VIRTUAL ELECTRODE ESTIMATION FROM MAGNETOENCEPHALOGRAPHIC RECORDINGS

Carnegie-Mellon University, Pittsburgh PA

Investigators

Linked publications & trials

Abstract

This subproject is one of many research subprojects utilizing the resources provided by a Center grant funded by NIH/NCRR. Primary support for the subproject and the subproject's principal investigator may have been provided by other sources, including other NIH sources. The Total Cost listed for the subproject likely represents the estimated amount of Center infrastructure utilized by the subproject, not direct funding provided by the NCRR grant to the subproject or subproject staff. The promise of magnetoencephalography (MEG) is to be the next great imaging modality. It provides sub-centimeter spatial resolution, millisecond time resolution, and direct access to functional processing by the brain. These are clear indicators for the potentiality of its uses for diagnosis of pathology, monitoring of treatment, identification of recovery milestones, and substantively advancing our direct access to and understanding of brain processes which underlie sensation, cognition, memory, attention, affect, movement, and pathologies which affect any one or more of these. But the devices are cumbersome, the analysis complex, the possibilities for application endless. The raw measurements provide voluminous data sets. Each of the typically hundreds of channels are, in general, a "superposition" of overlapping signals from numerous sources within (and outside) the brain, each of which is may or may not be of potential significance in understanding the problem under study. The overall goal of the project is to implement and demonstrate methods which provide the full range of information accessible via MEG and to then use that information to identify biomarkers which distinguish normal brain function and that for patients with mild Traumatic Brain Injury (mTBI). For every individual studied, the methods will reliably separate the signals from multiple sources within the brain, provide interpretable tracings from each of the chosen "virtual electrode" source locations, and will identify which sources are differentially activated under different conditions encountered during the recording session. Analysis of the resulting waveforms, parameters, and summary statistics obtained for each individual will provide "biomarkers" which reliably distinguish normal and pathological brain function. : In general, the signal observed at each MEG sensor is the superposition of magnetic field signals from many sources within the brain. Virtual Electrode Estimation simplifies this by casting the 306 MEG sensor signals into some 100 virtual electrode signals. Each of these is the estimate of the electric current at a specific location within the brain. This casts the raw MEG measures into a form which is "as if" 100+ wire recording electrodes had been placed directly in the brain. This both markedly "simplifies" the signals and allows interpretation of the variables which are retained by the stepwise classification analysis directly as locations within the brain. Virtual Electrode Estimation is carried out for all locations and for 30 msec at a time all at once. The noise rejection and signal identification properties of the method are very powerful and precise. The Virtual Electrode tracings are single trial evoked responses. 4. Identification of biomarkers which differentiate normal vs pathologic brain function. Rationale: Our working hypothesis is that changes due to mTBI in cognition, memory, and attention all arise from moderate slowing of white matter transmission speed. This slight slowing produces reduced synchronization in the arrival times for volleys of actions potentials, resulting in decreased activation in the receiving dendritic domain and consequent reduced likelihood of information transmission and subsequent information propagation. The decreased activation produced by information bearing action potential volleys gives rise to reduced ability to evoke memories. The decreased likelihood of information propogation gives rise to both cognitive and attentional deficits. In the short to medium term, to compensate for this, the tonic level of activation of the cortex is globally increased, increasing the likelihood that a moderately desynchronized volley of action potentials will produce sufficient activation in the receiving cell populations to induce action potential volleys and resultant persistence and propagation of information. Specific Aim 4: For each of the normal subjects and concussion patients, produce standardized reports containing the virtual electrode waveforms, classification accuracies, collapsed waveforms showing differential activation tracings, etc. Note that each concussion patient will have 2 studies, one acute and one 2-3 months post-injury. Provide reports from half of the normal subjects and half of the patients to each of the members of the clinician panel along with the diagnoses so that each clinician can independently identify those measures which they think will "stand up" as biomarkers of pathology. Then bring the panel together to discuss and refine those variables on which they agree. Finally, present each clinician, now blinded to the diagnoses, with reports from the other half of the subjects and patients and have them score and classify them. Hypothesis 4A: In the acute concussion, averaged evoked responses will show peaks with reduced amplitude and temporal prolongation. Hypothesis 4B: In the acute concussion, classification results will be less sensitive to the latency both following the stimulus and preceding the response. Hypothesis 4C: In the acute concussion, the collapsed differential activation tracings will show prolonged peaks. Hypothesis 4D: In the acute concussion, locations which are differentially activated will be distributed most widely in the brain.