Meeting Banner
Abstract #2147

Single-Trial EEG Discriminant Components Acquired During 3T FMRI

Jennifer M. Walz1, Jordan Muraskin1, Robin I. Goldman1, Truman R. Brown2, Paul Sajda1

1Biomedical Engineering, Columbia University, New York, NY, United States; 2Radiology and Radiological Science, Medical University of South Carolina, Charleston, SC, United States

We previously demonstrated that single-trial variability (STV) of EEG components recorded simultaneously with 1.5T fMRI can yield task-relevant BOLD activations that are unobservable using traditional fMRI analysis methods. A higher field MR system provides greater SNR of the fMRI BOLD signal, but reduces the SNR of the EEG due to increased magnitude of gradient, BCG, and motion artifacts. We have demonstrated the feasibility of our EEG-fMRI STV analysis methods in the more challenging environment of the higher-field 3T scanner, and we have surpassed our previous results in EEG discrimination performance, cluster size, and max z-score.

Abstract PDF Poster PDF

Keywords

able accuracy acquisition activations additional adjust analyses artifact artifacts attention attentional auditory bias biomedical blue bold bootstrapping brown caused challenging channel circles classification classifier cluster clusters comes components consistent correction correlated correlates correlations cortical cost cross dashed decreased despite determined determining diamonds discriminant discriminate discrimination dotted double enables endogenous engagement engineering environment epochs errors exceeded exceeding feasibility fidelity field finding findings frontiers functional gradient greater green height hypothesis internal interpreted larger latent leave locked logistic loss magenta magnitude modal modeled modulated motion negative negatively neural occipital oddball onsets orange output overlap paradigm participated pattern peak peaked performance positive power preprocessing previous previously prior psychology radiological radiology recorded reduced regressors relevant represent represents resolution response scanner scanning score scores selected seventeen significance since sliding south space specified statistical stimulus structures subject subjects supports system targets task threshold traditional training trial trials unobservable upon utilized validation variability varied widespread window windows yield yielded