetic inverse solution methods are available. Mosher and colleagues showed that a broad spectrum of linear inverse methods, ranging from model-driven minimum norms to data-driven beamformers, are equivalent in theory. That is, they all share a general formula that combines three basic mappings: (i) sensor mappings that capture observed correlation patterns of extracranial signals; (ii) forward mappings from intracranial source currents to extracranial sensor fields; and (iii) source mappings that capture modeled correlation patterns of intracranial signals. In practice, however, inverse mappings differ due to various ways of obtaining the sensor, forward, and source mappings. Noting that each mapping varies from person to person, I will propose ways to calibrate each of the basic mappings for an individual person: (i) using the statistical framework of switching linear dynamical systems to model latent states underlying EEG/MEG sensor mappings; (ii) using co-resonant modes of simultaneous EEG-MEG plus brain/head geometry derived from structural MRI to tune tissue conductivities (for skin, skull, CSF, skull, CSF, gray matter, and white matter) of the forward mapping; and (iii) using simultaneous EEG-fMRI to fit a multi-level source mapping model of cortical functional connectivity to observed sensor mapping data. The inverse mapping that results from the general formula is a hybrid method that that harmonizes model-driven and data-driven inverse methods.

HOST:

Jose Castillo

VIDEO: